gocr ep0797806-001
(PICTURE)(PICTURE) (PICTURE)Europäis hes Patentamt (1 9) (PICTURE) European Patent Offi_e (PICTURE) Offi_e euroPéen des brevets (1 1 ) E _ O _9_ 8O6 B_ (1 2) EUROPEAN PATENT _PECl FICATlON (45) Date of pUbliCation and mention (51 ) Int CI.6_. _O6F _ _J3O of the grant of the patent_. 14_O7_1999 BUlletin 1999l28 (86) InternationaI appIication number_. PCTICA95lOO685 (21 ) Application number_. 95939199.6 (87) International publication number_. (PICTURE)(22) Date of filing_. 11 .12.1995 wo g6l18g58 (2o.o6.1gg6 _azette 1gg6l28) (54) METHOD FOR BINARY-ORIENTED _ET _EQUENCING VERFAH REN ZUR BINÄR ORl ENTl ERTEN G RUPPENANREIHUNG (PICTURE)(PICTURE) PROCEDE POUR LA MISE EN SEQU ENCE D' ENSEMBLES A ORl ENTATlON BINAl RE __ _ O OO h _h Note.. w._th._n n._ne months from the pubI._Cat._on of the ment._on of the grant of the European patent, any person may g._ve o nOtiCe tO the EUrOpean Patent OffiCe Of OppOSitiOn tO the EUrOpean patent granted. NOtiCe Of OppOSitiOn Shall be filed in ii a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. w (PICTURE)99(1 ) European Patent Convention). Printed by Jouve, 75OO1 PARIS (FR)
gocr ep0797806-002
EP O 797 8O6 B1 Des_ription (PICTURE)BACKGR UND OF THE I NVENTION 5 1 . FieId of the Invention. [OOO1] The present invention discloses a method and apparatus for data organization, storage, retrieval and process- ing, that eliminates locational, structural or associative limitations. 1O 2. Description of Related Art. [OOO2] Currently, information organization is implemented using many methodologies, which often serve different and distinct purposes for different general kinds of information. If the information is composed of specific facts, figures, names, and relationships, then the current approach forces each specific application to provide its own way of defining 15 and using records. The onIy process who can possibIy know what the information is, is the one appIication which creates/maintains that kind of information. At the lower OS level, data can be any type as far as a database or any other application is concerned. So files and directories are used to organize information, where physical and logical organizations of information are one and the same. [OOO3] To find qualified and desired information, a process must first deal with files and directories. The process must 2O incorporate and refIect the physicaI directoy and fiIe hierarchy into its IogicaI information organization. Many current containment shells such as WINDOWS and DESQVl EW, attempt to provide a seamless gap between the OS and applications, so that a process does not have to deal with OS details. Aside from being unstable, such shells are still constricted by containment. That is, the logical and physical organizations of information become the same at some Ievel. That is the level at which logical expansions, reorganizations, and further associations become impossible for 25 containment. (PICTURE)SUMMA Y OF THE INVENTION [OOO4] To overcome the limitations in the prior art described above, and to overcome other limitations that will become 3O apparent upon reading and understanding the present specification, the present invention discloses a computer-im- plemented method for information organization, wherein atomic information can be both static and dynamic, but the compound information (e.g., associations, groupings, sets, etc.) of such atoms always remain dynamic. Unless other- wise directed, a compound information entity is always dynamically determined and generated. This determination is based on the processing of a defined condition, wherein all atoms qualifying the condition are included in the compound. 35 This dynamic determination eliminates the need to ''update'' the compound, when atoms and/or compounds common to two or more compounds are changed. Further, each information compound can be dynamically generated based on an existing definition for that compound. BRIEF DESCRIPTION OF THE DRAWI NGS 4O (PICTURE) [OOO5] Referring now to the drawings in which like reference numbers represent corresponding parts throughout_. FIG. 1 is a block diagram that illustrates one possible hardware environment for the present invention; FIG. 2 is a block diagram that illustrates the structure of an InfoFrame according to the present invention; 45 FIG. 3 is a bIock diagram that iIIustrates the associations present in a containment organization_, FIG. 4 is a block diagram that illustrates the binary associations according to the present invention; FIG. 5 is a block diagram that illustrates grouping in a containment organization to achieve a combined topic; FIG. 6 is a block diagram that illustrates grouping according to the present invention to achieve a combined topic; FIGS. 7A-7E are a block diagram that illustrates a current method of organization implemented according to the 5O present invention_, FIGS. 8A-8B are a block diagram that illustrates a compound logical structure according to the present invention; FIG. 9 is a block diagram that illustrates associative processing according to the present invention; FIG. 1 O is a block diagram illustrating the structure of the Universal Entity Identifier (UEl) according to the present invention_, 55 FIG. 1 1 is a block diagram illustrating the structure of the Endo-Dynamic Information Node (EDIN) according to the present invention; FIG. 1 2 is a block diagram that illustrates the valid combinations of the EDl N fields in terms of value according to the present invention; 2
gocr ep0797806-003
EP O 797 8O6 B1 FIG. 1 3 is a block diagram illustrating the structure of the Bond Information Record (BIR) according to the present invention_, FIG. 1 4 is a block diagram that illustrates the structure of an InfoFrame according to the present invention; FIG. 1 5 is a block diagram illustrating the structure of the Bond Flags portion of the BIR according to the present 5 invention_, FIG. 1 6 is a blockdiagram illustrating the structure of the Bond Organization Record (BOR) according tothe present invention_, FIG. 1 7 is a block diagram that illustrates logical bond organizations according to the present invention; FIG. 1 8 is a block diagram that illustrates the structure of a command line according to the present invention; 1O FIG. 1 9 is a block diagram that illustrates the structure of an Endo-Dynamic Information Statement according to the present invention; FIG. 2O is a block diagram that illustrates the structure of an Expandable Table Set according to the present in- vention_, FIG. 21 is a block diagram illustrating the structure of the Expandable Table Record (ETR) according to the present 15 invention_, FIG. 22 is a block diagram illustrating the structure of the Expandable Table Array Header (ETAH) according to the present invention; FIG. 23 is a blockdiagram illustrating the structure ofthe InfoFrame Control Record (l FCR) according tothe present invention_, 2O FlG. 24 is a block diagram that illustrates the structure of an Activation _ist Example according to the present invention_, FIG. 25 is a block diagram illustrating the structure of the Infobase Definition Record (l BDR) according to the present invention; FIG. 26 is a block diagram illustrating the structure of the Set Definition Record (SDR) according to the present 25 invention_, FIG. 27 is a block diagram illustrating the structure of the Module Definition Record (MDR) according to the present invention_, FIG. 28 is a block diagram that illustrates the structure of a Set Definition Equation according to the present in- vention_, 3O FIG. 29 is a block diagram that illustrates the structure of an Endo-Dynamic Set comprised of Set Definition Equa- tions according to the present invention; FIG. 3O is a block diagram illustrating the structure of the Operator Information Record (Ol R) according to the present invention; FIG. 31 is a block diagram illustrating the structure of the Parameter Definition Record (PDR) according to the 35 present invention; FIG. 32 is a block diagram that illustrates flat storage organization according to the present invention; FIG. 33 is a block diagram that illustrates the structure of an Endo-Dynamic Set hierarchical data tree according to the present invention; FIG. 34 is a block diagram that illustrates the structure of an Endo-Dynamic Set structure definition according to 4O the present invention; FIG. 35 is a block diagram that illustrates the structure of a program example according to the present invention; FIG. 36 is a block diagram that illustrates the structure of an Endo-Dynamic Set comprising an interpreted program according to the present invention; 45 (PICTURE)DETAl_ED DESCRl PTlO_ OF THE PREFERRED EMBODl ME_T [OOO6] In the following description of the preferred embodiment, reference is made to the accompanying drawings which form a part hereof, and in which is shown by way of illustration a specific embodiment in which the invention may be practiced. It is to be understood that other embodiments may be utilized and structural changes may be made 5O without departing from the scope of the present invention. (PICTURE)OV RVI EW [OOO7] Information can be generally described as either being atomic or compound, where atomic information is an 55 elementary unit and compound information encompasses any combination of atoms and other compounds to se_e a given purpose. The present invention, termed Binay Oriented Set Sequencing (BOSS), is based on the concept that the minimal common information structure for any body of data is binay. This binary view of data organization achieves an information management environment in which any information of any complexity and type can be represented, 3
gocr ep0797806-004
EP O 797 8O6 B1 viewed, stored, and processed. [OOO8] The present invention further adopts a view of information organization where atomic information can be both static and dynamic, but the compound information (e.g., associations, groupings, sets, etc.) of such atoms always remain dynamic. Unless otherwise directed, a compound information entity is always dynamically determined and 5 generated. This determination is based on the processing of a defined condition, wherein all atoms qualifying the condition are included in the compound. This dynamic determination eliminates the need to ''update'' the compound, when atoms and/or compounds common to two or more compounds are changed. Further, each information compound can be dynamically generated based on an existing definition for that compound. [OOO9] The present invention differs in several ways from conventional information storage and processing environ- 1O ments. These are listed below, and described in ensuing subsections_. - The InfoFrame - Binary Association - Universal Entity Identification 15 - Dimensia - Associative Processing [OO1 O] As a result of these elements, an environment becomes feasible, where the elements of logical organizations of all data are stored as nodes. Further, processes are able to view the same nodes in different ways. To be more 2O precise, BOSS can achieve any logical structure given the same basic set of information atoms. This means data is maintained the same way, regardless of the process manipulating it, and regardless of the process'sviewof the involved data. Instead of having to implement and store static single-purpose control structures and records, a process need only store a set definition equation which results in a particular view of the data. This promotes an environment in which a general pool of nodes for all kinds of different data can be used as the basis for dynamically generating different 25 views for all the different processes which use pieces of the collective data pool in different ways. (PICTURE)HARDW RE ENVIRONMENT [OO11] Figure 1 depicts the hardware architecture of the preferred embodiment in accordance with the principles of 3O the present invention. Generally, the present invention operates in a network environment 1 OO having a decentralized hardware architecture, including one or more servers 1 O2, and a plurality of user workstations 1 O4, all coupled together through the network 1 O6. In the preferred embodiment, the network 1 O6 is depicted as having a ring topology. Those skilled in the art will be able to bring to mind other known network topologies such as, but not limited to, a star or a bus configuration. Typically, the server 1 O2 will include a database 1 O8, although the workstations 1 O4 themselves could 35 store all or a part of a database 1 O8. (PICTURE)THE NFOFRAME [OO12] The present invention provides for the existence of a global (universal) Information Frame (lnfoFrame), where 4O all types of systems (lnfoBases) which include various programs and databases (modules), and various data structures and datavalues (views and nodes) can co-exist in the InfoFrame, and data can be created, modified, organized, shared and exchanged on a dynamic basis. Data exchange across the InfoFrame is trivial, no matter what the system type. This makes all the processing of data import/export and usage across the InfoFrame invisible to all client processes. [OO13] FIG. 2 is a block diagram that illustrates the structure of an InfoFrame 2OO according to the present invention. 45 The present invention orders all information in the following manner_. - Information Frame (lnfoFrame) 2OO, which is the overall grouping of all information. - Information Base (lnfoBase) 2O2, which is a set of information modules and other control information that provides a self-contained set of consistent modules which provide for needs of a BOSS client. 5O - Endo-Dynamic Information Node (EDl N) 2O4, which is a binary association of two information atoms, Subject and Attribute, as well as a Bond that binds them. - Endo-Dynamic Set (EDS) 2O6, which is a dynamically generated, possibly ordered, list of EDl Ns 2O4that describes, depicts, or embodies a subject, attribute, or bond. - Information Module (IM) 2O8, which is a set of EDS's 2O6 and other control information that provides a selfcon- 55 tained set of consistent EDS 2O6 which provide for needs of a BOSS client. [OO14] An EDS 2O6 may contain any number of EDl Ns 2O4 or be empty. The contents of an EDS 2O6 are dictated by a condition for that EDS 2O6, wherein the condition is provided by a Set Definition Equation (SDE). The more 4
gocr ep0797806-005
EP O 797 8O6 B1 complex the SDE, the more particular data and/or data relationships are required to satisfy membership in the resultant EDS 2O6. However, SDE complexity does not directly effect the size of a resultant EDS 2O6. The size of a resultant EDS 2O6 simply depends on the number of EDINs 2O4 which satisfy the associated SDE. This size depends on the kinds of data being organized, and how frequently instances of that kind of data occur. 5 [OO15] All EDS's 2O6 are dynamically generated using one or more modules as the source of generation. A module 2O8 is primarily a set of EDl Ns 2O4 with no particular order. Modules 2O8 can then be included in one or more InfoBases. Each InfoBase 2O2 provides evolved information access, searching, and processing, by including as many modules as required to account for all data and processes. The InfoFrame 2OO is the totality of all InfoBases 2O2 and is the Iargest possible data space. Evey individual computer or network has its own InfoFrame 2OO. When two or more 1O computers or networks are connected such that data exchange is possible, the InfoFrame 2OO has simply become Iarger. In this sense, there is only one InfoFrame 2OO on a global basis, and it is just a matter of what portion of the InfoFrame 2OO a user is connected to or has access to. [OO16] An EDS 2O6 differs from a set, as that term is understood in the art, in several important ways. First, an EDS 2O6 can have a heterogeneous meaning. That is, an EDS 2O6 can contain any number of different meanings, data or 15 representations. [OO17] Second, an EDS 2O6 does not have to conform to the concept of containment. Containment is where an element ''X'' physically exists or is ''contained'' within set ''Y''. In a non-containment environment, there is no predefined meaning between ''X'' and ''Y'', just because one contains the other. In set methodology, the only meaning that can be derived is that ''X'' is contained by ''Y''. In the present invention, ''X'' and ''Y'' can have any number of relationships 2O defined directly in EDl Ns 2O4 or streams of related EDlNs 2O4. [OO18] Third, it is possible to execute expressions to modify an EDS 2O6 or create a new EDS 2O6 in terms of a meaningful formula. This formula is based on operators which can affect EDINs 2O4 or EDS's 2O6 in a number of different ways. In contrast, set-based mechanisms are based on adding or extracting the meaning of the set by adding or extracting elements from zero or more sets. 25 [OO19] Since, by default, all modules are accessible across the InfoFrame 2OO, each added module increases the possibilities for different and new InfoBases 2O2 by a substantially large number. Clearly, this increase is non-linearly proportional to the number of modules 2O8. Formula A below gives the number of unique possibilities, where ''n'' is the number of modules 2O8. Formula B gives the absolute increase in unique possible combinations of modules 2O8, when one module 2O8 is added. Formula C gives the real number of increased possibilities, by assuming that 3/4 of all such 3O unique possibilities have no meaning and serve no purpose in reality. Table l shows the calculated numbers based on different module 2O8 numbers. A -- ni. 35 B -- (n + 1 )! - n! 4o C -- ((n + 1 )!- n!)/4 wherein the ''!'' operator indicates a factorial operation. As can be seen, this is slightly less than ''n x n''. 45 5O 55 5
gocr ep0797806-006
EP O 797 8O6 B1 (PICTURE)TA LE I 5 (PICTURE) 1O 15 2O 25 [OO2O] The ease of integration and data sharing, combined with the rapid increase in potential new InfoBases, pro- vides an environment where, as more data is added and as more processing takes place, the environment as whole becomes more stable and capable. Further, using automation and chaining for all levels of the InfoFrame 2OO (including the InfoFrame 2OO), clients can tie together InfoBases 2O2 in particular ways, such that automatic activities take place, 3O these activities including the monitoring, retrieval, storage, and determination of_. - Entity value; - Entity organization; - InfoBase 2O2 chaining and the determination of chains; 35 - Module 2O8 chaining and the determination of chains; and - InfoFrame 2OO native processes and settings. [OO21] Given a defined InfoFrame 2OO, client processes can use BOSS operators as to manipulate part or all of the InfoFrame 2OO in variety of ways to display, modify, process, search, etc. , the information. A BOSS client always calls 4O the Endo-Dynamic Processor (EDP), passing it a list of operators (and their parameters). The EDP is a software Com- mand Processor (CP), which accepts an operations list as input and executes each line in the order specified by the operations list. Note that the operations list is an EDS itself, providing for variable number of parameters for the oper- ators. In a pure BOSS environment, a top-level operation list (program) would be executed at power-on. This program is an infinite loop where exit is possible by satisfying monitor processes, and where each required InfoBase is located, 45 verified, and initiated. In a multi-tasking environment (e.g. , Windows), linear module chaining and InfoBase chaining is possible. In a multi-processing environment (e.g., Windows-NT), real-time (and therefore non-linear) InfoBase chain- ing and module chaining is possible. [OO22] In a multi-site (computer) environment, each site executes operations lists via its own EDP and accepts remote operations lists as well. Since a Universal Entity Identifier (UEl) identifies the site from which data originates (i.e. , is 5O located on). remotely located data is potentially slower, but is handled via EDP-to-EDP communication and data transfer that is invisible to the client. Therefore, it is possible for one site to initiate a process that will execute via the EDP of another site. thereby leaving the original site free to perform further immediate processing. BI NARY ASSOCIATION 55 (PICTURE) [OO23] In a logical information organization, an atom of information can be a logical representation for a topic, event, process, or entity which can exist, be identified, and requires processing. A logical organization exists when the infor- mation atoms are associated in different ways to produce a structure. In current information organization methodologies, 6
gocr ep0797806-007
EP O 797 8O6 B1 the only kind of association between two atoms of information is containment. This is true no matter how evolved the containment method may be. For example, in object-oriented programming methods, objects can have associated predefined processes, where this is accomplished by the object containing the processes or references to those proc- eSSeS. 5 [OO24] Current methods maintain child-lists for each parent to record logical organization. A child is only recognized as having an association if it is a parent itself. [OO25] FIG. 3 is a block diagram that illustrates the associations present in a containment organization. In FIG. 3, the atoms shown on the right, identified as B 3O4, C 3O6, D 3O8, and E 31 O, are the children of A 3O2. The containment approach is to store the atoms (or identifiers to atoms) B 3O4, C 3O6, D 3O8, and E 31 O in the logical control record for 1O A 3O2. For example, a directory in DOS or UNIX is simply a file containing a list of that directory's contents. [OO26] Unlike a conventional set-element, the EDl N provides two atoms (or identifiers of atoms)_. a Subject and an Attribute. Each of the atoms given by an EDIN is an information atom in the sense described above. The Attribute provides additional association, enhancement or qualification to a given Subject atom. As a result, an EDl N contains a binary-association within itself. The EDIN is an independent entity, capable of existing in any EDS whose condition 15 (SDE) can be met. The Subject and Attribute atoms can be used in various ways to express any number of different associations. As a result, an EDS (a set of EDl Ns) can contain any number of singular relationships to describe a more general association. [OO27] FIG. 4 is a block diagram that illustrates the binary associations according to the present invention. FIG. 4 shows four EDl Ns with a Subject of A 4O2 and Attributes B 4O4, C 4O6, D 4O8, and E 41 O. The condition for this EDS 2O is simply that the Subject of each EDl N must be A 4O2. Each EDlN provides a relation, but all EDl Ns together describe the same tree as shown in FIG. 3. In this way, a collection of EDINs provide binary-associations between Subject atom A 4O2 and the various Attributes 4O4-41 O. [OO28] Note that it is equally possible to dynamically generate another EDS where the condition (SDE) is that the Attribute must be B 4O4. This EDS would describe all atoms which have the Subject in a relation with B 4O4. Yet another 25 SDE could produce an EDS where all Subjects are A 4O2 and all Attributes are B 4O4. This EDS would describe all the possible associations of A 4O2 and B 4O4. [OO29] Containment methods also experience problems when two existing information organizations need to be combined. FIG. 5 is a block diagram that illustrates grouping in a containment organization to achieve a combined topic. The two existing organizations under A 5O2 and B 5O4 are combined to produce a combined organization 5O6. 3O No new associations are produced, the associations of A 5O2 and B 5O4 remain unchanged. Containment methods have no way of actually integrating the two organizations because identification of atoms is based on location. Consider the atom D 5O6 under topic A 5O2. This atom also occurs under topic B 5O4. In the resultant combined organization, atom D 5O6 is duplicated, since the location of the two were different prior to combining the organizations. This is actually not good as the duplicate atom in the new organization will confuse existing processes which access atom D 35 5O6. As a result, combing two logical containment organizations is often manual and always time consuming. Using BOSS, the same two organizations can be combined without the problems common among containment approaches. [OO3O] FIG. 6 is a block diagram that illustrates grouping according to the present invention to achieve a combined topic. FIG. 6 shows an EDS format of the same topics A 6O2 and B 6O4 as in FIG. 5. Again, atom D occurs both under topics A 6O2 and B 6O4 prior to the merge. The combined topic 6O6 simply contains all EDl Ns of both topics, where 4O the condition for the EDS is that the EDIN Subject must be either A 6O2 or B 6O4. The atom D occurs as the Attribute of two EDINs, but is not duplicated. That is, both EDl N Attributes identify the same exact information atom. [OO31] In this way, BOSS is independent of the location of information atoms in an organization. In other words, BOSS achieves a complete separation of physical and logical locations of data. Note that this is only possible because of the endo-binary nature of the EDIN, the derived binary association in an EDS, and universally unique data identifi- 45 cation and location. (PICTURE)UNIVERSAL ENTITY IDENTI FICATION [OO32] The next cornerstone of BOSS is the universally unique identification and location of items. In BOSS, each 5O set and each element is uniquely identifiable via a Universal Entity Identifier (UEl). This means that any topic/object can be uniquely identified, under all conditions. If this were not true and a containment-based approach were used, BOSS could be inadvertently rendered impotent because the Subject and Attributes of an EDl N could be ambiguous. The ambiguity is introduced because if an item (Subject or Attribute of an EDl N) cannot be uniquely identified, it cannot be uniquely or correctly associated with other items. 55 (PICTURE)DIM NSIA [OO33] The next element of BOSS is called ''Dimensia''. Dimensia loosely refers to contexts or levels of abstraction 7
gocr ep0797806-008
EP O 797 8O6 B1 that a method for information organization is able to achieve at both atomic and non-atomic levels. Most current systems use flat or static multi-level methods for information organization. A flat method or structure employs one level of abstraction. A static multi-level method provides a maximum of N predefined levels of abstraction. [OO34] In all current methods, an information atom qualifies and/or describes itself. For example, an object in an 5 object-oriented programming system maycontain the atomic element's 32 bit number, string, and date. While describing itself, the singular atom does not immediately provide a relationship to any other entities. For example, an atomic date object element simply tells you that fact; it does not provide you with any relationships it may have to any other entities. It is the object containing the elements which is known to have relationships with each atom. By further including object deriving procedures in an object, one or more atoms may be related to other entities, but only the object driver knows 1O this fact, and the identity of the associated entities. So even in evolved containment methods, only a limited set of the data relationships are given by the data itself; the rest is process dependant. [OO35] FIGS. 7A, 7B, 7C, 7D, and 7E are block diagrams that illustrate a current method of organization, e.g., a binary tree 7O2. To have this logical structure, the organization can be of two general types. It can be an array 7O4 where tree-traversal is performed via mathematically calculated indexes, based on current index. Or it can be a set of 15 records 7O6 with left and right pointers. ln either case, an inflexible control structure is used to achieve logical organ- ization. These structures are inflexible because they are static in nature. For example, should the order of the elements in the array 7O4 change, a different tree (if one is decipherable at all) is now represented. The record example 7O6 is free of this problem, but the control record is particular and can only be used for binary trees and linked lists. [OO36] In BOSS, each EDIN contains a Bond between a Subject and an Attribute. As described above and shown 2O by FlG. 3 and FlG. 4, an EDS 7O8 can represent a tree by collecting all EDl Ns with the same Subject. However, to implement all levels in a tree, Attribute atoms in EDl Ns occur as Subject atoms in other EDINs. In FIGS. 7A-7E, to record the example binay tree 7O2, an EDS 7O8 records enough EDINs to relate all required associations. Note that to discern left and right children, the Bond specified in the EDl N can be used. Note that since an EDS is dynamically generated, insertions and deletions to/from a weighted tree 71 O are trivial as shown in FIG. 7E, and do not involve 25 complex left or right sub tree rotation. [OO37] In general, by duplicating the EDIN (i.e., twice the control data) with same Subject and different Attributes, two or more relationships of the same Subject atom and various Attribute entities can be established. In each case, the particular relationship is identified. In this way, all the different relationships an atom has to one other entity, as well as expressing all the different entities with which that atom has a relationship, can be expressed. Therefore, using 3O BOSS methodology, the maximum number of possible individually defined relationships of one entity with others is infinite and is only constrained by the amount of available storage space. This is a major aspect of Dimensia. [OO38] When complexand/or compound logical organizations of data are used, current methodologies are alsoforced to employ (and implement) complex and/or compound processes totraverse the organization. Considerthe case shown in FIGS. 8A and 8B, which are block diagrams that illustrate a compound logical structure according to the present 35 invention. In FIGS. 8A-8B, each atom of the example binary tree 7O2 of FIGS. 7A-7E is also an element in a distinct and separate two-dimensional array 8O2. Using current methodologies, new structures must be introduced, i.e., eight two-dimensional arrays of DATA l DENTl FIER. The DATA l DENTIFl ER values of atoms ''A'' to ''H'' would then be stored in appropriate locations in one of arrays, Al to A8. Now the program dealing with the data must not only include proc- esses for the old binary-tree record 7O6, but also include processes to manipulate the arrays. 4O [OO39] To implement the example compound organization 8O2 using BOSS, new EDINs are introduced, not new structures or processes. FIGS. 8A and 8B illustrate a simple (and inefficient) two-dimensional array implementation, where the EDl Ns are sequenced based on a value calculated from two-dimensional values. The EDS 8O4 actually encompasses all the arrays, where sequenced subsets of EDl Ns represent two-dimensional arrays. In each EDl N subset, the atom which coexists in the binary tree 8O2 is shown; this atom would occur in the sequence resulted from 45 its array coordinates. Each such EDl N set is not a two dimensional array in the actual sense, and is very sparse. Again, the dynamic nature of an EDS, means that the EDS is sequenced upon generation. [OO4O] Note that the additional EDl Ns 8O4 to represent the arrays could be stored together with the EDl Ns for the binary tree 7O8 of FIGS. 7A-7E in a module. Upon loading the module, and depending on the Set Definition Equation (SDE) used, one or the other of the EDS's can be produced. In this way, Dimensia is made possible for information, 5O where no new processes or control structures are required. and only new SDEs and EDl Ns are introduced and proc- essed (as per before) to produce different and currently incompatible views of the same atoms of data. (PICTURE)ASSOCIATIVE PROCESSI NG 55 [OO41] The next cornerstone of BOSS is associative processing. As mentioned above, a Bond in a given EDl N iden- tifies a native BOSS process associated with the Attribute of that EDl N, and possibly involving the Subject of that EDl N. In a BOSS environment, it is possible to automatically establish and execute new associated processes, based on a given information set. As a simple example, consider the two EDINs given as 9O2 in FIG. 9, which is a block diagram 8
gocr ep0797806-009
EP O 797 8O6 B1 that illustrates associative processing according to the present invention. Based on these EDl Ns, the BOSS process is able to automatically derive and store the third EDl N 9O4. From then on, the item ''X'' has direct relationships with both ''Y'' and ''Z''. Note that the derived EDIN 9O4 can only be assumed to be correct when the relation is transitive in nature (i.e., X BOND Y -- Y BOND X). For example, the ''brother of'' relationship is transitive, while ''father of'' is not. 5 Aside from simple association, BOSS can derive associations whose correctness is not absolute. Consider the EDl Ns given as 9O6 in FIG. 9. Possible automatic derivations are shown as the EDl Ns 9O8. Such proposed EDl Ns can then be automatically checked for correctness by gathering all data about ''A'' and ''C'' and then performing an exhaustive cross check to establish one of the following general results_. 1O - Not enough data - The correct and incorrect EDl N(s) [OO42] Since EDl Ns with any contents and purpose can coexist in a given module, it is possible to automatically derive new relationships and associated processing and store such new information in the same module, thereby 15 expanding the lnfoFrame of BOSS on an automatic and dynamic basis. (PICTURE)HARDWARE EVOLUTION [OO43] The principles of BOSS outlined thus far are hardware compatible concepts. It is possible to reduce the vast 2O majority of BOSS operators directly into hardware. lndeed, most of the BOSS operators and the Endo-Dynamic Proc- essor have been designed such that they can be converted (or evolved into) hardware. [OO44] This simple fact renders BOSS one of the most powerful data organization approaches in existence. Besides being able to operate two to three orders of magnitude faster than conventional data organization approaches when implemented in hardware, BOSS is infinitely more flexible. 25 [OO45] Based on the evolution of hardware devices, the demand for order of magnitude solutions is greater than ever. Further, the existing approaches to solving the increasingly complex data organization, migration and integration issues are being limited by the engines used. [OO46] The BOSS methodology also promises interesting advances in CPU design. Consider that a UEl can also be a machine code mnemonic. A natural result of this fact is that the data of an EDl N can also be a program, under the 3O correct circumstances. Further, it is possible to also create processing actions based on the binary relation found in an EDIN. (PICTURE)COM ONENTS 35 [OO47] Unless specified otherwise, when any component or list is stored to file, a number of operations occur. First, a checksum of the component is calculated. Next, the checksum, followed by a size (or number of records) is stored at top of file. Finally, the component is saved. Loading performs the reverse actions. 4O (PICTURE) [OO48] FIG. 1 O is a block diagram that illustrates the structure of a Universal Entity Identifier (UEl), which is the heart of information location and identification in BOSS methodology. A UEl contains twofields to provide a universally unique Iocation for a physical body of data. These are the Site Owner Identifier (SOl) and the Site Entity Identifier (SEl). The SOl is the serial number of the EDP operating at a given site or some other unique identifier for an EDP. The SEl is a 45 unique incremental number per site, where the SEl is assigned and incremented each time a new data entity is created. An SOl and SEl together are called a Combined Data Identifier (CDl). A CDl combines the duties of identification and physical location into a single entity. This is contrary to many current methods, where location is derived or is cross- referenced based on a given identifier. 5O (PICTURE) [OO49] FIG. 1 1 is a block diagram that illustrates the structure of an Endo-Dynamic Information Node (EDIN), which comprises the elements in an EDS. The EDIN is the most atomic form of stored BOSS information. An EDl N is com- posed of four fields, i,e., a Subject UEl, an Attribute UEl, a Bond UEl, and a Sequence field. The Subject, Attribute, 55 and Bond UEls can occur in other EDl Ns and in other fields. For example, an Attribute UEl can be a Subject or Bond UEl in another EDl N. The Sequence field is used to enforce a predefined order for the EDl Ns in an EDS. 9
gocr ep0797806-010
EP O 797 8O6 B1 (PICTURE)EDIN CO BINATIONAL BEHAVIOR [OO5O] FIG. 1 2 is a block diagram that illustrates the valid combinations 1 2O2 of the EDl N fields in terms of value, i. e., non-null, null, and ''any'' (i.e., could be either null or a valid value) according to the present invention. Since any 5 field in an EDIN can contain a null value, it is prudent to specify the exact set of possible combinations and their meanings. [OO51] The first and most common combination is for all valid EDl Ns which are simply elements in a set. [OO52] The second combination is used when an item of information is ''nullified'' (see below). This has the effect of making the Attribute item inaccessible in all non-edit BOSS processes. 1O [OO53] The last two combinations are shown for completeness. These combinations, and all others not shown by FIG. 1 2, are illegal and invalid occurrences of EDINs. (PICTURE)NUL I FIED EDI NS 15 [OO54] As shown in FIG. 1 2, if the Attribute of an EDl N has a null value, it is called a Nullified Subject Node (NSN), where the Subject of a NSN is the item being nullified. When a NSN is created, all EDl Ns with the same Subject and all the EDl Ns with the same Attribute UEl, as the NSN subject, are now prohibited from being including on all subsequent EDS generations. This has the effect of hiding information, or hiding a particular section in an organization. To remove a nullification (un-nullify), the NSN is simply removed from a module. Now, all previously invisible items or hierarchy 2O branches are made visibIe again. [OO55] The NSN is strictly optional and it's presence or absence does not invalidate or limit the working of BOSS methodology. If used, NSNs can augment BOSS with an information hiding capability. BOND 25 - [OO56] The third EDIN field is a Bond UEl value. This ensures that Bonds are universally unique. A Bond value identifies a process where processing occurs based on an interpretation of the EDl N Attribute field; these include a noun, verb, adjective, adverb, action, action-sequence, etc. In all cases, the Bond is known to be between the Subject and the Attribute. 3O [OO57] FIG. 1 3 is a block diagram that illustrates the structure of a Bond Information Record (Bl R), which records Bond information. A Bl R has three fields. The first is a Bond to provide a key in the Bond Information T_ble (BIT). The BIT is a list of Bl Rs sorted by the Bond field. As shown in FIG. 1 4, the BIT is stored at the InfoFrame level. The second field of the Bl R is a flag to describe the basic properties of the Bond. The last field of the BIR is UEl which identifies the associated process to be executed (by the EDP). This is most often an EDO, but can also be a major subsystem 35 of the EDP which handles this and other similar bonds or a BOSS program. The images and any default values for bonds are stored in the l MAGE-ETS and DATA-ETS at the InfoFrame level. [OO58] As shown in FIG. 1 5, the Bond Flags field in the Bl R gives the properties of the Bond, as follows_. Active/Passive - Active Bonds institute immediate processing and interrupt the active process flow until they are 4O terminated. Passive Bonds are relations which make a statement of fact or existence; they do not instigate immediate processing, but are used in the various BOSS processes to generate and process EDS's. Operator - This flag indicates that the Bond is an EDO. Although redundant, EDOs are also Bonds recorded 45 twice. The UEI for the EDO Bond is identicaI to the EDO UEI given for that EDO in the Operator IT (forces active as flag enable). Call - This flag indicates that the Bond is a BOSS process including a SDE, BOSS program, EDP command list (forces active as flag enable). When this flag is off, the associated process is 5O assumed to be an OS binary program. Spawn - This flag indicates that a non-BOSS process is to spawn concurrently or multitask (forces active as flag enable and call flag disable). 55 User/Native - This fIag indicates whether the Bond is a native Bond as suppIied by the EDP, or a Bond created by other person or process. Whenever a Bond is created, this flag is set to User, since any native Bonds would be supplied by the EDP or shipped as upgrades. 1 O
gocr ep0797806-011
(PICTURE) EP O 797 8O6 B1 [OO59] In order for Bonds to make sense to a user, not only do they have to have names, but also some form of organization. The names/images for all Bonds are stored in the IMAGE-ETS, using the Bonds as the search key. [OO6O] In order to provide organization, the Bond Organizational Record (BOR) is used. A minimal form of the BOR is shown in FIG. 1 6. This BOR contains only two fields, a SELF and a RELATED Bond. Using this simple record, almost 5 any logical organization of Bonds can be achieved. [OO61] As shown in FIG. 1 7, anything from a multi-level tree 1 7O2 to a simple list 1 7O4 is possible. Depending on the running process, different logical structures can be adopted. [OO62] A Bond value should always occur in the Bond EDl N field. Consider a bond UEl which is recorded as a subject or attribute of an EDl N, with some other Bond value in the EDIN bond field. When this EDIN is processed, the bond 1O recorded as subject or attribute will behave as a subject or an attribute, and not as bond. This can cause errors in EDP processes and clients which require and recognize the subjected/attributed bond for their critical processing. 15 [OO63] Any dynamically generated or simply loaded list of EDl Ns is an Endo-Dynamic Set (EDS). An EDS always has a particular purpose and meaning, as known only to the process using the data. For example, an EDS generated from a program module could be a program data structure, a program data occurrence, or a procedure occurrence. The EDl Ns in the EDS also may be or may not be ordered, depending on the requirements ofthe data being represented by the EDS as a whole. 2O [OO64] EDS's are identified by UEls, but for the most part, this is done indirectly and not in the same manner as other entities. The UEl associated with most EDS's is actually the identifier for a Set Definition Equation (SDE). Given any module, the SDE can be used to (re)produce an EDS with the same exact membership conditions and potentially different elements. Instead of storing the distinct EDS's present in a module, only the equations (SDEs) need be stored. This is required to ensure EDS's generated via SDEs remain dynamic at all times, and is somewhat smaller since set 25 elements need not be duplicated. Since an SDE is itself implemented as an EDS, it is necessay to store the SDE- EDS in the same module. (PICTURE)BASIC DI N SEOUENCES 3O [OO65] Information can be generally categorized as being active or passive. In this view, EDl N sequencing in an EDS takes one of two basic forms_. active sequence and passive sequence. An active sequence is always an executable process of some form; a passive sequence always expresses the structure, existence, qualities, properties, values, etc., of some information. Put differently, an active sequence performs some activity, while a passive sequence provides data about some information. Further, BOSS methodology allows for any combination and number of occurrences of 35 both kinds of EDl N sequencing in the same EDS. However, this would involve overhead processing, and the availability of a client program to process the passive sequences. Some passive sequences have associated native processes which handle or drive those particular kinds of passive information required for BOSS operations. In both cases, the EDIN sequence field is used to establish the EDl N order. For active sequences, the EDl N sequence value starts from zero and goes up to the number of required EDl Ns, where an EDIN sequence value is never duplicated in an active 4O sequence. For passive processes, the EDIN sequence may or may not be required, depending on the information being represented by the passive sequence. For example, if the passive information comprises files maintained hier- archically in directories, which exist in volumes, the sequence field is not required. However, if the passive information is a data structure definition, with elements in some depth, the sequence fields are used to order the elements of the structure. The only EDl N sequence value which can be duplicated in a passive sequence is a ''null'' value. 45 [OO66] To express an active sequence, one or more Endo-Dynamic Command Lines (EDCL) are used, where the order of the EDCLs, as established by the EDl N sequence fields, embodies the required active sequence. To express a passive sequence, one or more Endo-Dynamic Information Statements (EDIS) are used, where the EDl Ns may or may not be ordered by the EDl N sequence field. 5O (PICTURE)(PICTURE) [OO67] The BOSS central process, the EDP, takes command lines as input. An Endo-Dynamic Command Line (EDCL) is dynamic in nature, and variable length. The basic EDCL 1 8O2 is shown in FIG. 1 8. First, any number of EDl Ns bond any number of parameters to a subject identification, the subject being an EDCL 1 8O2. Then, an EDIN bonds the EDCL 55 1 8O2 (same subject) to an executable entity, shown as ''XXX''. The EDl Ns are ordered by the sequence field to place parameters before the execution occurrence. The executable entity XXX could be an endo-dynamic operator, a BOSS process (including SDE, BOSS program, activation list, etc.), or an OS executable program. [OO68] EDOs form the ''instruction'' set available from the EDCL 1 8O2. EDO EDCLs 1 8O2 are the fastest to execute,
gocr ep0797806-012
(PICTURE) EP O 797 8O6 B1 and require the least amount of overhead processing. A BOSS process is any ordered list of EDCLs. This could be an SDE, a BOSS interpreted program, an activation list for an InfoBase or InfoFrame, etc. An OS executable program is externally executed and requires the most amount of overhead processing. [OO69] The EDCL 1 8O2 differs from conventional command lines in several ways. Clearly, the EDCL 1 8O2 is variable 5 length in that any number of parameters are possible. The EDCL 1 8O2 is also dynamic, in that parameter and execute EDINs (all EDl Ns for an EDCL 1 8O2) can be changed dynamically, and the EDCL re-executed. Note that the EDCL 1 8O2 trigger for the EDP is the Bond field of the EDl Ns, not the subject or attribute fields. This is an important aspect of BOSS methodology. Using the Bond as a trigger means that, in EDCL 1 8O2 processing, information subjects and attributes can occur freely and without affecting the process flow. 1O [OO7O] The EDCL 1 8O2 forms the basis of BOSS processing. Using combinations of EDCLs 1 8O2, any process what so ever, using any kind and number of parameters, can be accurately recorded and executed. Using the generic EDCL 1 8O2 enables all BOSS clients to dynamically create and modify any kind of EDCL group, and then have it executed and re-executed by the EDP. As should be obvious, the EDCL 1 8O2 provides a simple and powerful way of implement- ing, maintaining and executing genetic algorithms. Many of the EDP initialization, and default information processes, 15 are expressed and stored as an ordered list of EDCLs 1 8O2. Any ordered list of EDCLs 1 8O2 is referred to as an Endo- Dynamic Command Set (EDCS) 1 8O4. 2O [OO71] An Endo-Dynamic Information Statement (EDIS) is two or more EDl Ns which make a statement of fact about some subject. The basic EDIS 1 9O2 is shown in FIG. 1 9. In this figure, several attributes are bonded (possibly via different bonds) to the same subject ''A''. When order and hierarchy are required for the information, an EDIN attribute UEl (shown as ''B'') occurs as the subject of other EDINs, whose attributes further describe the UEl (i.e., ''B'') originally occurring as an attribute of a subject. 25 [OO72] This is an important aspect of BOSS methodology. The interchangability of the subject and attribute UEls means that any depth and breadth of information hierarchy can be achieved. Further, upward or backward links can be introduced into the information hierarchy, such that a workable information networMgraph is achieved. [OO73] The EDIS 1 9O2 is dynamic in nature, so that the expressed passive sequence is a dynamically established one. Since EDl Ns can be freely inserted into an EDS, and the EDS reordered, any information represented as a passive 3O sequence remains dynamic. In the example shown in FIG. 1 9, the EDl N sequence fields are not used. However, many passive sequences require this field to establish order among the EDl Ns. Any passive sequence of EDl Ns is called an Endo-Dynamic Statement Set (EDSS) 1 9O4. DATA AND IMAGES 35 (PICTURE) [OO74] So far, all information has been referred to in terms of UEls. While the UEl does provide all required information about an entity to a process, it means little to an end-user. For example, while a program can process and maintain an EDS identified by the UEl value ''1 1 2_. 1 O'', it would be pointless for that program to display those numbers to an end- user. Clearly, names and/or images must be associated with each unique entity, so that a program can use them in its 4O display interface. Hereafter, ''image'' refers to both a binay image and a name-string. [OO75] Aside from an image, an entity (as represented by an EDl N), may also have associated physical data. For example, a BOSS-applied database program would use EDINs to record logical relationships and groupings, but it could not directly use EDl Ns to store the different data values being maintained by the database. [OO76] Both images and physical data are examples of Variable Length Data (VLD). To maintain and store VLD in 45 general, a format called ''Expandable Table Set'' is used. [OO77] As shown in FIG. 2O, an Expandable Table Set (ETS) 2OO2 is afile or memoy pair, consisting of an Expandable Table Array (ETA) 2OO4 and an Expandable Table Composite (ETC) 2OO6. The ETA 2OO4 and ETC 2OO6 must exist together or not at all. The ETA 2OO4 is a sorted list, where each element is an Expandable Table Record (ETR) 2OO8. Each ETR 2OO8 identifies information about (and the location of) an Expandable Table Block (ETB) 2O1 O within an 5O ETC 2OO6. As shown in FlG. 21 , the ETR 2OO8 is a record containing a UEl key, a flags field, an ETB size, an ETB checksum and an ETB file address. The ETRs 2OO8 in the ETA 2OO4 are sorted based on the UEl key. The ETC 2OO6 is simply a binary dataset composed of a number of variable-length ETBs 2O1 O, in any order. [OO78] To find an associated piece of VLD, a binay search is performed of the ETA 2OO4 for the input UEl. The UEl comparison is binary, so if any field of a subsequent input UEl is different, a different associated VLD occurrence exists. 55 Since the ETA 2OO4 is ordered by UEl value, one ETS 2OO2 can be used to store all VLDs of all data. Where two or more VLDs are required for a single UEl, separate ETS's 2OO2 must be used. [OO79] As shown in FIG. 2O, each ETA 2OO4 contains an Expandable Table Array Header (ETAH) 2O1 2 at the top, followed by the actual ETA 2OO4 (list of ETRs 2OO8). As shown in FIG. 22, an ETAH 2O1 2 contains_. a flags field, a self- 12
gocr ep0797806-013
EP O 797 8O6 B1 ETS UEl, an ETA size (number of ETR's) an ETA CHECKSUM field to enable verification of the ETA file upon loading an ETC size, an ETC CHECKSUM field to enable verification of the ETC file upon loading, an ETA memory address, an ETC memory address, an ETC buffer size, a current starting ETR identifier, and a current last ETR identifier. [OO8O] Through usage, VLDs will come and go in a system. That is, when entities are deleted, their associated VLDs 5 are also deleted. This would leave holes of unused space between the used ETBs 2O1 O of an ETC 2OO6. Fortunately, the process to optimize an ETS 2OO2 is trivial. First, a temporary ETC 2OO6 buffer is allocated. Then, starting from the first ETR 2OO8, and by keeping a current pointer, all valid ETBs 2O1 8 are copied, back-to-back, into the temporary ETC 2OO6. To finish, the ETC 2O1 O is overwritten with the temporary ETC 2OO6 buffer and the temporary ETC 2O1 O buffer is de-allocated. 1O [OO81] If memory is scarce, the optimization can be performed using a buffer as large as the largest ETB 2O1 O. In such cases, ETBs 2O1 O would be swapped (using unused holes) until they are in a back-to-back order. Unlike the first scheme, using a single ETB 2O1 O buffer, the ETBs 2O1 O in the resultant ETC 2OO6 may not be ordered in the same order as the ETRs 2OO8. [OO82] Since a BOSS element can have an image and have associated physical data, two ETS's 2OO2 are used for 15 each element. FlG. 2O also shows the minimum set 2O1 4 of ETS's 2OO2 required at any level to enable BOSS VLD maintenance. (PICTURE)I NFOFRAME 2O [OO83] FIG. 1 4 illustrates the components 1 4O2 of the Information Frame (lnfoFrame), which represents the highest Ievel of logical and physical data organization in BOSS. The InfoFrame is a definition of the collection and usage of all InfoBases found at a site, and other sites that may be connected to the home site. [OO84] FIG. 23 illustratesthe components of an InfoFrame Control Record (IFCR), which is contained in the InfoFrame to describe the default InfoBase processing, if any, for a site. The l FCR contains a Local Name UEl field to provide a 25 local name for the InfoFrame known to the current site that is used as key into the l MAGE-ETS for the InfoFrame (at this site). A Flags field is used to record InfoFrame processing configurations. An SOl field recorded from the serial number of the installed EDP is used to create all UEls generated at the current site. A Next SEl field is used to provide the next available SEl value across the current site, and this value is incremented, once read, by the EDP processes which create UEls. A Modifiers field is used to provide operational thresholds and guidelines for the InfoFrame, wherein 3O these modifiers are_. OLDEST VALl D EDS, START TlME, EDS MODl Fy OCCURRENCES, STOP TlME, and EDS MODl FY FREQUENCY. [OO85] To absolutely determine when EDS's require regeneration, it would be required to examine each EDl N in each EDS, to determine all possible EDS's which that EDS is dependant upon (in some way). Clearly, this is atime consuming and an infeasible methodology to adopt. Instead, the OLDEST VALl D EDS is used. This is a time scalar, indicating 35 how old a valid EDS can be. If this is a low number, EDS's are quickly deemed invalid and in need of regeneration. If a high number, generated EDS's are deemed valid for long periods of time. [OO86] While the EDP can record occurrences when distinct changes are made to individual EDS's or SDEs, this fact is not enough to estimate when an EDS requires regeneration. For this reason, the OLDEST VALl D EDS value is used. 4O [OO87] While this number can be assigned, an End-Dynamic Operator, ''DETERMINE-OLDEST-VALl D'', can be used at anytime to automatically determine a value forthis number. The START Tl ME, and the three modifiers EDS MODl FY OCCURRENCES, STOP Tl ME, and EDS MODl FY FREQUENCY, are used by the DETERMl NE-OLDEST-VALl D op- erator. When first initiated, this EDO records the START Tl ME, sets the EDS MODl FY OCCURRENCES to zero, and enables the RECORD MODl FY flag in the IFCR flags. This flag indicates that each subsequent EDS modification 45 requires an increment of the EDS MODl Fy OCCURRENCES modifier. Finally, this EDO prompts for a time duration, and records a STOP Tl ME. At the appointed stop time, the EDS MODl FY FREQUENCY is calculated based on the other assigned/accumulated modifiers. This frequency is then used to determine an estimated OLDEST VALl D EDS value. [OO88] The InfoFrame also contains a Default InfoBase List (Dl L), whose elements are EDl Ns and which comprises 5O an EDCS. FlG. 24 shows an example DlL 24O2 with three EDCLs. First, for each parameter required for an lnfoBase activation, an EDIN occurs. No parameter EDl Ns are present if the InfoBase requires no parameters. After the param- eter EDl Ns, the last EDIN associated with the InfoBase occurs, where the ''activate InfoBase'' EDO performs all tasks associated with locating and activating a particular InfoBase. [OO89] A Default Module List (DML) is used in the InfoFrame, whose elements are EDl Ns. The DML is an EDCS, 55 exactly as the DlL 24O2, except that the EDlN subjects are all a UEl generated for the lnfoFrame DML . The lnfoFrame- DML is used and loaded before the Dl L 24O2, and InfoBase DMLs. This enables the EDP to load native modules which may have a hand in loading and activating InfoBases. [OO9O] The InfoFrame also contains an InfoBase definition List (l BDL), where each element is an IBDR. The l BDL 13
gocr ep0797806-014
EP O 797 8O6 B1 is frequently updated to ensure any newly added InfoBases are included. The InfoFrame contains Data and Image ETS's to record such data associated directly with the InfoFrame. The InfoFrame contains an Operator Information Table (OlT), to identify and describe all Endo-Dynamic Operators. The InfoFrame contains one or more EDO program files, each containing the executable code for one or more operators. The InfoFrame contains a Parameter ETS, to 5 describe all parameters for all EDOs. The InfoFrame contains a Bond Information Table to describe all bonds. The InfoFrame contains a Default Command List (DCL), to provide a ''default dynamic program'' which EDP always (and possibly continuously) executes. I NFOBASE 1O (PICTURE) [OO91] FIG. 1 4 illustrates the structure of an Information Base (lnfoBase), which is a conglomeration of one or more information modules. An InfoBase Definition Record (l BDR) is used to provide image and processing options for the InfoBase. An l BDR file exists for each InfoBase for import/export purposes. All regularly used l BDRs are stored on an InfoFrame basis. 15 [OO92] As shown in FIG. 25, the l BDR is composed of_. a FLAGS field to provide processing switches, a self-UEl field to uniquely identify the InfoBase, and an image-UEl field to provide a key into the InfoBase assigning an image for the InfoBase. The Flags field is identical to the one in the l FCR. [OO93] A Module Definition List (MDL) is used to provide a list of included modules in the InfoBase. Each element of the MDL is an MDR as described under module section. 2O [OO94] A Default Module List (DML) is used for the InfoBase structured exactly as the DML stored at the InfoFrame Ievel. [OO95] The following modules identified by the InfoBase DML are loaded and activated upon InfoBase activation. The Data and Image ETS's are used to record such data associated directly with the InfoBase. [OO96] Modules can be included in an InfoBases in two ways_. shared, and exclusive. A shared module physically 25 occurs once across all lnfoBases in the current lnfoFrame, but may be included in all lnfoBases. ln a BOSS-applied environment where concurrent processing is possible, the usual precautions and preprocessing must be applied before access is granted. An exclusive module is what all modules are by default, one that is exclusive to a particular InfoBase. An exclusive module only appears in the InfoBase it is exclusive to. While other InfoBases can access an exclusive module, any such access is regulated by the owner InfoBase. 3O [OO97] An InfoBase can store a large amount of data and processing. In general, an InfoBase will have one or more modules containing data in one or more data organizations, and one or more modules containing programs which process that data. The modules containing programs which process that data are optional, in that the programs that process the BOSS data need not be written as BOSS programs; they could be any binary program. 35 (PICTURE)MODULES [OO98] FIG. 1 4 illustrates the structure of an Information Module (l M), which is a collection of EDl Ns and ETS's to record the associated images and physical data. The minimum set of required ETS's is used as described in the previous section. These ETS's store all images and data for the module as well as for all EDl Ns in the CNL. When 4O saved EDS's are present, images associated with saved EDS's are also stored in these ETS's. [OO99] A Collective Node List (CNL) stores all EDl Ns, in arbitray order, which together make up all the EDS's which can be generated from that module. The CNL is always loaded upon module activation. Most EDS generation operators require the specification of one or more modules to use as a source of generation; in such cases, all associated CNLs must be loaded (in turn) and used as a source for generation. 45 [O1 OO] A Set Definition List (SDL) maintains a list of ''saved EDS's''. Each element in the SDL is a Set Definition Record (SDR). As shown in FIG. 26, each SDR contains a self- identifier UEl field identifying the EDS, a GENERATlON- PROCESS UEl field, a Flags field, a LAST-GENERATlON field, an EDS size field, and a memoy address. The LAST- GENERATlON and memory address fields are only used at runtime, after the EDP has loaded a particular module, and provide the current location and size of an EDS in memoy. The GENERATlON-PROCESS UEl identifies a process 5O which will generate the EDS_, this can be either an SDE, or another process. The LAST-GENERATlON field is also only used at runtime; it is a date and time stamp of the last generation. This field provides a measure of how valid or up to date the EDINs pointed to by EDS address fields are. This field is compared to (current time - oldest valid eds), and if older, the associated EDS is deemed to be invalid and in need of regeneration. The SDR flags field is used to record which EDS's are temporary and which are not. Further, it identifies whether the EDS generation process is an SDE, 55 or other process. All newly generated EDS's are by default temporary. Using EDOs, a newly generated EDS can be made permanent, or a generation process can be made to result in a permanent EDS. [O1 O1] When an EDS is saved to a module, only unique EDl Ns are added, or old EDl Ns updated in the CNL; EDl Ns are never duplicated in the CNL. Next, the EDINs which make up the equation (SDE), used to generate the EDS, are 14
gocr ep0797806-015
EP O 797 8O6 B1 also added to/updated in the CNL. Finally, the UEl for the SDE is added to/updated in the SDL. Now, upon subsequent module loading, a client can first re-generate the SDE, then execute the SDE to regenerate a (new version of a) previously saved EDS. [O1 O2] When a new EDS is dynamically generated, a new unique UEl is assigned to it, and a new SDR created in 5 the SDL. The self-identifier field of the new SDR is assigned from the newly created UEl value. All SDR flags are cleared, the last generation date is set, and the EDS size and address fields assigned from the newly generated EDS buffer. The SDE is set from the LAST-SDE global variable; this variable is cleared in each EDP cycle, and is set by the Iast line of any SDE. As a result, it can be used by the EDP processes to determine the associated SDE (or NULL for none). 1o [O1 O3] As shown in FIG. 27, the Module Definition Record (MDR) provides a UEl for the module image (stored in the module ETS's), as well as default processing flags for a module. These are the same flags as for the l FCR and l BDR. The MDR for a module is always stored in a separate file; this file is only used when importing or exporting modules. The MDR in this file (along with all other modules used by an InfoBase) are duplicated in the Module Definition List as defined for an InfoBase. So, in reality, the MDL contains the latest version of all MDRs, and when import/export 15 is required, the MDR file is generated and used. This is done to avoid potentially long update periods every time a module is modified in some way, and poses no problems because the MDR file is not used in regular processing; only for import/export. [O1 O4] A Default EDS List (DEL) is used, where each element is a UEl identifying an EDS to generate (i.e., identifying an SDE/process to execute). All default EDS's are generated upon module loading. 2O [O1 O5] The l M is a self-contained package of information, providing values, images, data organization(s), data as- sociation(s), and data processing. The l M is always constructed to serve the needs of a client BOSS process. Since an EDS can always be dynamically generated from a CNL, it is possible to place incongruent or inconsistent information in the same module; although this is not recommended, it poses no problems to the BOSS environment, and valueS, organizations, and associations remain unaffected. Some module examples follow_. 25 - A BOSS program, where procedures, data-structure definitions, and data occurrences are recorded, and later generated as, EDS's, which are processed by eitherthe EDP ora client BOSS program interpreterto run a program. - A BOSS menu system, where menus are recorded, and later generated as, EDS's, which are processed by the BOSS menuing client. 3O [O1 O6] In general it is bestto group information common tothe same compound information entity in the same module. While out of context EDINs in a module do not create problems by themselves, out of context SDEs and EDl Ns would create potentially fatal processing problems. For example, consider an out-of-context SDE which generates an EDS for a data structure definition by default, for a module whose purpose is menuing. This would more than likely hang 35 the menuing client. For these reasons, a BOSS client can construct SDEs which ''filter'' all input EDl Ns for consistency. Such SDEs can check for particular types and allows and disallow the input. So if the module is a menuing system, data types like ''procedure'' could be optionally disallowed. 4O (PICTURE) [O1 O7] As mentioned above, EDS's are based on Set Definition Equations (SDEs). An SDE is an expression com- posed of Endo-Dynamic Operators (EDOs) and operands. An Endo-Dynamic Operator (EDO) can be almost any kind of operator. FIG. 28 shows an example SDE 28O2 with a C-like format. A new EDS called MY-EDS will be the result of resolving the right hand side of the equation. The atomic binary SDE units are shown and numbered 1 , 2, and 3 45 from the deepest to the outermost SDE unit. The SUB, SEQ, and ATT mnemonics are EDOs that perform filtering based on different fields of the EDl N. The l NTERSECT mnemonic is a logical EDO and signifies that the resultant setS of both operand expressions must be intersected. The expression shown in FIG. 28 dictates that all EDl N's in the ''MY- EDS'' EDS will have a Subject equal to ''W_.X_.'' and an Attribute equal to ''A_.B_.''. The ''MODULE-N'' module is the module used here for all operators. except l NTERSECT. 5O [O1 O8] The SDEs are always binary in nature. No matter how complex the equation, it can always be broken down into binary (and unary) SDE-units. As a result, an SDE is easily implemented as an EDS. [o1o9] FIG. 29 shows an EDS 29O2 for the SDE depicted in FIG. 28. This shows the Subject and Attribute fields of the EDIN as UEls. This EDS 29O2 also shows the Bond and sequence field values. As can be seen in FIG. 29, the SDE is simply an EDCS. In this case, these EDCLs are shown, i.e., one for each EDO showing in FIG. 28. 55 [o11 O] In this way, an EDS 29O2 can be used tostore equations (SDEs) which define howother EDS's are dynamicalIY generated. The subject fields of all EDl Ns will always contain a unique SDE-UEl associated solely with MY EDS. The SDE by itself does not result in anything. But when the SDE is applied to an existing EDS or module, a new EDS can be generated. As a result, a single module, with multiple SDEs, can provide different dynamically generated EDS's. 15
gocr ep0797806-016
(PICTURE) EP O 797 8O6 B1 [O111] As mentioned above, Endo-Dynamic Operators (EDOs) are tothe Endo-Dynamic Processor (EDP) as instruc- tions are to a processor. An EDO is any executable body of code requiring any number and type of parameters. While 5 the code for most EDOs is in the form of a binary executable OS program (or procedure), EDOs expressed as EDCS's can also be created and used. As should be obvious, an EDO occurrence with all its required parameters forms a complete EDCL. So it is possible to construct an EDO and EDCL, such that the EDCL activates the EDO (via the EDP), wherein the EDO is itself an EDCS processed by the EDP. This forces the EDP to be re-entrant, where the EDP must be capable of correctly processing any number of EDCS's in as many streams of processing as initiated by various 1O processes. [O112] Unlike conventional ''instructions'', the EDO is not limited in size or complexity. The EDO can be anything from a one line procedure to a whole system (program). Further, EDOs can freely call each other without interfering with EDP process leveling or the EDP stack. [O113] Incorrect process streams which are potentially fatal are terminated, mostly before and sometimes after a 15 fatal process error has occurred. All stack data regarding the process(es) which were involved in a process stream resulting in the fatal error(s), can be safely and accurately removed from the EDP stack, such that pursuant EDP processing, and other existing processes can continue. [O114] For each EDO available for use, there exists an Operator Information Record (OlR). As shown in FIG. 3O, each Ol R contains_. an OPERATOR UEl field to provide a key in the OlT, a NUMBER OF PARAMETERS field to give 2O the total number of input and output parameters required by the operator (the number of PDRs in the associated list), an Endo-Dynamic-Libray UEl to identify the library that contains the executable code of the EDO, and a CALLl NG ADDRESS field to provide a memory address forthe operatorthat is onlyvalid at runtime afterthe operator's executable code has been loaded into memoy. The OlRs are used at run time to verify calls to, and execute operators. As shown in FIG. 1 4, all Ol Rs are permanently maintained in the Operator Information Table (OlT), maintained at the InfoFrame 25 level. The OlT is a sorted list, wherein a binay search locates a given OlR. The OlT is updated when EDOs are imported or modified. [O115] To record parameter data requirements for EDOs, the PARAMETER-ETS is used. As shown in FIG. 1 4, this ETS is stored at the InfoFrame level. The ETRs in this ETS have EDO UEls as the keys. The ETBs store ordered lists of records, where each record is a Parameter Definition Record (PDR). As shown in FIG. 31 , a PDR contains_. a Flags 3O field to identify general l/O type of the parameter, a Data Type field to identify the required data type to internal BOSS processes, a Data Size field to give the size of the identified data type, a Type Image UEl field that identifies an image for the data type stored in the InfoFrame l MAGE-ETS and a Default Value UEl field that points to a default value occurrence for the parameter in the InfoFrame DATA-ETS (if no default value is supplied, this field contains a null UEl). Both the OlT and the Parameter ETS are used at run by the EDP to perform verified dynamic enty and execution of 35 EDCLs. (PICTURE)EDO- NFOBASE [O116] An Endo-Dynamic Library (EDL) is an Information Module which provides a means fortransporting and storing 4O all information regarding a given set of EDOs. All EDOs in an EDL should be related in some general way; this is often (part of) the name for the library. Note that hereafter and throughout the document and figures, ''Library'' is used inter- changeably with ''EDL''. [O117] A strictly logical entity called Endo-Dynamic Group (EDG) is used to organize all EDLs in various ways. Note that hereafter and throughout the document and figures, ''Group'' is used interchangeably with ''EDG''. 45 [O118] All EDLs are collected by the Endo-Dynamic Operator InfoBase (EDO-lnfoBase). The EDO-lnfoBase is an InfoBase like any other, but also encompassing any additional program files required by the EDLs. The EDO-lnfoBase provides a way of accessing all available libraries and libray information. Further, using the InfoBase DML, a certain base set of libraries are always activated (i.e., loaded and ready to be processed via EDP). The EDO-lnfoBase is supplied with each EDP program/package, and is necessary for the operation of the EDP. 5O [O119] While the OlT and the parameter ETS provide for quick EDP processing at thetop level, the bulkofthe required data for the EDO processing is provided by the EDO-lnfoBase. The module components as shown in FIG. 1 4 are used as follows for a libray. A library module is no different than any other module, except that in some cases additional program files are also associated with the module. [O12O] The Module Definition Records (MDR) comprises normal module information, wherein the System flag is 55 enabled. [O121] The Collective Node List (CNL) stores EDl Ns in no particular order. As well as normal SDE recording and processing, these EDl Ns are used in two ways. [O122] First, these EDINs are used to organize and specify EDOs in the libray, the fields need to be set a certain 16
gocr ep0797806-017
EP O 797 8O6 B1 way. The Subject field should contain a UEl for an EDL,or a UEl for an EDLG. The Attribute field should contain a UEl for an EDO. The Bond field should contain an ''EDO Occurrence,'' which indicates that there is an occurrence of the EDO given by the attribute in the libray given by the subject. Finally, the Sequence field is not used. [O123] In addition, these EDl Ns are used to record EDOs programmed as EDCLs, all fields are set as per an EDCL. 5 All EDl Ns in this CNL can be sorted by the two keys, i.e., subject and attribute, to provide an overall hierarchy of the operator groups and libraries. In addition, the CNL can be filtered for a particular subject (library or group) to generate EDS's which can be used as menus, which are then traversed, generated a new menu EDS at each traversal step. When an EDIN in library menu EDS is selected, any number of further information is available for the EDO identified by the EDl N's attribute (e.g., PDRs, image, code, etc.). The client process can then perform further processing using 1O the EDO information. [O124] The Image-ETS stores all images associated with all EDOs (and their parameters) in the library, as well as the image(s) for the library itself. [O125] The Data-ETS uses a UEl key. Associated with the library (i.e., using the module UEl), is an ETB containing a list of OlRs. This provides a list of all EDOs in the libray. Associated with each EDO (i.e., using the EDO UEl), is an 15 ETB containing a list of PDRs, describing the parameters of the EDO. [O126] The SDL identifies SDEs (stored in the CNL) to generate an EDS. For SDE-1 , EDINs are sorted by the two keys_. by subject and attribute to provide an overall hierarchy of operator groups and libraries. [O127] For SDE-2, EDl Ns are filtered for a particular subject (library or group) to generate EDS's which can be used as menus, which are then traversed, generating a new menu EDS at each traversal step. When an EDl N in libray 2O menu EDS is selected, any number of further information is available for the EDO identified by the EDl N's attribute (e. g. PDRs, image, code, etc). The client process can then perform further processing using the EDO information. [O128] For SDE-3, EDINs are filtered for a particular EDO subject, and sorted by the sequence. This SDE generates an EDCS executable by the EDP. [O129] For DEL, this has one default EDS_. EDS for the top-level group. This is the same as SDE-2 above, where 25 the source subject is predefined. [O13O] If the code associated with an EDO is an EDCS, all EDINs required to make up the EDO's code body are also stored in the CNL. If the code associated with an EDO is not an EDCS (i.e., if the EDO code is some form of OS executable code body), in addition to all normal module files, an OS program file is also stored. This program file has a filename derived from the associated EDO UEl values, plus the normal OS executable extension. All such executable 3O EDO program files are stored at the InfoFrame level. [O131] When any libray information is loaded, imported, or modified, appropriate updates are made at the InfoFrame Ievel. After any such events, the OlT, parameter ETS, and any EDO program files are updated as required, using the just saved library data. While updating the system EDO information is a simple procedure of replacing records and files, the effects of such updates on user data containing references to the updated operators could potentially be a 35 difficult to determine and diagnose. (PICTURE)LI BRAR ES AND OPERATORS [O132] The EDP requires a minimum basic set of libraries to operate. These are_. 4O - Set Filtration Libray - Control Flow Library - Physical Manipulation Library 45 [O133] The general membership requirement(s) and a minimum set of EDOs are described for each library by the following sections. In addition to any listed EDOs, any other qualifying EDO can be added to a library. However, all such dynamically created EDOs are always tagged as ''user'' in the associated Ol R. [O134] When EDOs are also made into bonds (a matter of creating bond control records, since the process already exists as the EDO), a viable but limited language is realized for defining and executing BOSS programs embodied by 5O information modules, complete with data definitions as realized by EDS's (each is an EDS in the module) and executable code as realized by EDCS's (each is an EDS in the module). The more evolved and/or complex EDOs that are intro- duced, the more robust such a language will become, but it will do so non-linearly. This is because any introduced EDO can call others in any (meaningful) combination that it sees fit. Each added EDO increases the number of new possibilities combinatorially. 55 [O135] Further, given sufficient numbers of added EDOs, any number of such dynamic programming languages are simultaneously possible, where languages can interface invisibly to any of the involved specific language processes. Any and all such languages are simply an implementation of several BOSS concepts and the specific usage of several BOSS entities disclosed in this patent. 17
gocr ep0797806-018
EP O 797 8O6 B1 (PICTURE)SET FI TRATION LI BRARY [O136] An EDO in this library must process EDl N list(s), based on any kind of input, to produce subsets of that EDl N list, or new EDIN list(s). The minimum required set of filter EDOs are described below. The EDOs listed below constitute 5 the minimum required set of EDOs in the filtration library_. _ [O137] This EDO combines two or more input EDS's into a third output EDS. 1O (PICTURE)Intersection [O138] This EDO examines two or more input EDS's for common EDl Ns and outputs a third EDS containing only the common EDl Ns. 15 (PICTURE) [O139] This EDO searches the input EDS for EDl Ns having a match in their subject field with an input subject and returns those EDl Ns in a new EDS. 2O (PICTURE)Attr'bute-match [O14O] This EDO searches the input EDS for EDl Ns having a match in their attribute field with an input attribute and returns those EDl Ns in a new EDS. 25 (PICTURE)Bo d-match [O141] This EDO searches the input EDS for EDl Ns having a match in their bond field with an input bond and returns those EDl Ns in a new EDS. 3O (PICTURE) [O142] This EDO searches the input EDS for EDl Ns having a match in their sequence field with an input sequence and returns those EDINs in a new EDS. 35 (PICTURE)(PICTURE) [O143] This EDO sorts the EDINs in the input EDS by their subject field and then assigns sequence numbers to those EDINs based on their sorted order. 4O (PICTURE)Remove-nodes [O144] This EDO searches for and then deletes the input EDIN from the input EDS. 45 (PICTURE)Descendants [O145] This EDO searches for any EDINs in the input EDS that are the descendant of the input EDIN. Ancestors 5O (PICTURE) [O146] This EDO searches for any EDINs in the input EDS that are the ancestor of the input EDl N. _ 55 [O147] This EDO searches for any EDINs in the input EDS that are the siblings of the input EDIN. [O148] All of the above EDOs provide the basis for constructing SDEs to formulate and process anything from a simple database query, to data structure collection and processing, to evolved, multi-level queries where specific in- formation is qualified to any degree and extent. Each specific application requiring filtration would introduce SDEs 18
gocr ep0797806-019
(PICTURE)(PICTURE)(PICTURE)(PICTURE)(PICTURE)(PICTURE)(PICTURE) EP O 797 8O6 B1 which use the above listed system-EDOs in various combinations with other EDOs to accomplish further specific fil- trations. Any one such client procedure or program (in a client program module) can be made into a user-EDO, and incorporated into the currently known InfoFrame. Clients would create all EDOs associated with a general purpose in the same EDO-library, and add the libray to the EDO-lnfoBase. This makes the client supplied EDO accessible by all 5 BOSS clients in the lnfoFrame. [O149] In this way, flexible, dynamic, and custom-made information search engines can be built and supplied as EDOs. Such EDOs would be then used by even bigger BOSS clients such as an expert system, to unify, simplify and speed up minor information gathering and simple correlation tasks. 1O (PICTURE)CO_T_O_ F_OW _lB_A_y [O15O] An EDO in this library must be associated with process flow of the EDP, or that of a BOSS client. The following lists and describes the minimum required EDOs for this libray. Although more complex control flow EDOs are possible, such EDOs would simply be ''implementations'' of the technology disclosed by this patent. 15 _P h [O151] This EDO pushes the parameters onto the EDP stack. 2O _P [O152] This EDO pops a value from the stack into the parameters. Peek 25 - [O153] This EDO uses a stack index number to determine a stack enty from the top. and return the value stored therein into the given parameters. Poke 3O - [O154] This EDO uses a stack index number to determine a stack entry from the top. Then the parameters are stored into the stack enty. 35 (PICTURE)(PICTURE) [O155] This EDO returns a true or false value dependent on the condition of the stack. (PICTURE)St ck-fuII 4O [O156] This EDO compares total stack space against currently used space and returns true or false value dependent on the condition of the stack. 45 [O157] This EDO locates, loads, and executes the binary OS program identified by the input UEl. This EDO waits for the program to terminate before returning. 5O [O158] This EDO locates, loads, and spawns the binary OS program identified by the input UEl, as a concurrent process. This EDO does not wait for the program to terminate before returning. 55 [O159] This EDO locates, loads, and executes/spawns the BOSS program identified by the input UEl. This EDO may or may not wait for the program to terminate before returning, depending on availability of concurrent processing in the environment. 19
gocr ep0797806-020
(PICTURE)(PICTURE)(PICTURE)(PICTURE) EP O 797 8O6 B1 [O16O] This EDO sets the global variables associated with input to the input values, thereby performing an uncondi- tional jump to another EDCL. 5 (PICTURE) [O161] This EDO performs a jump as per the jump EDO, but based on a condition. The condition is a BOSS process (sets of ordered EDCLs), which returns true or false. In most cases, the condition can be automatically generated as 1O a SDE. (PICTURE)PHYSICAL MANI PULATION LI BRARY [O162] An EDO in this libray must manipulate EDS's and EDINs at a physical level, where a possible input parameter 15 for a physical EDO is a physical memory address. Some of these physical EDOs are_. -Sort [O163] This EDO sorts the EDl Ns in the input EDS. 2O (PICTURE) [O164] This EDO removes duplicate EDl Ns in the input EDS. 25 _L th [O165] This EDO determines the length of the input EDS. Generate-eds 3O (PICTURE) [O166] This EDO generates an EDS for the input EDl Ns. (PICTURE)Activ te-moduIe 35 [O167] This EDO activates the module created by the EDl Ns in the input EDS. (PICTURE)Activ te-InfoBase [O168] This EDO activates the InfoBase created by the EDl Ns in the input EDS. 4O [O169] The other physical EDOs are listed and described below. Mostly these EDOs are combinations of calls to other EDOs already described. 45 [O17O] This EDO loads the input file into an allocated memory buffer, performing checksums, and returning the ad- dress of the allocated buffer. 5O [O171] This EDO will irretrievably purge previously deleted BOSS data by deleting entries in trash files. The input parameter WHERE is a UEl. If the value is null, the purge will occur for all deleted data in the InfoFrame. If a non-null value, the UEl either identifies an InfoBase (find an l BDR matching the UEl) or a module (find an MDR matching the UEl). In these cases, all deleted data in the located InfoBase or module will be purged. The DAYS parameter is optional and specifies the number of days to keep deleted information. 55 (PICTURE) [O172] This EDO will retrieve previously deleted, but not purged, information. The input parameter WHERE is a UEl. 2O
gocr ep0797806-021
(PICTURE)(PICTURE)(PICTURE) EP O 797 8O6 B1 If the value is null, the restore will consider all deleted data in the InfoFrame. If a non-null value, the UEl either identifies an InfoBase (find an l BDR matching the UEl) or a module (find an MDR matching the UEl). The DATA-UEl identifies the deleted data to be restored. If this value is null, all deleted data in the identified location is considered for restoration. Ifthe DATE parameter is non-null, it is usedtogetherwith the DIRECTlON parameterto restore occurrences of qualifying 5 information deleted on, before, or after a specific date. If the AUTO parameter is non-null, this EDO will make a best guess for all information restorations when duplicate deleted data is encountered. Otherwise, this EDO will prompt an operator with a choice of duplicate deleted information with different dates. The best guess is arrived at by grouping deleted data by date stamp, then restoring the set of data with the most recent date. 1O (PICTURE) [O173] This EDO retrieves and returns the image (or name) associated with input UEl, from an l MAGE-ETS. The search starts from current module l MAGE-ETS and expands to parent InfoBase and InfoFrame if not found at the module level. 15 (PICTURE) [O174] This EDO retrieves and returns the image (or name) associated with input UEl, from an l MAGE-ETS. The l MAGE-ETS is located in the same manner as get-image EDO. When located, the associated ETB image contents 2O are replaced with the input l MAGE. [O175] This EDO creates a new node in the input module's CNL, using the given input parameters to set the node's 25 fields. The sequence field can be supplied as ''null'' when not required. This is how information is added to BOSS at it's most primitive level. This process can be triggered from any environment, so long as the UEls provided are valid, or will have meaning. 3O (PICTURE) [O176] This EDO moves all EDl Ns in the CNL associated with input module, which are binary-equal to the input EDIN, NODE, to an associated trash CNL. 35 (PICTURE) [O177] This EDO receives a memory address, POlNTER, to start of a list of EDINs in (some) memoy. This is NOT an active EDS at this time. Also receives a COUNT to indicate the number of EDl Ns in the list. This operator creates a new active EDS (not stored) containing the list of EDl Ns, and returns a newly assigned UEl value. This operator is useful for processes that either automatically or via user input, create EDl Ns from scratch. The new EDS is always 4O added to the CNL associated with an existing module identified by input MOD-UEl. [O178] This EDO receives a MOD-UEl to identify a CNL to add nodes to, as well as a memory address, POl NTER, 45 to start of a list of EDlNs, and a COUNT of those EDlNs. The given EDlNs are then added to the identified CNL. This operator will not add binay equal EDl Ns which already exist in the CNL. 5O [O179] This EDO makes a new EDS with the same contents as the EDS given by input EDS-UEl and returns a unique UEl to the new EDS. The new EDS is created in the module identified by the input MOD-UEl (this must exist) The EDS image remains the same. 55 (PICTURE) [O18O] This EDO first generates the input EDS-UEl via generate-eds EDO, if required. Next, it calls the intersect EDO and the assigns the CNL of the input module to be the resultant ''xor-eds'' (i.e., the EDS containing EDINs in the CNL but not in the generated EDS). The generated EDS is then added to the associated trash CNL. Finally, the asso- 21
gocr ep0797806-022
(PICTURE)(PICTURE)(PICTURE) EP O 797 8O6 B1 ciated SDL and DEL entries matching EDS-UEl are moved to the associated trash files, if they exist. 5 [O181] This EDO is probably the most time and space consumptive EDO. It assumes that an EDS (EDS-UEl) was previously generated using the generate-eds EDO and then the EDINs in the memoy image were modified by some client process. At this point, the memory image of the EDS needs to be updated in the module CNL to reflect any changes made by the client process. For example, consider a module containing several procedures of a program. Each EDS can then be generated from the program module on a dynamic basis. Once generated, the procedure can 1O be dynamically modified by a programmer. Finally, the procedure is saved again in the program. [O182] Unlike specifically deleted EDINs and data, replaced EDINs and their associated data cannot be recovered, unless steps are taken by an Endo-Dynamic Editor. [O183] In concurrent environments, where multiple processes can access the same the module, it is best to let the currently used Endo-Dynamic Editor handle all such issues. Being dynamic, the EDE can be called by a BOSS client 15 process to safely load, edit, and/or save EDS's in modules. [O184] This EDOcreates anewmodule called NAME in the InfoBase given by InfoBase-UEl. This EDOfirst generates 2O a new unique UEl for the new module. Then it creates an associated MDR in the lnfoBase's MDL, with the newly generated module UEl. Next, all required module files (as shown in FIG. 14) are created (empty except for control). Now the input module name is inserted in the new module image ETS, by creating another new UEl for the image. The image UEl is also stored in the new MDR. Finally, the inputs FLAGS are assigned to the new MDR. The module can now be used as a source of any physical manipulation to add data, and later as a source of filtration to generate 25 new EDS's. [O185] This EDO first generates a new unique UEl for the new module. Next, all module files for module MOD-UEl 3O are copied to files with the same extensions and the new UEl for filename. Then, a the MDR for MOD-UEl is copied into the MDL for the input InfoBase (lnfoBase-UEl), and its self identifier set to the newly generated module UEl. The new module name remains unchanged if the NAME parameter is null. If NAME is a valid image, it will be copied to the newly copied IMAG-ETC, replacing the existing value. The image UEl value need not change. 35 (PICTURE) [O186] This EDO first appends the contents of all module files to their appropriate trash files, and then deletes all module files, as well as associated control records. 4O (PICTURE)(PICTURE) [O187] This EDO creates a new InfoBase called NAME in the currently known InfoFrame. This EDO first generates a new unique UEl for the new InfoBase. Then it creates a new IBDR in the IBDL, with the newly generated InfoBase UEl. Next, all required InfoBase files (as shown in FIG. 11 ) are created (empty except for control). Now the input 45 lnfoBase name is inserted in the new lnfoBase image ETS, by creating another new UEl for the image. The image UEl is also stored in the new IBDR. Finally, the inputs FLAGS are assigned to the new IBDR. The InfoBase can now be used as a source of any physical manipulation to add data, and later as a source of filtration to generate new EDS's. 5O (PICTURE) [O188] This EDOfirst generates anew unique UEl forthe new InfoBase. Next, all InfoBasefilesfor InfoBase InfoBase- UEl are copied to files with the same extensions and the new UEl for filename; this DOES NOT include module files for all modules encompassed by the InfoBase. Then, a the IBDR for InfoBase-UEl is copied in the IBDL, where the self identifier is changed to the newly generated InfoBase UEl. Finally the new InfoBase UEl is returned. If NAME is 55 a valid image, it will be copied to the newly copied lnfoBase lMAGE-ETC, replacing the existing value. The modules are shared by InfoBases. An InfoBase encompasses modules by including MDRs for modules in its MDL. 22
gocr ep0797806-023
(PICTURE) EP O 797 8O6 B1 [O189] This EDO first appends the contents of all InfoBase files to their appropriate trash files, and then deletes all InfoBase files. finally all associated control records at the InfoFrame level are trashed. If the CONTENTS parameter 5 is non-null, all modules encompassed by the InfoBase are trashed via calls tothe delete-module EDO, priorto all above steps. (PICTURE)PHYSICAL TORAGE OF BOSS ENTITIES 1O [O19O] Each of the BOSS entities described above, as well as all those shown in FIG. 1 4, is identified and located by a unique UEl. The identification methods have been described in the sections above. The location method is irrel- evant to the BOSS technology, and any method will do, which given a UEl, can locate the physical data associated with that UEl. The following describes one such method of physical data storage and location. [O191] Using a conventional containment storage system (e.g. UNIX, DOS, Windows), create an InfoFrame directoy 15 in a storage media attached tothe computer. The location of this directory, in the existing directory hierarchy, is recorded in the EDP program, such that when EDP is run from anywhere, it will be able to locate the directoy. Further, all files shown in FIG. 1 4 (at all levels) are stored directly in the InfoFrame directory. This flat and simple model is depicted in FIG. 32. Two distinct file-sets are distinguishable at InfoFrame level_. system and information files. System files contain those BOSS data entities which are critical to the correct operation of BOSS processes. Information files are generated 2O as a result of information stored in BOSS format. At the lnfoFrame level, there is exactly one of each file 32O2 shown in FIG. 32. At the InfoBase and module levels, there are many files. In FIG. 32, the number N represents the total number of InfoBases known to the InfoFrame, the number M represents the total number of modules in all InfoBases, and the number E represents the total number of available EDO program files. Each InfoBase, module, or EDOfilename is a string derived from the UEl value identifying that InfoBase, module, or EDO; this derived string is shown as '''', '''', or '''' in FlG. 32. Since UEls are guaranteed to be unique, there is no possibility of conflicting filenames in the InfoFrame directoy. Now given a UEl for an entity, all filenames for all files associated with that entity can be constructed and immediately located in the InfoFrame directory. [O192] Looking at FIG. 32, it should be clear that the connecting link between an IBDR and the InfoBase files, and the connecting link between an MDR and the module files, is a UEl. For example, the InfoBase UEl identifier in an 3O l BDR is used to generate the filename string associated with all files for that InfoBase. (PICTURE)BOSS I N ORMATION DELETION [O193] When information is deleted in a BOSS environment, it is always via a physical EDO involved in data deletion/ 35 restoration. These EDOs define the methods of information deletion in BOSS. The delete-node, and delete-eds EDOs do remove information such that subsequent generations will not find the ''deleted'' information. However, until ''deleted'' data is automatically or manually purged, it can be restored. Data is purged via the purgeclata EDO, or as part of automatic BOSS processing (again via purge-data), where a concurrent process is initially started upon EDP startup. This process would regularly schedule purges, based on initial user input or default values for to-purge durations. To 4O enable this, all associated BOSS files (shown in FIG. 1 4) will have an associated ''trash'' file, where a trash file is structured as its real counterpart, where in addition toa record, a date/time stamp of deletion is also stored. For example, while the entries in a real CNL are just EDl Ns, a trash CNL contains records of the following structure_. - EDIN 45 - Tl ME STAMP [O194] Entries are duplicated in a trash file until purged. To restore information, the restore-data EDO is used. The nature of this EDO enables manual as well as automatic data restoration, where all information links and data are also restored as before. 5O (PICTURE)DEFAU T COMMAND LIST [O195] As shown in FIG. 1 4, the Default Command List (DCL) is an EDCS. The DCL is executed by the EDP after InfoFrame initialization is complete. If no DCL is defined, the EDP goes into an idle state, where it waits for input EDCLs. 55 The best usage of the DCL is to implement a procedure containing an infinite loop, where the loop body activates/ reactivates any systems, InfoBases, modules, and/or EDS's required to realize a continues and changing overall proc- ess. The DCL is dynamic, in that any process activated as a result of an EDCL in the DCL can alter the contents of the DCL and return. This has the effect of altering the continuous EDP process in arbitrary ways. The DCL enables 23
gocr ep0797806-024
EP O 797 8O6 B1 the EDP to institute dynamic perpetual processing. [O196] The InfoBase and module activation processes will insert EDCLs in the DCL, as and when required. As a result, the DCL is automatically created/modified after InfoFrame initialization is complete. 5 (PICTURE) [O197] The BOSS environment requires at Endo-Dynamic Editor of some type. An EDE can use combinations of EDOs to perform the real BOSS data edit. An EDE can also use any kind of graphical user interface (GUl) or other input/output interface. An EDE includes means to dynamically interact with the EDP. An EDE includes means for dealing 1o with editing EDS's, modules, and InfoBases in a an environment where concurrent or multi- processing is possible. An EDE includes means to maintain previous versions of modified data, so that data recovery is possible via the EDE. An EDE includes means to interact with the restore-data and purge-data EDOs, such that a seamless ''UNDO'' can be implemented as a combination of these EDOs and the data recovery means of the EDE itself. An EDE includes means for displaying in ''bare mode'', where all EDl Ns are shown and regular EDS processing is not performed, optionally 15 including means to show regularly generated EDS's in separate windows as required. An EDE includes means for human as well as process interface, so that both a human and a process can operate the EDE. (PICTURE)BASIC BOSS CLIENTS 2O [O198] These are some basic general usages of BOSS that cover several important corner stones of computing. Each model embodies a different process-view of an EDS. In each case, the storage formats and some required Bond values are assumed and given. Also in each case, an SDE is supplied if applicable. [O199] Each general model presented may be used by any number of different real application programs. The basic boss client-models are_. 25 - Data-Traversal comprises data entry, storage, retrieval, with logical hierarchy/organization; - Structure-Definition comprises data structure definition and usage; - Program-Execution comprises program, procedure, parameter, variable, code, lines, etc., storage and usage (i. e., execution); 3o - Analyze-Data comprises the function of deriving meaning and output any required actions given assumption data, and new input. [O2OO] Note that a DB interface is different from a program execution interface only in how it processes the data. It's just a matter of perspective, on the same BOSS data (namely EDl Ns in various lists and orders). Even the analyze- 35 data client can express all of its required rules and data as EDl Ns (and ETS data). There is a difference between the first three above listed clients, and the last. The first three are almost entirely composed of calls to EDOs to accomplish their real processing. The only processing in such clients requiring additional code (to EDO-calls), is the code required for a particular interface required by the client. All real processing can be accomplished using EDOs. This means given an operating EDP and its required initial data, these client models can be constructed almost immediately, especially 4o in a GUl environment wherein interface construction is vastly simplified. The last client listed above requires further code (preferably BOSS executable procedures) to perform deductive, and possibly heuristic processing. Each are of these clients are described in ensuing subsections. 45 (PICTURE) [O2O1] Many kinds of applications fall under this client-model. In fact, this encompasses any process which requires data storage, retrieval, and maintenance, where data exists in some hierarchy/organization, and where such data is then presented to a user in some depiction of the data organization. Examples of a Data-Traversal client are_. 5O - a system which maintains hierarchical data in a directory-like organization, - a system which defines and processes data structure definitions, and - a dynamic menu (or window) definition, traversal, processing and maintenance system. [O2O2] Further, many BOSS clients will need to incorporate a Data-Traversal client (as well as othercode) to automate 55 the tasks of data storage, location, retrieval, and maintenance. Examples of this are_. a system which dynamically processes (interprets) program code, and any kind of database [O2O3] The tree 33O2 of FIG. 33A is some hierarchy of data, where, A, B, C, and E are logical entities, providing hierarchy for data entities identified by F, G, H, l, and K. This tree 33O2 could be a directoy tree, or a menu-tree, or 24
gocr ep0797806-025
(PICTURE) EP O 797 8O6 B1 any other kind of tree (a tree is hierarchical by nature). Although not shown, the data organization is not limited to trees, and any kind of graph, or other more compound data organization scheme can be used. In FIG. 33B, the leaf nodes G, H, l, K of the tree 33O2 do not appear as Subjects in the EDS 33O4. If a process were to filter for EDl Ns with Subject fields matching these IDs, the process would get an empty EDS for all of them. In fact, just such a process can be 5 used to determine all leaf nodes in a tree-hierarchy. [O2O4] Bond can be used to establish EDl N-typing, and subject-attribute relations. For example in a database, the statement_. ''Car55 is Crimson'' can be expressed as a single EDl N, as follows_. 1O [O2O5] By using SDEs, which call EDOs, simple and complex data queries can be constructed that apply to all kinds of data for any kind of BOSS client. For example, a system which maintains a directory-like data organization, would 15 ConStrUCt SDES WhiCh Will loCate fileS and direCtorieS, While a databaSe ConStrUCtS SDES to loCate data With defined criterions and constraints. (PICTURE)DATA STRU TURE DEFINITION AND USAGE _o [O2O6] When adopting a method for defining data structures, usually there are two basic types_. primitive (e.g. string, signed 32-bit number, etc), and compound, where a structure's elements are composed of primitives and other com- pounds. As described above, a BOSS defined record structure can have any kinds of elements whatsoever. This is mainly because a UEl uniquely identifies an entity anywhere. The other major factor is the already discussed inter- changeability of the source and target fields in the EDIN. _5 [O2O7] FIG. 34 shows an example minimal rendition of BOSS oriented data structure definition. At the top of FIG. 34, two compound records, ''A'' 34O2 and ''Z'' 34O4, are shown. Record ''A'' 34O2 contains an element with data type ''Z''. Both compounds contain elements with primitive data types. At the bottom half of FIG. 34, the corresponding EDS's 34O6 and 34O8 for each record is shown. The Bond field is used to define a binary relation as in any BOSS application. The key is what those Bonds are. In this case, Bonds describe elements of data definitions, primitive/compound data 3o types, and size. [O2O8] The EDS shown for record ''A'' 34O6 (and record ''Z'' 34O8) would only contain all shown EDINs, if complete structure traversal is performed. That is, all Attributes of a Subject match are themselves Subject-matched, until empty sets are reached. At each step, EDl Ns are accumulated. Such a traversal process could be set to terminate at any Ievel in the data hierarchy. If set to ''maximum'' or ''all levels'', the EDS 34O8 shown for record ''Z'' would be part of the 35 resultant EDS 34O6 for record ''A''. [O2O9] In the EDS 34O6 shown for record ''A'', the Attributes of four EDl Ns with a Subject of ''A'' represent the four elements of record ''A'' 34O2. The element-sequence field in these EDINs establishes the order of elements in the data structure. The rest of the EDl Ns describe the characteristics of each of the elements of record ''A'' 34O2. [O21 O] Now consider element ''B'' in record ''A'' 34O2. This element generates a total of three EDl Ns_. one with a 4o Subject of ''A'' and Attribute of ''B'', and two with a Subject of ''B''. The first signifies ''B'' as an element of ''A''; the next two give the data type and size of element ''B''. The Attribute value shown as ''STRING'' would be a UEl value in reality. Aside from the shown Bonds, any number of other Bond values could be implemented to provide more detailed de- scriptions of an element (e.g. associated process, input-coordinates, display-attributes, etc). [O211] Aside from employing a data-traversal BOSS client, the Structure-Def BOSS client uses SDEs to dynamically 45 construct and generate data structure definitions. When an element in a structure is created where the type is the bond ''call'' or ''execute'', it is a simple matter to construct an EDCL from all element EDl Ns of equal subject. The EDCL can then be executed in the normal manner. Since the EDP is capable of executing EDCLs from anywhere and at any level (limited by stack size), this enables BOSS defined data structures to contain processes which are executed when the data definition is accessed, regardless of the calling process, or any other active process. 5O (PICTURE)PROGRAM STORAGE AND EXECUTION [O212] Using EDINs, the EDP, and EDOs, it is possible to store, maintain, and execute any kind of program. As described above, under the EDO section, a given set of EDOs can form the instruction set for a programming language, 55 when all such EDOs are also made available as bonds. A BOSS interpreted program can use the following EDl N implementations_. - store an EDCS for each procedure, 25
gocr ep0797806-026
(PICTURE) EP O 797 8O6 B1 - use a Structure-Def client model to store and process program data structures, or - use a Data-traversal client model to save, locate and retrieve program code and data. [O213] Since a client can also create construct and supply its own bonds and EDOs (tagged as ''user''), any missing 5 bonds/EDOs can be implemented by the client and seamlessly integrated into the BOSS environment. The new EDOs will be processed as per all EDOs, by the EDP. Further, the associated bonds can now be used to insert new code lines (EDCLs) into the interpreted program. [O214] BOSS is ideally suited to interpreted programs, although it does support compiled programs as well. At a minimum, a BOSS oriented program must contain definitions of_. data-types, variables, parameters, procedures, and 1O lines of code. If imperative programs are required, the associated module must be tagged as ''static''. This indicates to the EDP that no EDS generation should take place and that the CNL should be loaded and taken for the EDS(s) in question, in some order. While interpretive programs can be directly executed by the EDP, imperative programs require a ''program execution'' system, program or shell which facilitates the execution of the imperative BOSS program. Such a shell would construct/establish EDCL groups for the EDP, where the last EDCL in each group returns control back 15 to the shell, until all program processing is complete. Even using a static module, the normal control flow EDOs may not operate correctly in an imperative program. This depends on whether the operator performs a change of context or not. For example, the ''edp-pop'' EDO changes the next EDCL to be executed by the EDP, thereby changing the context of current processing. Such EDOs almost always cause a regeneration upon successful termination, to ensure updated data. By introducing a lot of controls in such EDOs (and complicating them), it is possible to detect an imperative 2O process, and return the parameters of would-be context change, and it's associated action(s) to the calling process. This enables the calling process to always be in control of process flow (except for fatal errors and the like). [O215] The program called CCOPY, shown in FIG. 35 as 35O2, performs a Conditional Copy. It has two parameters, and two variables. It also has one procedure which is not shown; only called. The parameters of CCOPY are two UEls to two distinct files_. Source File (SFl LE) and Target File (TFl LE). First, the program 35O2 gets the current date of each 25 input file and stores the dates in SDATE and TDATE. This is accomplished by calls to the GET-DATE program proce- dure. This procedure is not shown, but it should be easy enough to picture it as an OS call. Finally, the program 35O2 compares the two dates and if the source file date is greater than that of the target file, it copies the source file, over- writing the target file. Both the ''if'' and ''copy'' are accomplished by OS or shell calls (i.e., native to BOSS interface). [O216] The EDS 36O2 shown on FIG. 36 depicts the required EDINs 36O4 to define and be able to execute, the 3O CCOPY program. The sequence field establishes an order, where a set of EDCLs are realized. To further define a parameter or variable, the same methodology as for a Structure-Def client is used. For example, to define the SFl LE parameter, the following two EDl Ns 36O4 could be added to the program module CNL_. 35 [O217] The SDEs can be used to dynamically generate components of the program, from the CNL. As a result, all 4o program components remain dynamic. So if a data structure changes, the processes using that structure, or data in that structure, will immediately experience the effects ofthe change. As with anydynamic interpreted program, sufficient safeguards must be taken to enure only changes without destructive effects take place. (PICTURE)CONCLUSION 45 [O218] This concludes the description of the preferred embodiment of the invention. In summary, [O219] The foregoing description of the preferred embodiment of the invention has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Many modifications and variations are possible in light of the above teaching. It is intended that the scope of the 5o invention be limited not by this detailed description, but rather by the claims appended hereto. Claims 55 1 . A method for dynamically organizing and processing data in a computer having a memoy and a data storage device coupled thereto, the method characterized by the steps of_. (a) generating an information structure and relationship in the memory of the computer as one or more Endo- 26
gocr ep0797806-027
EP O 797 8O6 B1 Dynamic Sets (EDS), the EDS comprising a list of one or more Endo-Dynamic Information Nodes (EDINs), the EDl Ns each representing an atomic component of data, and the atomic component of data comprising a subject identifier identifying a topic, an attribute identifier identifying information pertaining to the topic, and a bond identifier defining a relationship between the subject and attribute identifiers; 5 (b) associating each bond