[PIC Micro Pascal] Nouvelle version du compilateur PMP [News]

[Alcatîz]

Bonjour,

PMP est un cross-compilateur gratuit pour les processeurs PIC. Il fonctionne avec la suite MPASM/MPLINK de Microchip; il comprend son propre EDI.
Il supporte pratiquement tous les composants PIC18, PIC16, PIC10 et PIC12.

Téléchargez la dernière version en date :
http://pascal.developpez.com/telecha...C-Micro-Pascal

08/11/2010 : Sortie de la nouvelle version 1.5.0.
10/11/2010 : Suite à un package buggué, sortie de la version 1.5.1.
15/11/2010 : Sortie de la version 1.5.2.
16/11/2010 : Sortie en urgence du release 1.5.3.
20/12/2010 : Nouvelle version 1.5.4, plus stable avec quelques nouvelles fonctionnalités.
02/06/2011 : Sortie de la version 1.6.0.
02/06/2011 : Version de maintenance 1.6.1.
27/01/2012 : Sortie de la version 1.6.2.

Si vous avez déjà testé ce compilateur, n'hésitez pas à nous faire part de vos impressions, trucs et astuces...

[Alcatîz]

Nouvelle version 1.3.20 :
http://pascal.developpez.com/compilateurs/pmp/

Il s'agit d'une révision mineure sans ajout de nouvelles fonctionnalités.
De légère modifications de syntaxe ont entre-temps été implémentées dans la version 1.3.17.

[Alcatîz]

Sortie aujourd'hui de la nouvelle version 1.3.25, avec une grande nouveauté : un explorateur de code.
http://pascal.developpez.com/compilateurs/pmp/

Informations de version : http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

La version 1.3.27 est à présent disponible, toujours en version beta.
Un effort tout particulier a été mis sur la résolution de bugs et un meilleur support des PIC18.

http://pascal.developpez.com/compilateurs/pmp/

Informations de version : http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[PPA94]

Bonjour,

La nouvelle version 1.4.0 de PIC Micro Pascal (PMP) vient de sortir.

Principales nouveautés :

• Le support des PIC16 enhanced mid-range (alpha).
• Le support de la virgule flottante pour les PIC16 (en plus des PIC18).

Alcatîz, merci de mettre à jour la page http://pascal.developpez.com/compilateurs/pmp/

[Alcatîz]

Attention, si vous avez téléchargé PMP 1.4.0 : cette version est entachée de bugs. Les développeurs mettent à votre disposition la version 1.4.1 (en version beta) :

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

Nouvelle version 1.4.2 : correction de bugs et ajout de fonctionnalités pour les PIC18, notamment un MUL(byte,byte) dans un word et des divisions 8 bits DIV 10, DIV 100, MOD 10 and MOD 100 super rapides.

L'équipe de développement est toujours à la recherche de beta-testeurs :
http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

Nouvelle version 1.4.3.

Il s'agit d'une version de maintenance, corrigeant quelques bugs et ajoutant quelques fonctionnalités :

• Added : Syntax for pointer assignment with a constant, variable or sfr address.
• Added : Syntax for CONST array of pointers.
• Added : ShortInt type support (-128..127).
• Added : @ operator on EEPROM variables returns EEPROM index 0..n (EEREAD address).
• Added : New syntax EEREAD(<ADDR>, <SIZE>) and EEWRITE(<ADDR>, <VALUE>, <SIZE>) where <SIZE> may be any simple type or constant equal to 1, 2 or 4.
• Added : New warning for RMW problem if using writes to PORTx bits rather than writes to LATx bits; processors with LATx registers.
• Fixed : Editor's caret and position in window was not always saved correctly.
• Added : Error if FOR loop variable assignment (check in local proc/func/main block only: loop variables may be global in PMP, so no check is made if the loop variable is assigned outside of the proc/func/main block.
• Fixed : Bad RAM data reservation code generation; all processors.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

Une semaine après la 1.4.3, l'équipe de développement sort en catastrophe la version 1.4.4, à cause de bugs très gênants :

• Fixed : ABS() accepted on ShortInt.
• Fixed : Unexpected error for loop variable assignment; all processors.
• Fixed : Infinite loop on some syntax errors; all processors.
• Fixed : Unexpected syntax error on empty CASE block.
• Fixed : IDE behavior - right click on a multi-line block may loose the block selection.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

Sortie de la version 1.4.6

Correction de nombreux bugs :

Fixed : Bad generated code for ROL() and ROR() with more than 1 position; all processors.
Fixed : Internal error for ROL() and ROR(); PIC16 enhanced processors.
Fixed : Unexpected syntax error on label definition; side effect on better syntax checking for semicolons.
Fixed : Unexpected syntax error on forward label definition.
Fixed : Unexpected syntax error in asm block when using assembler instructions starting with 'end'.
Fixed : Internal error on array indexing; PIC16 enhanced processors.
Added : Compiler errors and warnings file names are now displayed with a relative path to save space.
Fixed : Internal error on variable clear; PIC16 enhanced processors.
Fixed : Bad generated code for a local variable declared absolute to a VAR parameter; all processors.
Fixed : Bad generated EEPROM initialization data; PIC18 processors (use of CODE_PACK).
Fixed : Bad configuration bits decoding for some old parts as 16F88.
Fixed : Another issue with goto / labels; also changed labels naming convention; all processors.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[Alcatîz]

Nouvelle version 1.4.7 :

Ajout du support de GPUTILS :

Fixed : Unexpected type mismatch or internal error when defining a variable absolute to a const variable; all processors.
Fixed : Bad IDE behavior with Highlighter options when selecting a scheme that has more attributes than the current one.
Fixed : A bad side effect of a previous code update impacts on booleans based on another variable; bad generated code for booleans; all processors.
Added : A confirmation dialog box in PMP uninstall to confirm the PMP folder full erase.
Added : Devices files path is now accessible in Project's options.
Added : GPUTILS support (alpha).
Added : Expression evaluation for absolute variable declaration.
Fixed : Bad generated code for ADCON1 initialization when $INIT ANALOGS is used; some PIC12 and PIC16 processors.
Fixed : No generated code for ANCON0/ANCON1 initialization when $INIT ANALOGS is used; some PIC18 processors.
Fixed : Internal error for jump tables trying to use PCLATH; PIC10 & PIC12 processors without PCLATH.
Fixed : Related to above, same error for CASE jump tables; PIC10 & PIC12 processors without PCLATH.
Added : Copy errors to clipboard command (right click menu).

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

[krachik]

Nouvelle version 1.4.8 :

Fixed: Bad banking for small processors without shared memory (several banks but not all mapped to bank 0).
Changed: Removed the last "Hidden file functions": now "Export" to Any/rtf/html is in the main and context menus.
Fixed: Bad generated code (broken code due to new optimization side effect) for longint/longword SHR/SHL <CONST> in some cases, with SPEED optimization; all processors (behave differently).
Fixed: Related to above, internal error is some longint/longword SHR/SHL cases when SPEED optimization and <CONST> > 2; all processors (behave differently).
Fixed: Related to above, optimization for <VAR> := <VAR> SHR/SHL <CONST> was not effective (should shift "in place" for small <CONST>); all processors.
Fixed: Boolean typed CONST was not accepted.
Added: Delay_ms now accepts a word sized non constant expression.
Added: Delay_cy new built-in procedure (delay a nb of cycles, may be constant or word expression).
Changed: Compiler search paths in project's options are now stored as relative paths.
Fixed: Bad parsing of some string concatenations.
Fixed: Bad generated code for function call returning integer and followed by an AND/OR operator.
Fixed: Syntax error using a boolean as FOR loop variable.
Modified documentation: DECLARED() and DEFINED() pseudo functions was useable in normal code expression despite of what the manual said.
Changed: Boolean variables memory allocation is now global so that host bytes are shared between units/main program, using less memory.
Changed: Internal asm representation now uses a dot as separator instead of @ character. Warning: Compatibility issues may occur within asm blocks, libraries have been adapted.
Optimized: Pagesel was still generated for some 10F and 12F devices with only one bank.
Fixed: Another syntax error that hangs PMP (semicolon before ELSE in some cases).

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=320

Concernant la PMP V1.4.8.64 release (beta version), l'équipe de développement est à la recherche de beta-testeurs

[Alcatîz]

Déjà une nouvelle version 1.4.9

A titre expérimental pour l'instant, apparaissent les enregistrements dynamiques (c'est-à-dire des enregistrements avec des méthodes statiques). PMP semble donc évoluer vers la programmation orientée objet.

Egalement quelques corrections sur cette version :

• Fixed : Side effect on the new assembler naming convention: STR32/STR32S generates assembler errors.
• Fixed : /q assembler option had vanished.
• Fixed : STRCAT routine parameter passing was broken, generating string concatenation / assignments errors. PIC10 to PIC16 Processors (except enhanced).
• Fixed : Bad generated code for bit test within a field within a record. All processors.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=756

[Alcatîz]

Version 1.5.0

Les auteurs de PMP avertissent leurs utilisateurs que, suite à une nouvelle convention de nommage, il se peut qu'apparaissent des problèmes de compatibilité au niveau de l'Assembleur inline.

• Modified : STATUS is preserved at startup so special bits may be tested by program.
• Added : $PUSH/$POP directives for enhanced management of local compiler options.
• Added : $SCRIPT directive to override normal linker script name.
• Added : $OPTIMIZE_PARAMS directive to switch ON/OFF the procedure/functions parameters.
  WARNING : Possible compatibility issue with pure EXTERNAL procedures and functions.
• Fixed : Bad generated code (old bug) in IN internal subroutines; all processors.
• Fixed : Bad file loaded if the required source also exists in the "current" windows directory (std FileSearch behavior...).
• Added : Better assembler block : now identifiers are automatically translated to internal format.
• Added : Now VAR arrays may be in EEPROM, with optional initial values. EEPROM VAR arrays cannot be passed as CONST or VAR parameters.
• Fixed : Syntax error after a label inside a loop in another loop.
• Modified : Better analysis of interrupt context save : called procedures/functions are now analyzed at full nesting deep.
• Added : Open array parameters.

[Alcatîz]

Version 1.5.1

Le package FP pour PIC16 étant buggué, une mise à jour s'impose.
http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=813

[Alcatîz]

Version 1.5.2

Une autre release qui corrige quelques effets de bord introduits par une nouvelle methode d'optimisation :

• Fixed : If $INTERRUPT UNIQUE, context saving was not always complete; all processors.
• Fixed : New ASM enhanced features produces syntax errors with decimal notation .nnnn and d'nnnn'.
• Fixed : Bad generated code due to a wrong pseudo stack usage / optimization when using slices. Visible impact on A2D unit; all processors.
• Fixed : Bad generated code with CASE jump table in some circumstances. PIC18 processors.
• Fixed : Bad generated code for array read with a byte index in some circumstances; PIC18 processors.
• Fixed : Side effect of new optimization: Warning nnn directive was not applied locally.
• Fixed : Side effect of new optimization: Defining a String constant by concatenation was rejected.
• Fixed : IDE file was not always focused after loading.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=819

[krachik]

Version 1.5.3
Encore une autre release pour corriger quelques effets de bord introduits par une nouvelle méthode d'optimisation :

• Fixed: A ShortInt was not accepted as a FOR loop variable!!!
• Fixed: Bad generated code for mixed signed/unsigned DIV in some circumstances; all processors.
• Fixed: Bad generated code for mixed 8/16/32 bits operators - operand promotion in some circumstances; old bug, all processors.
• Fixed: Big regression bug; if there's an IF statement that evaluates to FALSE at compile time, the whole current procedure code is not generated after this point.
• Added: After too much side effects / regression errors, the enhancement of the validation system is in progress, with more new automatic checks based on Delphi code comparisons.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=823

[krachik]

Version 1.5.4
Une nouvelle version plus stable avec quelques nouvelles fonctionnalités :

• Fixed: Side effect of new optimization: Bad generated code for $B- AND / OR expressions; all processors.
• Fixed: Side effect of new optimization: Possible banking problem in MUL16/DIV16 MUL32/DIV32; all processors.
• Fixed: Bad generated code if a function call is not the leftmost operand in an expression in some circumstances; all processors.
• Fixed: Bad generated code if a function call is within a FOR loop in some circumstances; all processors.
• Fixed: processors.cfg was not complete for PIC16F5? processors, generating bad banking instructions.
• Fixed: Side effect of new optimization: bad parameter passing if only one parameter and if it may be evaluated as boolean (0/1); all processors.
• Added: A "save as defaults" functionality in the project's options dialog to define options to be used for a new project.
• Changed: IDE search/replace function preselects "selected text only" if something is selected in the current window.
• Fixed: PIC12F629/675 false stack overflow warnings.
• Added: $INIT COMPARATORS to initialize comparator-aware pins to all digital (comparators off) - CMCON in small devices.
• Fixed: Removed unnecessary banksel instructions; some small devices (PIC10, PIC12).
• Fixed: Compilation error in disabled conditional compilation blocks with a $VECTORS directive for another processor range.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=835

[Alcatîz]

Nouvelle version 1.6.0
Plus de stabilité et plusieurs nouveautés. Jugez plutôt :

Fixed : Always read volatile 16 variables LSB first and write them MSB first to manage PIC18 TMRx 16 bits mode.
Fixed : Unexpected syntax error when an argument of WRITE std procedure is a function call with parameters and with a char or string result type.
Added : In Extended Syntax mode, an optional BY clause to the FOR statement to specify the loop variable increment (MODULA/OBERON-like syntax).
Added : In Extended Syntax mode, an optional ELSIF clause to the IF statement (MODULA/OBERON-like syntax). Also added $ELSIF as an alias of $ELSEIF that stays for compatibility.
Added : In Extended Syntax mode, a "LOOP ... END" block equivalent to a "REPEAT ... UNTIL FALSE;" block (MODULA/OBERON-like syntax).
Added : In Extended Syntax mode, a "RETURN <Expression>" statement for functions that generates an optimized "RESULT := <Expression>; EXIT;" block.
Added : A "FOR <Loop Variable> IN <Enumerable type or variable> DO <Block>" iterator construction (now standard in DELPHI/FPC).
Fixed : Unexpected "undefined forward reference" error on a non-forward LABEL, due to an early modification side effect.
Changed : For PIC18, the default procedure interrupt level is now HIGH.
Added : For PIC18, a new attribute for interrupt procedures: FAST may be used to optimize interrupts when there's only one interrupt level. FAST mode is the default when there's only one interrupt procedure.
Fixed : With $INTERRUPTS UNIQUE the interrupt vector did not be generated if the interrupt procedure was in a unit.
Fixed : Bad generated code for STR() routine for bytes; bad conversion of some numbers; PIC18 processors.
Added : UPCASE built-in function.
Fixed : Bad generated code for MOVE procedure if the source parameter is a ROM identifier; all processors.
Fixed : Bad generated code for a by-value parameter if the passed value is in ROM (initialized ROM arrays); all processors.
Fixed : Bad dependency check between units when a unit is added to a uses list, avoiding rebuild in some circumstances and generating linker errors.
Fixed : Unexpected undefined symbol with local variables with a procedure/composite composite name larger than 32 characters (failure of the MPASM limit workaround).
Added : Support of ROM array parameter passing (CONST ROMABLE, see the new manual).
Fixed : Various formatting problems in ASM blocks.
Fixed : Bad side effect on STR() built-in procedure that generated unexpected errors.
Fixed : Bad dependency check that produced unnecessary unit rebuilds.
Changed : Symbols table management / rebuild has been optimized so that the compiler is a bit faster.
Fixed : In an ASM block double quoted strings was altered due to the Pascal parser that does not recognize them as strings.
Fixed : In an ASM block assembler hex format nnH was altered due to the Pascal parser that does not recognize them as hex numbers.
Updated : processors.cfg to include 16F630 & 16F676 specifics.
Fixed : Bad error message position in some circumstances; PIC16ENH & PIC18 processors.
Fixed : FP divide code generation was broken due to a bad optimization; all processors.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=924

[Alcatîz]

Nouvelle version 1.6.1

Il s'agit d'une version de maintenance :

• Fixed : The optimization of a simple EE variable read to a RAM variable of the same size has been altered and generated bad code; all processors.
• Fixed : Some FP routines was still generating internal errors / bad code in some circumstances; all processors.
• Fixed : Standard constants and SFRs was no longer displayed in the IDE's code tree.
• Fixed : Write/Writeln bad behavior if no width specified; PIC18 and enhanced PIC16 processors.
• Added : IDE's editor tabs can now be re-arranged through mouse drag'n drop.
• Fixed : Some IDE's behaviors, such as editor page focus after loading.
• Fixed : Bad generated code for FP compare if both terms are negative: indeed a bug in the original Mike Gore's FP_CMP routine; all processors.
• Modified : STR() behavior on small signed values (shortint, integer): minus sign is now inserted at first position instead of overwriting first character.
• Fixed : Link error due to a bad RAM memory allocation when performing a partial compilation of a project with nested USES.

http://www.pmpcomp.fr/plugins/opendo...ode=one&pg=933