Gpc February 2004

gpc@gnu.de

28 participants
40 discussions

Re: Help with blockread/blockwrite
by Peter Norton 08 Feb '04

08 Feb '04

Chuck F wrote: > An integer requires 4 bytes, and must be aligned at a multiple of 4. Fairly say I never heard about it must be aligned in record. I use Pascal some years now (not GNU) and always mixed different types in records without problems. Thanks for explanation. I take my words (buggy GPC) back ;-) Peter. __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

2 1

Re: Help with blockread/blockwrite
by Peter Norton 07 Feb '04

07 Feb '04

Frank wrote: > That's the GCC version. Run `gpc -v'. Oh yes. GPC is version 20030830. BTW it seems realy buggy on windows ( or windows is buggy itself ;-) Today I found that my program can't read BMP (image) header with GPC. The record I use for this is started from two bytes ( I tried also two char, shortword, shortint ) followed by integer, integer ... It reads first two bytes OK, then skips two bytes more and starts to read first integer from offset 4 ??? Magic. Peter. __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

2 1

Re: Help with blockread/blockwrite
by Peter Norton 07 Feb '04

07 Feb '04

Frank wrote: >BTW, which OS and which GPC version? Windows 98SE, MinGW/GPC 3.2.3 Peter __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

2 1

Re: Help with blockread/blockwrite
by Peter Norton 06 Feb '04

06 Feb '04

Frank, >I cannot reproduce it, but I suppose it's just a misunderstanding of >the `BlockRead' semantics. If the 4th parameter is given, it doesn't >have to read the whole buffer because the 4th parameter tells how >many were read. I always thought that forth parameter in 'BlockRead' contains information how much units was read. How much to read is controlled by the third parameter. Newertheless your code works fine - it just make one more loop on the eof. I'm still interest what prevent to fill the full buffer (512 bytes) on this magic poiner 16384. I've debugged this sample and found that with big files the second magic spot appears somewhere about 100773 filepos. But it's may vary. Thanks, Peter __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

2 1

GDB Debugging Nightmare!
by Bryan Meredith 06 Feb '04

06 Feb '04

Hello chaps, I have looked through the list but cant find anything definitive on this one... I am using -gstabs to compile as -g on its own gives "internal error - unimplemented function unk_lang_create_fundamental_type called." (as per an earlier post). I can get things to display by casting them to their type Type(Var) with initial letters capitalised. Where I get really stuck is trying to use this technique on a thingy (its a union in C - I dont know much pascal so I dont know what its called - sorry). This isnt real code - its so I can try to explain a little better... TYPE header = RECORD item1 : WORD; item2 : WORD; item3 : LONGWORD; END; type1 = PACKED RECORD hdr ; header; custom1 : BYTE; END; { you get the idea...} thingy = RECORD CASE INTEGER OF 1: (member1 : type1); 2: (member2 : type2); 3: (member3 : type3); END; VAR dummy1 : header; dummy2 : type1; dummy3 : thingy; >From within GDB when the program is running, I can inspect the contents of dummy1 by doing p Header(Dummy1) ( I would expect a simple p dummy1 to work but any debug is better than none). How do I get it to display member2 (assuming that member2 is the relevent member). I can tell which structure to use as there is a variable that has been set. Out of interest, is there a way of telling which member is the relevent one just by looking at dummy3? Sorry if this is really amateurish - I am a C programmer that has suddenly inherited a whole load of Pascal and I must admit that I am struggling with the debugger. It works fine for me in C - I am hoping that it is my ignorance that is the cause of the problem. If there are any publications, online or in print (I have the GPC documentation but it didnt help on this one) that you recommend, please post away. Thanks for any insight, Bryan "Brain Murders" Meredith

3 8

Help with blockread/blockwrite
by Peter Norton 05 Feb '04

05 Feb '04

I have simple file copy example: var infile,outfile: file; buff: array[1..512] of byte; copied: integer; begin assign(infile, 'input'); reset(infile, SizeOf(byte)); assign(outfile, 'output'); rewrite(outfile, SizeOf(byte)); repeat blockread(infile, buff, 512, copied); blockwrite(outfile, buff, copied); until copied < 512; close(infile); close(outfile); end. It works fine. But if before repeat loop I placed: blockread(infile, buff, 1); blockwrite(outfile, buff, 1); it copy only 16384 (2^14) bytes of input. Always. If input file is smaller than 2^14 bytes then it copied all bytes. If larger then it cutted down to 2^14 bytes. I have this behaviour with GPC only. No problems with FPC, VP... What is wrong? Thanks in advance, Peter __________________________________ Do you Yahoo!? Yahoo! Finance: Get your refund fast by filing online. http://taxes.yahoo.com/filing.html

2 1

packed records' address again
by Francesco Bonomi 05 Feb '04

05 Feb '04

Dear friends, It turns out I have been too optimistic. I had two thoughts: 1) I can unpack/repack the structures and work on them. FALSE! Here I have records, not arrays! 2) I cannot assign the array of chars in a single instruction, but I can assign a single char: all I need is a loop. FALSE! My situation is in effects more problematic, probably because also some of the sub-structures are packed... See below These are the declarations I have (stripped down) TYPE ITEN = packed RECORD KEY :PACKED ARRAY [1..MAX] OF CHAR; PUNT:integer; END; ND = packed RECORD POS : integer; IDX : packed ARRAY [1..ONDM2] OF ITEN; END; VAR DATA : ND; This assignment fails: DATA.IDX[H1].KEY:= <array of char>; I thought I could loop through this assignment: DATA.IDX[H1].KEY[x] := <char>; But it still fails. The compiler says: Error: cannot take address of packed record field `IDX' BUT if the problem is with IDX, why this assignment is ok??? DATA.IDX[H1].PUNT := <integer>; At the moment, the workaround I have found is: VAR DATA : ND; TMP : ITEN; TMP:=DATA.IDX[H1]; TMP.KEY:=<array of char>; DATA.IDX[H1]:=TMP; I cannot really understand why (and if) this should be safer (from the compiler's poin of view) than doing DATA.IDX[H1].KEY:= <array of char>; Any thought? Ciao Francesco Bonomi Francesco Bonomi Si.lab s.r.l. Via Citille 13 50022 Greve in Chianti FI- Italy Tel +39 - 055 8544805 Fax +39 - 055 8547418 http//:www.silab.it

2 1

TLIST, STRINGLIST, etc
by Paul Davidson 03 Feb '04

03 Feb '04

Not sure if this is proper forum for this. Flame if appropriate :) Are there ports of TSTRINGLIST object and its friends, TLIST, TOBJECTLIST, et al? Thank you p davidson

2 1

Re: label used before containing beinding
by Tom Verhoeff 03 Feb '04

03 Feb '04

>Nick Ioffe wrote: > >> I've encountered the message in the subject trying to compile >> the following: >> >> program x(input, output); >> label >> x_label; >> var >> z:integer; >> w:string; >> function f:string >> begin >> f:='f'; >> end; >> >> begin >> if z = 1 >> then >> begin >> writeln('to x_label'); >> goto x_label; >> end; >> >> w:= 'z:'+f; >> >> x_label:; >> >> end. >> >> Compilation fails and I get: >> x.pas:3: label 'x_label' used before containing binding contour. Have you tried it with numeric labels (which at least the original Pascal definition required)? program ...; label 99; begin ... ... goto 99 ... 99: end. An alternative is to put the code in a separate procedure, and use Exit to leave it early. Exit is basically a jump to the end of the procedure. You can also use Exit in the main program to terminate execution (jump to the end). See documentation for details on Exit. Tom

1 0

label used before containing beinding contour problem
by "Nick Ioffe 02 Feb '04

02 Feb '04

Hi! I've encountered the message in the subject trying to compile the following: program x(input, output); label x_label; var z:integer; w:string; function f:string begin f:='f'; end; begin if z = 1 then begin writeln('to x_label'); goto x_label; end; w:= 'z:'+f; x_label:; end. Compilation fails and I get: x.pas:3: label 'x_label' used before containing binding contour. If I surround w:= 'z:'+f; with 'begin/end' then everything compiles. also if summation is not on string everything compiles. I use "gpc version 20030507, based on gcc-3.2.1" I've found in the archive message #2535 descring the same problem. However the message seems to be unanswered, at least the answer was not on the archive.... Thank you in advance, Nick Ioffe ===================================================== message #2535 From: Berend de Boer Subject: label used before containingn binding contour Date: 29 Jan 1999, 21:01:21 Since about 5 years I try to compile lex for extended pascal with gpc. However, I still don't succeed. This program is a real test case I assume :-) Here the latest error, boiled down to the most minimal code: ------------------------ program Test; type string255=string(255); function Test(filename, ext : String): string255; label 1,2; begin Test := ''; goto 1; goto 2; 2: Test := filename+'.'; 1: ; end; begin end. ------------------------ Groetjes, Berend. =======================================

2 1

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

Gpc February 2004