FR2833095A1 - Personal digital assistant for storing/processing relational databases, comprises copying field key format information from first to second table descriptors and establishing links between tables - Google Patents

Abstract

The personal digital assistant has a microprocessor (1), read only memory (2), user memory (3a) and data acquisition means (5,6). During database construction data defining field key formats is copied from a first table descriptor to a second table descriptor such that each field key has the same format in the two tables. A linkage is then made between the two tables to identify identical memory space keys

Description

<Desc / Clms Page number 1>

The invention relates to a device for the digital processing of data of the PDA ("Personal Digital Assistant") type: handheld computer such as those marketed under the names "PDA" (PDA). Palm &commat; Pocket PC e, PSION (B) or EPOC &commat; as well as any other portable computing device (electronic diaries, communicating mobile phones and other terminals) and / or low processing and storage capacity (on-board computer per vehicle, metering and meteorological data terminal, etc.). Such a PDA device comprises: processing means comprising at least one microprocessor (equipped with at least one internal memory called register); means, said non-volatile memory read-only memory directly accessible read-only by the processing means, and consisting exclusively of one or more nonvolatile electronic memory (s); means, said user memory, storage directly accessible in read and write by the processing means for storing programs and / or user data (programs and / or personal data specific to a user of the device and / or a particular use of the device, data acquired, that is to say, entered by a user and / or measured by measuring means of the device or related to the device-meteorological data for example-and / or transmitted to the device, etc. .), and consist exclusively of one or more electronic memory (s); autonomous power supply means; means forming a user interface comprising at least user data acquisition means and means for restoring said data; software means stored in the read-only memory and / or in the user memory adapted to manage at least one relational database for organizing the storage of user data in the form of table records each comprising a plurality of fields.

<Desc / Clms Page number 2>

 A PDA device is in particular free of magnetic or optical recording medium mass memory requiring the intervention of a reading device using mechanical means: the user data and programs are in fact stored in an electronic memory, which can be, for example, a non-volatile electronic memory type "flash" (rewritable non-volatile electronic memory without specific equipment, direct access and relatively fast), or a volatile memory electronic memory type "RAM" or memory "SRAM" (device RAM, direct access and very fast), the device before, in this case-which is the most common-, be constantly supplied with electrical energy by a battery operating off.

 Professional or personal software applications are mostly based on databases, including relational databases. The explosion of the PDA market has created the desire to create new applications for these devices or to implement existing applications.

But traditional database technologies are unsuitable for PDAs, due to the low storage capacity and the limited processing power of the latter.

 A relational database consists of tables, each formed of a plurality of records of the same nature (each corresponding to a row of the table) and a plurality of fields (each corresponding to a column of the table), each field receiving a specific value for each record. The fields can be of various type (character string, boolean, date, time, integer, decimal, binary ...), fixed size or variable size. The fields are ordered according to a predetermined order (order of the columns of the table).

 Two tables are linked together by key fields common to both tables. The same table can be linked to several tables: it therefore has common key fields with a first table, common key fields with a second table, etc., the key fields common to the first table may be different from those common with the first table. second. In the following, when two linked tables are concerned and when there is no possible ambiguity, the term "key fields of the first one of said tables" key fields of said first table

<Desc / Clms Page number 3>

 common with the second table concerned. The linked records of two tables are those for which the values of the key fields of the first table are equal to the values of the key fields of the second table. Making a join between the first and second tables consists in comparing the values of the key fields of the records of the first table with the values of the key fields of a record of the second table (or vice versa), and selecting and storing at less information representative of the fact that one or more record (s) of the first table is (are) related to said record of the second table.

To perform this operation, the current PDA devices are programmed, according to the so-called SQL model ("Structured Query Language", structured query language for the creation, management and consultation of relational databases, implemented on known PDAs), to perform an operation commonly called "dot product" of the two tables, consisting of creating and storing in a working memory a search table whose each line comprises a row of the first table (that is to say the values of the different fields of a record of the first table) and a row of the second table following one another, and to search, among the rows of the search table thus created, those simultaneously checking the value equalities ( Table 1 - field i) = value (Table 2 - field i), i taking successively the values 1 to n, where the fields field. l to field. n are the key fields common to said first and second tables, respectively denoted Table-field. i and Table2 - field. i. The command of such an operation requires the writing of the following query lines in SQL language: "select * from Tablel, Table2 where Tablel ~ field.l = Table2 ~ field.l and Tablel ~ field.2 = Table2 ~ field. 2 and Tablel ~ field n = Table2 ~ field n
If the first table has 10,000 records and the second table has 100,000 records, the dot product of these two tables will be 109 rows. Such an operation requires a memory and a processing capacity often higher than those available in the current microcomputers of the personal computer type of

<Desc / Clms Page number 4>

 office. This is why SQL search engines all have a query optimizer to reduce the number of rows before making the scalar product.

However, despite the existence of these optimizers, the SQL model is totally unsuited to PDAs, which have resources well below those of said microcomputers.

Processing times are therefore often incompatible with the use of SQL engines.

 It should be noted, moreover, that it is common for the corresponding fields Tablel ~ field. i and Table2 ~ field. i (taking successively the values 1 to n) have different recording formats (in terms of type and / or size).

To check each of the previous equalities, the known devices are therefore adapted to read the memory spaces occupied by the fields Tablel ~ field. i and Table2 ~ field. i, copy the electronic data (series of bits) they contain into a working memory internal microprocessor called register, converting said electronic data into data that can be compared with each other, that is to say, rewrite the values of the Key fields in the register according to a common registration format allowing their comparison. A step of conversion and interpretation of the data recorded in the key fields of the first and second tables is therefore necessary to achieve their join, consuming memory and additional time.

 It should also be noted that SQL search engines have historically been created for server-type devices, with a higher processing capacity than PDA devices and equipped with an optical or magnetic recording medium (peripheral memory) such as a hard disk) for storing the database. Access to such a memory is particularly slow, a search performed on the database of a server involves the loading (copy) in a working memory of the "RAM" type of data (tables) involved in the search, in to perform the various search operations (including the scalar product of a possible join operation) in this working memory access and fast processing. Previous PDAs incorporating databases have database and search engine management systems utilizing this method. In these devices, the results of a search, stored in a working memory and provided to the user, therefore correspond to a copy of the value of one or

<Desc / Clms Page number 5>

 several fields (depending on the desired parameters, entered by the user) records checking the search criteria. This copy is dissociated from its original record (s). It is therefore impossible, in the previous PDA devices, to search from the results of a previous search, the relevant records are not known to the processing means.

 The aim of the invention is to overcome these disadvantages by proposing a device of the PDA type making it possible to simply, safely and quickly make joins between tables of a relational database. In particular, the invention aims to provide a device capable of performing such joints almost instantaneously despite a processing speed and a limited working memory.

 Another objective of the invention is to propose a construction of relational databases adapted to the specific features and performances of the PDAs and making it possible to go up from one field of a table to a plurality of fields of another table.

 Another objective of the invention is to propose a device of the PDA type equipped with a search engine using a simplified query language, requiring for its operation and the storage of its applications a reduced electronic memory space, and making it possible to realize single-criteria or multi-criteria searches on several tables in the database ten to one hundred times faster than previous PDA devices.

 Another objective of the invention is to propose a device of the PDA type able to carry out a search for data on the basis of the results of a previous search.

 The invention relates to a device (PDA type) for digital data processing, comprising processing means comprising at least one microprocessor; means, said non-volatile memory read-only memory directly accessible read-only by the processing means, and consisting exclusively of one or more nonvolatile electronic memory (s); means, said user memory, storage directly accessible read and write by the processing means for storing programs and / or user data, and consisting exclusively of one or more memory (s)

<Desc / Clms Page number 6>

 electronic (s); autonomous power supply means; means forming a user interface comprising at least acquisition means (capture, measurement, reception ...) of user data and playback means (display, sound, transmission ...) of said data; software means stored in the read-only memory and / or in the user memory adapted to manage at least one relational database constructed according to specificities defined by a person, called an operator, and making it possible to organize the storage of the user data in the form of table records each comprising a plurality of fields, said database comprising at least two tables linked together having, for this purpose, one or more common fields, called key fields (thus designating only the key fields common to the two tables ), which occupy, for each record of said tables, a memory space said key storage space of the recording. The device according to the invention is characterized in that it is suitable for: during the construction (definition of tables, fields, relational rules, etc.) of the database, recording in the user memory a format of the records of one-first-of the two tables, as defined and entered by the operator, in a descriptor said descriptor of the first table, and copy in the user memory, from the descriptor of the first table, the information defining the format of each of the key fields of said table (those common with the second table) to a descriptor of the second table defining a format of the records of the other-second-of said tables, so that each key field is defined with the same format in both tables, - allow the realization of at least one join between said tables in order to determine the record (s) of the-first-of said tables linked to a record the other-second-of said tables, by directly comparing the electronic state of the key memory spaces of the recording of the second table and of at least one record of the first table, and storing in the user memory at least an information, called join result, representative of the fact that one or more record (s) identified (s) of the first table has (s) a key memory space electronically identical to that of the registration of the second table.

<Desc / Clms Page number 7>

 It should be noted that the term "descriptor" of a table designates both the data describing the format of the records of the table and the memory block in which these data are stored. The descriptor describes the type and the size of the fields constituting said table, in the predetermined order of said fields. A reference to point to the descriptor is inserted into each record of the table (at the beginning of the record), to indicate to the database management system which table the record belongs to and in what format it has been memorized, in particular to enable him to locate the different fields within the recording. The device according to the invention is therefore adapted to import the format of the key fields from the descriptor of one of the tables to the descriptor of the other table. In a preferred version of the invention, this import is done automatically: during the construction of the database, the operator defines the first of the two tables (type and size of his fields) by means of his descriptor , it defines the other second of the tables by means of its descriptor, excluding the fields common to the first table which serve only to establish a relation between the two tables, it defines finally a unique field, said field relation first and second tables, formed by the combination of the key fields, as defined in the descriptor of the first table; the device according to the invention is advantageously adapted to memorize the definition of the relation field, as entered by the operator, and to copy, from the descriptor of the first table, the information defining the format of each of the fields composing said relationship field to the descriptor of the second table, and this automatically following the definition of the relationship field of said tables.

 By "direct comparison of the memory spaces" is meant a physical comparison of said memory spaces, that is to say a comparison of the electronic state (0 or 1) of the corresponding different bits of said spaces (two corresponding bits occupy the same place within the two memory spaces), without prior operation of conversion and interpretation of the stored data. Such a comparison, which is particularly simple and rapid, is made possible by the fact that, thanks to the invention, the corresponding fields of the two linked tables necessarily have the same format, and that consequently, in case of equality of values, they contain, we have

<Desc / Clms Page number 8>

 necessarily electronic identity (same number of bits and same state (0 or 1) of corresponding bits) corresponding memory spaces.

 In a device according to the invention, making a join between two tables is therefore a particularly simple and fast operation, consuming little memory. It comes down to a comparison operation of two memory blocks (copy memory blocks in two registers memories, then comparison of the electronic state of the copied blocks, without formatting the aforementioned blocks or masking unused bytes). In addition, thanks to the invention, the definition of the queries making it possible to carry out searches on the database is greatly simplified, as will be explained hereinafter.

 The device according to the invention is also advantageously adapted to calculate and store, for each record of the database, an identifier allowing direct access to said record in the user memory. The identifier may be an integer of fixed size. It is listed and stored in an address table mapping the physical address (or possibly a logical address) of each record and its identifier, and thus to locate the location of the record in the user memory.

 Advantageously and according to the invention, the device is also adapted to perform, when performing a join between the first and second tables, one or more of the following operations: - compare byte by byte, successively, the spaces key memories of the recording of the second table and a record of the first table. As a variant, a comparison word by word (1 word = 2 bytes), or long word by long word (1 long word = 4 bytes), or double by double (1 double = 8 bytes) is also in accordance with the invention. ; in the case of a comparison byte by byte-respectively word by word, long word by long word, or double by double-, interrupt the comparison of the key memory spaces of the recording of the second table and a record of the first table, as soon as two bytes-respectively two words, two long words, or two double-correspondents compared present a different electronic state;

<Desc / Clms Page number 9>

 select for each key field, according to the format (and in particular the type) of said key field, a comparison mode (byte by byte, or word by word, or long word by long word, or double by double). If one of the key fields to be compared is a real or integer number stored on 2 bytes (respectively 4, or 8 bytes), the device is adapted to carry out a comparison word by word (long word by long word, or double by double) key memory spaces-corresponding to said field-of the recording of the second table and a record of the first table. If one of the key fields is a string, the device is adapted to perform a byte comparison of the key memory spaces of said records corresponding to said key field.

 storing, as a result of the join, the identifier of the records of the first table having a key memory space which is electronically identical to that of the record of the second table; selecting successively the records of the first table whose key memory spaces are to be compared with that of the recording of the second table, in their order of storage in the user memory. Alternatively, the device is adapted to randomly select said records; compare key field by key field the key memory spaces of the record of the second table and a record of the first table; and / or interrupting the comparison of the key memory spaces of the recording of the second table and of a recording of the first table as soon as the memory spaces of two corresponding key fields compared present a different electronic state; and / or in the case where at least one of the key fields is an indexed field, successively selecting the records of the first table whose key memory spaces are to be compared with that of the record of the second table, by performing a dichotomous search on the index of the indexed key field. To do this, the device according to the invention is adapted to develop, update for each new record entered a predefined table by the operator, and store in the user memory, an index of a field of said table, said indexed field, predefined by the operator, said index comprising the identifiers of the records of said table stored one at the

<Desc / Clms Page number 10>

 following from the others so that the values of the indexed field of said records are ordered, that is to say ordered according to an order predefined by the operator (alphabetical order for alphabetic character strings, ascending or descending order for numbers whole or decimal, chronological order for dates, etc.).

 In this case, the device is adapted to select each record of the first table to be compared with the record of the second table by performing a dichotomic search on the indexed field: if the comparison of the record of the second table with a selected record of the first table indicates that the value of the indexed key field of said record of the first table is lower (respectively greater) than the value of the key field of the record of the second table, the device is adapted to select, as a new record to compare, the one whose identifier is located, in the index, in the middle of the list of identifiers delimited by the identifier of the previously selected record and the last (respectively the first) identifier of the index (in the case of a ranking of the values of the indexed field in ascending order in the index).

 The combination of the key fields constitutes a primary key of the first one of the two tables, and a foreign key of the second. By primary key is meant a field or a combination of fields making it possible to individually identify each of the records of a table, that is to say taking a different value for each record. By foreign key of a second table is meant a field or a combination of fields making it possible to establish a relation with a first table: the fields used as foreign key of the second table correspond, for this purpose, to the fields defining a primary key of the first table.

According to the invention, during the construction of the database, the device is preferably adapted to copy, from the descriptor of the first table, the format (type and size) of the fields defining the primary key of said first table to the first table. descriptor of the second table. It is further adapted to define the combination of the fields thus imported as being a foreign key of the second table.

 In the case where the combination of the key fields constitutes a primary key of the first table and a foreign key of the second (and that the join must

<Desc / Clms Page number 11>

 allow to define the recording of said first table linked to a given record of said second table), the device is advantageously adapted to interrupt the comparisons between the key memory space of the recording of the second table and the key memory spaces d records of the first table, as soon as a record of the first table having a key memory space electronically identical to that of the recording of the second table has been identified (this record is indeed the only one verifying the identity), i.e., in other words, as soon as information representative that a record of the first table is related to the recording of the second table has been stored.

 The record of the second table, the starting point of the join of the second and first tables, is usually the result of a query included as part of a user-initiated search of the database. or during the execution of an application previously programmed by the operator. According to the invention, the device is suitable for: in response to a search made in the database from search criteria, entered by a user or an operator and comprising at least designation information of a table, said search table, a field of said table, said search field, a comparison operator, a reference value, and another table, called the result table, select the record (s) (s) of said search table verifying said search criteria (this is the record (s) for which the search field value is equal, higher or lower (according to the comparison operator) to the reference value), - make at least one join between the search table and the result table to determine the record (s) of the linked result table (s) at each record, previously selected, of the search table (the device performs as many join operations as the selected search table records), - store, in the user memory, the identifier of the record (s), previously determined (s) ), of the result table.

 A search whose search criteria are limited to those listed above is called a single-criteria search (the search concerns only one search

<Desc / Clms Page number 12>

 search field on which we perform a single comparison) on several tables.

Note that the device also makes it possible to perform single-criteria searches on a single table (these searches do not require any join operation), or even multi-criteria searches (involving several reference values and possibly several comparison operators) on multiple tables, combining multiple single-criteria searches on one or more tables, and performing one or more joins as needed.

 The device according to the invention thus provides the database management software means, as a result of a search, means (identifiers) for accessing the searched records. This feature is particularly advantageous in view of the fact that the database is stored on a fast and direct access memory. It also makes it possible to search based on the results of the previous search.

 It should be noted that, thanks to the invention, the realization of a join between the first and second tables can be programmed by means of a single command line, in a simplified specific query language which uses in particular the relationship-formed field by combining the key fields common to the first and second tables-defined during the construction of the database, as will be explained below. The device according to the invention comprises for this purpose software means adapted to compile and / or assemble and / or edit relationships and / or interpret applications using this specific query language.

 Advantageously and according to the invention, the user memory is chosen from a random volatile electronic memory of the RAM (Random Access Memory) type, a static volatile electronic memory of the SRAM (Static Random Access Memory) type, a non-volatile memory type electronic memory. flash (memory "EEPROM": "Electrically Erasable Programmable Read Only Memory", erasable in blocks), a memory space of one of these memories dedicated to storing data and user programs. In all cases, it is a direct access memory by the processing means, that is to say an electronic memory.

<Desc / Clms Page number 13>

 The invention extends to a method of constructing a relational database for storing and processing user data in a device (of the PDA type) comprising processing means comprising at least one microprocessor; means, said non-volatile memory read-only memory directly accessible read-only by the processing means, and consisting exclusively of one or more nonvolatile electronic memory (s); means, said user memory, storage directly accessible read and write by the processing means for storing programs and / or user data, and consisting exclusively of one or more electronic memory (s); autonomous power supply means; means forming a user interface comprising at least user data acquisition means and means for restoring said data; software means stored in the read-only memory and / or in the user memory adapted to manage at least one relational database constructed according to specificities defined by an operator, said database comprising at least two tables linked together by one or more common fields, called key fields, which occupy, for each record of said tables, a memory space said key storage space of the recording. The method according to the invention is characterized in that it comprises the following steps: storage in the user memory of a format of the records of the first one of the two tables, as defined and entered by the operator , in a descriptor of the first table, - storage of the definition, as entered by the operator, of a field, said relationship field of the first and second tables, formed by the combination of the key fields (those common to both tables and already defined in the descriptor of the first table) and intended to make it possible to establish a relationship between the two tables (the definition of this relationship field is equivalent to the definition of relational rules linking the two tables), - copy of the information defining the format (type and size) of each field composing the relationship field, from the descriptor of the first table to a descriptor of the second table, defining a format of the records of

<Desc / Clms Page number 14>

 the other-second-table, so that each key field is defined with the same format in both tables.

 According to the invention, the method of constructing the database preferably comprises a step of calculating and storing, for each record of said database, an identifier making it possible to directly access said record in the user memory.

 It also comprises the preparation, updating for each new record entered of a table predefined by the operator, and the storage in the user memory of an index of a field of said table predefined by the operator. said indexed field, said index comprising the identifiers of the records of said table stored one after the other so that the values of the indexed field of said records are classified according to an order predefined by the operator. The classification of a new record in the index is done automatically by dichotomy.

 The invention also extends to a method of making a join between two linked tables of a database in a device of the PDA type according to the invention, said base being constructed according to the aforementioned construction method according to the invention. , to determine the record (s) of a first of said tables linked to a record of the other second of said tables.

Advantageously and according to the invention, this method comprises a step of direct comparison, by the microprocessor processing means (s), of the electronic state of the key memory spaces of the recording of the second table and of at least one recording of the first table, and a step of storing in the user memory at least one piece of information, called the result of the join, representative of the fact that one or more record (s) identified (s) of the first table present ( nt) a key memory space electronically identical to that of the recording of the second table.

 The method according to the invention for making a join between a first and a second table advantageously has one or more of the following characteristics:

<Desc / Clms Page number 15>

 the comparison of the key memory spaces of the recording of the second table and of a record of the first table is done byte by byte successively. Advantageously, this comparison is interrupted as soon as two corresponding bytes compare have a different electronic state. Alternatively, the comparison is performed word by word (or long word by long word, or double by double), and is interrupted as soon as two words (or long words, or double words) corresponding compared present a different electronic state; the comparison of the key memory spaces of the registration of the second table and of a record of the first table is effected key field by key field, and in the case where at least one of the key fields is a field indexed, the processing means successively select the records of the first table whose key memory spaces are to be compared to that of the registration of the second table, by performing a dichotomic search on the index of the indexed key field; as a variant, the processing means successively select the records of the first table whose key memory spaces are to be compared with that of the record of the second table, in their order (or reverse order) of storage in the user memory; in the case where the combination of the key fields constitutes a primary key of the first table and a foreign key of the second table, the processing means interrupt the comparisons between the key memory space of the registration of the second table with the spaces key storage memories of the first table, as soon as a record of the first table having a key memory space electronically identical to that of the recording of the second table has been identified (and information representative of the fact that a record of the first table is linked to the record of the second table has been stored); the processing means stores, as a result of the join, the identifier of the records of the first table having a key memory space electronically identical to that of the record of the second table.

 The method of performing join is notably implemented automatically in the device in response to a search carried out in the database of

<Desc / Clms Page number 16>

 data from search criteria entered by a user or by an operator, said criteria comprising at least a first table, called result table, a field, called search field, of a second table (different from the result table) ), called search table, a reference value and a comparison operator, the join performed on said search table and result table.

 The invention extends to a search method, in a database built according to the aforementioned construction method according to the invention, from search criteria entered by a user (comprising at least one result field). The search method according to the invention comprises, where appropriate, the realization of at least one join between a first and a second table according to all or some of the features described above, the storage in the user memory, as a result of the search, the identifier of each of the records verifying the search criteria, and the display on the display screen of the value of the result field (s) of each of said records.

The search method according to the invention is implemented by means of a simplified query language (notably for the programming of joins) using the relations fields of the linked tables.

 The invention also relates to a PDA-type computing device, a relational database construction method, a method of making join between two tables of a database constructed according to said method of construction, a search method in a such database, characterized in combination by all or some of the features mentioned above and below.

 Other objects, features and advantages of the invention will become apparent on reading the following description which refers to the appended figures (and to the tables below) representing preferred embodiments of the invention given solely for the purposes of the invention. non-limiting examples. Figure 1 is a schematic view of a device according to the invention; Figure 2 is a diagram illustrating the organization of the various components constituting a device according to the invention; Tables 1 and 2 are examples of tables of a relational database implanted in a device according to the invention; FIG. 3 represents a part of a memory block of a recording of the

<Desc / Clms Page number 17>

 table 1; FIG. 4 represents a part of a memory block of a record of the table 2.

 A PDA according to the invention, illustrated in Figures 1 and 2, comprises a microprocessor 1, small in size and cost, but also low processing capacity (about ten times lower than that of a standard desktop). The microprocessor 1 is connected to a non-volatile electronic memory 2 of the "ROM" type directly accessible read-only by the microprocessor, in which are stored general operating programs (and in particular an operating system, of the type of those marketed under PalmOS names &commat;, PocketPC &commat;). It is also connected to a volatile electronic memory 3 of the "SRAM" type directly accessible for reading and writing by the microprocessor, a part of which 3b, called working memory, is dedicated to the execution of operations related to the operation of the device and comparison operations and / or calculation for example, and another part 3a, said user memory, is dedicated to the storage of the database and any applications added by the user. Said data and user applications are thus instantly available as soon as the device is put into operation (unlike a desktop computer whose applications and data are stored in a hard disk and must be loaded into the working memory in order to be used), and instantly saved when they are modified (no copying operation on a hard disk is necessary ...).

 Since the memory 3a for storing the database (user memory) of the device is volatile, it must constantly be supplied with electrical energy by a battery 13, which may be an alkaline battery or, preferably, a rechargeable battery. lithium or nickel-cadmium for example. In order to avoid the loss of the data stored in the RAM memory 3 in the event of failure or of stopping of said batteries, the device comprises means of communication with external devices in order to allow the export of said data to these external devices, as well as all synchronization operations (harmonization of the databases stored in the PDA and in the external devices): a connection port 10 of the "USB" port type ("Universal Serial Bus") to allow the wired connection of the device according to the invention to a desktop computer 8 and / or a modem 9, means

<Desc / Clms Page number 18>

 infrared transmission 12, and means 11 for transmitting waves for transmitting files to a desktop computer via an internet server (by sending an email).

 The device is furthermore equipped with a display screen 4, preferably a liquid crystal display, and data input means such as a keyboard 5, control buttons 7 programmed to trigger certain functionalities, and a pen 6b for handwritten input of data on a touch screen 6a. The device comprises software character recognition means stored in the "ROM" memory 2 in order to allow the handwritten input of data by means of said pen 6b.

 A device as illustrated in Figures 1 and 2 is distributed to each physician of a hospital, to enable said physician to note information about his patients during consultations it provides, to establish his care records to consult if necessary a database listing the various existing drugs, their composition and properties, etc. Daily, the doctor synchronizes his database with a general database, stored on a high-capacity computer, gathering data collected and / or used by all doctors in the hospital.

 A first Patients table contains general information about each patient in the hospital. A second table Exams (not described) gathers medical information concerning each patient, and allows a daily monitoring of its state (date, temperature, tension, various symptoms ...). A third table Care identifies the sheets of care established by each doctor. A fourth table Prescriptions (not described) allows storing prescriptions prescribed by each doctor (drugs, dosage, duration of treatment, additional tests to be performed, ...).

<Desc / Clms Page number 19>

<Tb>
<Tb>

Patients <SEP> Care
<tb> Name <SEP> Care ~ ID <SEP> (number <SEP> of <SEP> the <SEP> sheet <SEP> of
<tb> First name <SEP> care)
<tb> Person <SEP> - <SE> SS <SEP> 1 <SEP> (sex) <SEP> PERSONALIZEDSSL <SEP> (sex)
<tb> Person ~ SS2 <SEP> (year) <SEP> Personal Care ~ SS2 <SEP> (year)
<tb> Person ~ SS3 <SEP> (months) <SEP> Personal CareSS3 <SEP> (months)
<tb> PersonSS4 <SEP> (department) <SEP> Care ~ person ~ SS4 <SEP> (department)
<tb> PersonSS5 <SEP> (number) <SEP> PersonalCareSS5 <SEP> (number)
<tb> Address <SEP> DoctorID
<tb> PostalCodeAddress <SEP> Date
<tb> Adresseville <SEP> OrderID
<tb> Telephone <SEP> Code <SEP> consultation
<tb> Profession <SEP> AmountHT
<tb> Group ~ Sanguine
<tb> Mutual <SEP> ID
<Tb>
Note: the social security number is advantageously divided into five fields Person ~ SSl to Person ~ SS5 each corresponding to information on the patient (sex, year, month and department of birth, number).

 Figure 3 shows a portion of the memory block occupied by a record 100 of the Patients table. The 12 bytes 111 to 134 are used to store the Name field; byte 141 corresponds to the Person field ~ SSl (integer stored on 1 byte); byte 142 corresponds to the Person field ~ SS2 (integer of one byte); byte 143 makes it possible to store the field Person ~ SS3; byte 144 stores the Person field ~ SS4; the 10 bytes 151 to 172 are reserved for the First Name field; bytes 173 and 174 correspond to the field PersonSS5; on the next 16 bytes (partially represented) is stored the value of the Addresser field, etc ...

enregistrement 200 de la Table Soins. Les octets 201 et 202 servent au stockage du champ Soin~ID ; l'octet 203 correspond au champ SoinPersonneSSl ; l'octet 204 correspond au champ Soin~Personne~SS2 ; l'octet 211 correspond au champ FIG. 4 represents a part of the memory block of a

200 registration of the Care Table. Bytes 201 and 202 are used to store the field Care ~ ID; byte 203 corresponds to the field PersonalCareSS1; byte 204 corresponds to the field Care ~ Person ~ SS2; byte 211 corresponds to the field

<Desc / Clms Page number 20>

 Care ~ Person ~ SS3; byte 212 corresponds to the field Care ~ Person ~ SS4; bytes 213 and 214 correspond to the field Care ~ Person ~ SS5; bytes 221 and 222 are used to store the value of the Doctor ~ ID field; the next 4 bytes (partially represented) correspond to the Date field, etc.

 The fields constituting the social security number of each patient, stored on bytes 141 to 144 and 173-174 for the record 100, are chosen to define a primary key of the Patients table. The care table has these same fields, corresponding to bytes 203-204 and 211 to 214 for the record 200, which constitute for this table a foreign key allowing, starting from a record of the table Care, to go back to the corresponding record of the Patients table. A patient can be the subject of different care sheets, such a relationship is called 1-N relationship. The Exam and Patient tables are linked in the same way by means of the fields constituting the patient's social security number. The Prescription and Care tables are linked by means of the Prescription ~ ID field (denoted respectively Ordinance ~ ID and CareOrderID).

 In order to guarantee a perfect identity of the formats of the field Person ~ SSi of the table Patients with the field Care ~ Person ~ SSi of the table Care (i successively taking the values 1 to 5), the device according to the invention is adapted to import the setting the format of the Person ~ SSi fields from the descriptor defining the format of the records in the Patients table to the descriptor defining the format of the records in the Care table. This operation is carried out automatically following the definition of the relational rules linking the two tables (definition of the primary keys, foreign, relations fields ...).

 It should be noted that, in the device according to the invention, it is not necessary for the Person-SSi fields, which compose the primary key of the Patients table and the relation field with the Care table, to be defined afterwards. each other in the descriptor of said table, and thus stored after each other in the user memory when inserting a record. On the other hand, the definition of the PersonalCareSSi fields in the format file of the Care table is automatically performed by a single operation to copy the descriptor data from the Patients table relating to the said fields, they are therefore automatically defined afterwards.

<Desc / Clms Page number 21>

 from each other in the Care table. The key memory space dedicated to the combination of the Care ~ Person ~ SSi fields is therefore a compact memory block.

la table Patients et le champ de résultat est le champ Patients-Nom. The foreign and primary keys of the Care and Patient tables allow, for example, from any information concerning a given care sheet (for example the identification number of the care sheet, denoted SoinID), to access the information the subject of the care sheet. Such a search is said to be single-criteria on several tables. In the example mentioned above, the search table is the Care table, the search field is the Care field ~ ID, the reference value is the identification number of the care sheet concerned, 123456789 for example, the operator comparison is the sign =. If the purpose of the search is to find the name of the patient concerned, the result table is

the Patients table and the result field is the Patients-Name field.

This search results in the following operations: a) - find the record of the Care table for which SoinID = 123456789: reading of the values of the Care ~ ID field of the records of the Care table, comparison of these values with the reference value 123456789, storing the identifier of the record sought, b) -make the join between the Care and Patient tables. This operation is written with a single command line: "Join many to one Care. During the construction of the database and the definition of relational rules between the tables, a Person-ID relation field has been defined as grouping the Person ~ SSi fields (i taking the values 1 to 5) and constituting, moreover, a primary key of the Patients table. Moreover, the fields
Person care was defined as constituting a single field forming a foreign key of the Care table pointing to the primary key Person ~ ID of the Patients table.

The preceding query line causes the processing means to successively select records from the Patients table, for example a record 100, and to directly compare the key memory space (memory space occupied by the fields of the foreign key of the Care table corresponding to the personID relationship field, the care table also having other foreign keys defining relationship fields with other tables) of the registration of the care table,

<Desc / Clms Page number 22>

enregistrement 200 par exemple, vérifiant Soin~ID=123456789 (dont l'identificateur a été mémorisé lors de l'étape précédente), avec les espaces mémoires clefs (occupés par les e champs de la clef primaire Personne~ID de la table Patients) de chacun des enregistrements de la table Patients successivement sélectionnés. La comparaison s'effectue, par exemple, octet par octet : les moyens de traitement comparent successivement les octets 141 et 203, puis 142 et 204, puis 143 et 211, puis 144 et 212, puis 173 et 213, puis 174 et 214. Lesdits octets sont lus et directement comparés, sans aucun formatage ni interprétation préalables des données qu'ils contiennent. De plus, l'opération s'effectue sans création ni mémorisation dans la mémoire de travail d'une table de recherche (produit scalaire) comportant les enregistrements des tables Patients et Soins. Dans l'exemple présent, il n'existe qu'un seul enregistrement de la table Patients vérifiant l'égalité. Le dispositif est adapté pour interrompre les comparaisons dès que l'enregistrement recherché à été identifié (commande"many to one"). c)-mémoriser, à titre de résultat de la recherche, l'identificateur de l'enregistrement de la table Patients retenu par l'étape précédente, et lire et afficher la valeur du champ Patients-Nom de cet enregistrement.

record 200 for example, checking Care ~ ID = 123456789 (whose identifier was memorized in the previous step), with the key memories (occupied by the e fields of the primary key Person ~ ID of the Patients table) of each of the records in the Patients Successively Selected table. The comparison is made, for example, byte by byte: the processing means successively compare the bytes 141 and 203, then 142 and 204, then 143 and 211, then 144 and 212, then 173 and 213, then 174 and 214. The said bytes are read and directly compared, without any prior formatting or interpretation of the data they contain. In addition, the operation is performed without creation or storage in the working memory of a search table (dot product) containing the records of the Patients and Care tables. In this example, there is only one record in the Patient Equality Table. The device is adapted to interrupt the comparisons as soon as the searched record has been identified ("many to one" command). c) -Store, as a result of the search, the identifier of the record from the Patients table retained by the previous step, and read and display the value of the Patients-Name field of this record.

 This last step advantageously uses the fact that the user data are stored in a memory 3a directly and quickly accessible for reading and writing (electronic memory) by the microprocessor. It allows searches based on the results of previous research, extremely fast. Thus, the address of the person who has been the subject of the care sheet n 123456789, and whose name (Martin for example) appears on the display screen in response to the previous search, is directly accessible through the identifier of the record, still in memory as a result of the previous search.

This operation is a simple read operation in the user memory, instantaneous.

 For comparison, in an earlier device with an SQL search engine, find this address is to formulate the new search: search the record of the Care table for which Care ~ ID = 123456789 (a search on the name Martin in the Patients table would be very likely to lead to several results); perform the product of this record with all the

<Desc / Clms Page number 23>

enregistrements de la table Patients ; rechercher, parmi les lignes créées, celles vérifiant Personne~SSi = Soins~Personne~SSi (i prenant successivement les valeurs 1 à 5) ; mémoriser et afficher les valeurs des champs Patients-Adresse-rue, PatientsAdresseCodePostal et Patients-Adresse-ville de la ligne retenue lors de l'étape précédente.

records from the Patients table; search, among the created lines, those verifying Person ~ SSi = Care ~ Person ~ SSi (i successively taking the values 1 to 5); memorize and display the values of the Patient-Street Address, PatientPostalCode and Patient-City Address fields of the line selected in the previous step.

 It should also be noted that for the database manager, the combination of the Care ~ Person ~ SSi fields is defined as a single field (foreign key) to establish a relationship with the Patients table.

At the same time, each of the Care-Person-SSi fields is individually defined as a field in the Care table (whose format has been imported from the descriptor in the Patients table), so that it is also possible to perform on this table. Care of research on only one of these fields.

 The device according to the invention therefore makes it possible to perform joins between linked tables of a database (the relational rules must, in return, be defined during the construction of the database) in a safe and extremely fast way, thus considerably reducing search times in said database. It uses a greatly simplified query language; the corresponding applications occupy a reduced memory space.

 The various functions according to the invention, described above, can easily be performed by simple programming using traditional tools, using in particular the C ++ language.

 It goes without saying that the invention may be subject to numerous variants with respect to the embodiments previously described and shown in the figures.

 In particular, the device according to the invention is adapted to allow the realization of a join to determine the unique record of a first table linked to a record of a second table that may contain several records related to the recording. searched for the first table (relation called 1-N and corresponding to the example described above: the same person, identified by a single social security number, can be the subject of several sheets of care). Such a search is written using a single command line: "Join many to one Care.

<Desc / Clms Page number 24>

 In the case where several records from the second table have been previously selected (example: we are looking for the patients of a given doctor) A first stage of research made it possible to identify all the care sheets delivered by this person. doctor), the device is adapted to perform as many joins of said first and second tables as of selected records The set of joins is programmed by means of a single command line: "Join multiple many to one Care. PersonneID ".

 The device also makes it possible to perform a join to determine the records of a first table related to a record of a second table, it being specified that said second table contains only one record related to the searched records of the first table. (Relationship Nl: it is for example to find the sheets of care which was the object a given person, the first table being the table Care, the second table Patients). Such a search is also written using a single program line, using a "one to many" command: "Join one to many Care. As previously, in the case where several records of the second table have been previously selected (looking care sheets of a series of patients exercising a given profession, a first step to identify said patients), the device performs a plurality of joins under the following single command: "Join multiple one to many Care.

Claims (14)

1 / -Device for digital data processing, comprising processing means (1) comprising at least one microprocessor; means (2), said non-volatile memory read-only memory directly accessible read-only by the processing means, and consist exclusively of one or more non-volatile electronic memory (s); means (3a), said user memory, storage directly accessible read and write by the processing means for storing programs and / or user data, and consisting exclusively of one or more electronic memory (s) ( s); autonomous power supply means (13); means forming a user interface comprising at least means (5,6a, 6b) for acquiring user data and means (4) for restoring said data; software means stored in the read-only memory and / or in the user memory adapted to manage at least one relational database constructed according to specificities defined by a person, called an operator, and making it possible to organize the storage of the user data in the form of table records each comprising a plurality of fields, said database comprising at least two tables linked together by one or more common fields, called key fields, which occupy, for each record of said tables, a memory space said key memory space of the recording, the device being characterized in that it is adapted for: - during the construction of the database, recording in the user memory (3a) a format of the records of the one-first-of the two tables, as defined and entered by the operator, in a descriptor of the first table, and copy in the user memory, except ir of the descriptor of the first table, the information defining the format of each of the key fields of said table to a descriptor of the second table, defining a format of the records of the other-said tables, so that each key field is defined with the same format in both tables, allow for at least one join between said tables to determine the record (s) of the first one of said tables <Desc / Clms Page number 26>  linked to a record (200) of the other-second-of said tables, by directly comparing the electronic state of the key memory spaces (203,204, 211-214,141-144, 173,174) of the recording of the second table and at least one record (100) of the first table, and storing in the user memory at least one piece of information, called the result of the join, representative of the fact that one or more record (s) identified (s) of the first table has (s) a key memory space electronically identical to that of the registration of the second table.  2 / -Dispositif according to claim 1, characterized in that the device is adapted for, during the construction of the database, store the definition, as entered by the operator, a field, said field relationship of first and second table, formed by the combination of the key fields as defined in the descriptor of the first table, and automatically copy the information defining the format of the fields composing the relationship field from the descriptor of the first table to the descriptor of the second table table in response to storing the definition of said relationship field.  3 / -Dispositif according to one of claims 1 or 2, characterized in that it is adapted for, when performing the join, compare byte (141,203) by byte the key memory spaces of the recording (200) of the second table and a record (100) of the first table, and interrupt the comparison as soon as two corresponding bytes compare have a different electronic state.  4 / -Dispositif according to one of claims 1 or 2, characterized in that it is adapted for, when performing the join, compare word by word, respectively long word by long word, respectively double by double, the key memory spaces of the recording of the second table and a record of the first table, and interrupting the comparison as soon as two words, respectively two long words, respectively two double, corresponding compared show a different electronic state.  5 / -Dispositif according to one of claims 1 to 4, characterized in that it is adapted for, when performing a join, select a comparison mode for each key field depending on the format of said key field. <Desc / Clms Page number 27>  6 / -Dispositif according to one of claims 1 to 5, characterized in that it is adapted for, when performing the join, compare key field key field key memory spaces of the registration of the second table and a record of the first table, and interrupt the comparison as soon as the memory spaces of two corresponding key fields compared present a different electronic state.  7 / -Dispositif according to one of claims 1 to 6, characterized in that it is adapted to calculate and store, for each record of the database, an identifier for accessing said record directly in the user memory.  8 / -Dispositif according to claim 7 characterized in that it is adapted to develop, update for each new record entered a predefined table by the operator, and store in the user memory, an index of a field of said predefined table by the operator, said indexed field, said index comprising the identifiers of the records of said table stored one after the other so that the values of the indexed field of said records are ordered.  9 / -Dispositif according to claims 6 and 8, characterized in that it is adapted to allow the realization of at least one join between a first and a second tables linked by one or more common key fields of which at least one is indexed, and to successively select the records of the first table whose key memory spaces are to be compared to that of the record of the second table, by performing a dichotomic search on the index of the indexed key field.  10 / -Dispositif according to one of claims 1 to 8, characterized in that it is adapted to successively select the records of the first table whose key memories are compared to that of the record of the second table, in a storage order of said records in the user memory.  11 / -Dispositif according to one of claims 1 to 10, wherein the combination of key fields is a primary key of the first table and a foreign key of the second table, characterized in that it is adapted to interrupt comparisons between the key memory space of the registration of the second table and the <Desc / Clms Page number 28>  key memory storage spaces of the first table, as soon as a record of the first table having a key memory space electronically identical to that of the recording of the second table has been identified.  12 / -Dispositif according to one of claims 1 to 11 and according to claim 7, characterized in that it is adapted to store, as a result of the join, the identifier of the record (s) of the first table having a key memory space electronically identical to that of the registration of the second table.  13 / -Dispositif according to one of claims 1 to 12 and according to claim 7, characterized in that it is suitable for: - in response to a search made in the database from search criteria, enter by a user or an operator and comprising at least designation information of a table, called search table, of a field of said table, called search field, of a comparison operator, of a reference value, and from another table, called the result table, select the record (s) of said search table satisfying said search criteria, - make at least one join between the search table and the result table in order to to determine the record (s) of the result table (s) associated with each record, previously selected, of the search table, - store in the user memory, the identifier of the record (s) ), previously d finished (s), of the result table.  14 / -Dispositif according to one of claims 1 to 13, characterized in that the user memory (3a) is chosen from a dynamic volatile electronic memory of the RAM type, a static volatile electronic memory of the SRAM type, a non-volatile electronic memory flash memory type, a memory space of one of these memories dedicated to the storage of data and user programs.