Gpc October 1997

gpc@gnu.de

31 participants
158 discussions

Problems with strings in GPC beta 971001
by Jesper Lund 22 Oct '97

22 Oct '97

I have encountered a couple problems with strings in the most recent GPC beta version. Most of the problems are due to missing range checks, e.g. when assigning a string to a shorter string, or passing a string to a procedure with a shorter string as formal parameter. 1) The program below illustrates some of these problems: program Test5; var s1 : string [500]; s2 : string [10]; type str5 = string [5]; procedure p1 (s : str5); begin Writeln (s.capacity, ' ', Length (s)); end; begin s1 := 'This is a relatively long string'; s2 := s1; { Copies all of s1 to s2; not just the first 10 characters } p1 (s1); { Passes all of s1 to procedure p1 } p1 (s2); end. Before exiting with a SIGSEGV exception (under DJGPP), the program prints 5 32 5 32 2) Another facet of the problem occurs when reading strings from a text file with Read[ln] (F, stringVar). If you call Readln, the program will read from the file until the next CR/LF, no matter the size (length) of stringVar, and try to copy the entire line to the variables stringVar (which will almost surely crash the program somehow; but not necessarily in the call to Readln. The SIGSEGV exception could come later, and with a stack trace which makes no sense). On the other hand, Read (F, stringVar) only reads a maximum of Length(stringVar) characters [as it should]. Jesper Lund

1 0

Re: Uses in implementation part of an unit
by Peter Gerwinski 22 Oct '97

22 Oct '97

According to Jesper Lund: > GPC does NOT allow "uses" in the implementation section of a unit. [...] Fixed: 8< ---- diff for parse.y ---------------------------------------------------- @@ -6123,6 +6136,7 @@ ; unit_implementation: + import_part /* This may be empty */ any_declaration_part optional_init_and_final_part ; 8< -------------------------------------------------------------------------- ... but the Automake mechanism does not handle this correctly. To work around, compile one of the units manually before automaking the program. I will try to fix this for the next beta. > In most (Borland-style) programs, moving "uses b" to the interface part (an > obvious workaround; or *solution* if the current GPC behavior is a feature > and not a bug) should not make a difference, but I imagine that there could > be a few exceptions. Yes - not only a few ones. You cite it: > The Turbo Pascal 7.0 manual says about this: > "[...] It also enables you to construct > mutually dependent units." Greetings, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [971005] maintainer GNU Pascal [971001] - http://home.pages.de/~gnu-pascal/ [971005]

1 0

Uses in implementation part of an unit
by Jesper Lund 22 Oct '97

22 Oct '97

GPC does NOT allow "uses" in the implementation section of a unit. That is, the following piece of code cannot compile. unit a; interface <something> implementation uses b; <something> begin end. In most (Borland-style) programs, moving "uses b" to the interface part (an obvious workaround; or *solution* if the current GPC behavior is a feature and not a bug) should not make a difference, but I imagine that there could be a few exceptions. The Turbo Pascal 7.0 manual says about this: "A uses clause in the implementation section allows you to further hide the inner details of a unit, because units used in the implementation section aren't visible to users of the unit. It also enables you to construct mutually dependent units." Jesper Lund

1 0

Rewrite/Reset with filename as second parameter
by Jesper Lund 22 Oct '97

22 Oct '97

Should GPC allow passing the filename to Rewrite/Reset as the second parameter, rather than calling Assign prior to the call to Rewrite/Reset? (several Pascal compilers outside the Borland family have this feature). Currrently, the the program below compiles correctly, but at run-time it stops with the error message ?GPC runtime error: (gpc-rts) Internal file namei wrong (#700) program Test; var F : Text; begin Rewrite (F, 'test.out'); Writeln (F, 'This is a test ....'); Close (F); end. Jesper Lund

1 0

floating point constants
by The African Chief 22 Oct '97

22 Oct '97

Hi all, Why do these lines not compile? Const MAXDOUBLE = 1.79769313486231570e + 308; Const MINDOUBLE = 2.22507385850720140e - 308; Const MAXFLOAT = 3.40282346638528860e + 38; Const MINFLOAT = 1.17549435082228750e - 38; Best regards, The Chief Dr Abimbola A. Olowofoyeku (The African Chief, and the Great Elephant) Author of: Chief's Installer Pro v4.01 for Win16 and Win32. Homepage: http://ourworld.compuserve.com/homepages/African_Chief/ E-mail: laa12(a)keele.ac.uk

3 2

Re: Gpc not accepting non-standard (?) systax.
by Peter Gerwinski 21 Oct '97

21 Oct '97

According to richard.kerry(a)quantel.com: > > type > ArgDefs = ArgDefTable value ( > ((' ', 1, 0), 0, OptionalArg, NullArg), > (('Output_File ', 1, 11), 2, OptionalArg, FileArg), > ... ); According to gpc-971001: > test.pas:28: parse error before `Value' > test.pas:29: invalid schema discriminants Uh - sorry, gpc-971001 does not recognize `value' in this context any more. This is connected with the requirement that "value" must be a redefinable identifier according to ISO-7185 Standard Pascal - which was not the case in earlier versions of GPC. A fix follows below. (Due to other changes, the new line numbers are in fact not correct, but `patch' won't care.;-) Hope this helps, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [971005] maintainer GNU Pascal [971001] - http://home.pages.de/~gnu-pascal/ [971005] 8< --------------------------------------------------------------------------- --- gpc-971001/p/parse.y Wed Oct 1 23:32:34 1997 +++ gpc/p/parse.y Tue Oct 21 23:57:02 1997 @@ -1473,12 +1475,15 @@ { if (! flag_what_pascal || (flag_what_pascal & B_D_PASCAL)) lex_const_equal = 1; + enable_keyword ("Value"); } type_denoter absolute_or_value_specification { tree d, init = $8; tree type = TREE_VALUE ($7); + disable_keyword ("Value"); + immediate_size_expand = $<itype>4; lex_caret = $3;

1 0

Re: also Pre compiled version
by Peter Gerwinski 21 Oct '97

21 Oct '97

According to Omar Abu Al-Haija: > I get precompiled version , Djgpp from gnu pascal , > I try to compile a file , it give me an error message : - > > Gpc.exe : installation problem , cannot exec 'as': No such file or directory > ( Enoent ) . > > [...] > > Ps I had no dpmi loader , and no gnu assembler , or debugger . You need the DJGPP development tools. The GNU Pascal FAQ list says: 8< ---- GPC FAQ ------------------------------------------------------- [22]What do I download? As discussed in section 2.2, other than GPC itself, you need an assembler, linker and friends, a C library and possibly a debugger. From your local djgpp mirror, you can get these as: v2/djdev201.zip (C library) v2gnu/bnu270b.zip (assembler, ....) v2gnu/gdb416b.zip (debugger) The rest is up to you; `make' (v2gnu/mak375b.zip) can be useful, The latest RHIDE test release (an IDE with Borland-look) has support for GNU Pascal, and can be downloaded from: http://www.tu-chemnitz.de/~rho/rhidetest.html 8< -------------------------------------------------------------------- Remarks: * `rhidetest.html' is now `rhide.html'. * You can download this from any SimTel mirror. One is ftp://ftp.ruhr-uni-bochum.de/pub/simtelnet/gnu/djgpp/, but you will probably find one closer to your place. Hope this helps, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [971005] maintainer GNU Pascal [971001] - http://home.pages.de/~gnu-pascal/ [971005]

1 0

Re: complex numbers
by Peter Gerwinski 21 Oct '97

21 Oct '97

According to fulvio ciriaco: > I notice that complex and real numbers cannot be mixed > in mathematical expressions. > The standard tells nothing, to my knowledge, about real > being converted automatically to complex, but most > compilers of other languages do exactly that. > I think the following expressions would be not only more > confortable but also more efficient than the equivalent > explicit conversion, since one cannot directly access to > the real and imaginary fields of complex records: [...] I agree in complete. I will see how difficult it is to implement these features. If they don't make too much trouble they might be in the next beta. Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [971005] maintainer GNU Pascal [971001] - http://home.pages.de/~gnu-pascal/ [971005]

1 0

Re: Gpc not accepting non-standard (?) systax.
by Frank Heckenbach 21 Oct '97

21 Oct '97

richard.kerry(a)quantel.com wrote: > Thank you for your help with the C/external problem, I've removed > 'external' and that part is now happy. Fine! :-) > Regarding the second problem, concerning a structured constant, or at least > what seems to be the gpc technique to achieve the effect of a structured > constant : > > The full chain of defs is as follows : > > const > mArgName = 16; > > type > ArgType = (OutputFileArg, InputFileArg, OptionsArg); > ArgClass = (StringArg, FileArg, NullArg); > ArgStatus = (RequiredArg, OptionalArg); > > ArgName = packed array[1..mArgName] of char; > iArgName = 0..mArgName; > > ArgDef = record > name: record > txt: ArgName; > amin: iArgName; > len: iArgName; > end; > pos: integer; > status: ArgStatus; > aclass: ArgClass; > end; > > ArgDefTable = array[ArgType] of ArgDef; > > type > ArgDefs = ArgDefTable value ( > ((' ', 1, 0), 0, OptionalArg, NullArg), > (('Output_File ', 1, 11), 2, OptionalArg, FileArg), > ... ); > > var > kludge : ArgDefs ; > I hope this is sufficient for you to sort out my problem. It compiled > successfully under gpc-970824 but fails under 971001. I think this is a form of the known problem. I'm surprised it did compile under gpc-970824 (I don't have this version here to check). Anyway, what you could do (until this problem is solved which might take some time) is to declare the constant values not with the type but rather with the variable(s). Though structured constants do not yet work completely like ISO requires, I guess your example should work. Try something like: [types until ArgDefTable, no type ArgDefs] var kludge : ArgDefTable value (...); Of course, if you have multiple such variables, you have to do this multiple times, which is not so nice. There could be a possible "dirty" solution using the preprocessor, something like: {$define ArgDefs ArgDefTable value (...)} var klugde : ArgDefs; -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Re: Gpc not accepting non-standard (?) systax.
by richard.kerry＠quantel.com 20 Oct '97

20 Oct '97

heckenb(a)mi.uni-erlangen.de on 20/10/97 15:14:59 To: gpc(a)hut.fi cc: (bcc: Richard Kerry/Quantel Research) Subject: Re: Gpc not accepting non-standard (?) systax. richard.kerry(a)quantel.com wrote: > The 'C' directive on external procedures. > > FUNCTION _p_paramcount : Integer; C; > EXTERNAL ; > FUNCTION _p_paramstr (num: Integer; VAR str: String): Boolean; C; > EXTERNAL ; > > It says : > : parse error before `External' The "C" directive implies "external". Earlier GPCs have accepted the above, the newer versions don't anymore. Just using "C" should do what you want. (Apart from this, there are still plans to make a more consistent syntax for C/External/AsmName... sometime...) > If I comment these out it fails on a structured constant definition using > 'value'. > > type > ArgDefs = ArgDefTable value ( > ((' ', 1, 0), 0, OptionalArg, NullArg), > etc... > > It says : > : parse error before `Value' > : invalid schema discriminants > > If I comment out 'value' it says > : parse error before `,' > : missing expression > : discriminated type is not a schema or string > : parse error before `,' > > Is there some configuration I need to do to make these work ? Or have I > found a bug ? If ArgDefTable is an array or record (or a combination of these), this is a known problem (since I don't know if this is required by the standard, I'm not sure if it's a bug or just a missing feature ;-). It's listed in my To-Do list at http://www.mi.uni-erlangen.de/~heckenb/gpc-todo_toc.html in the section "Records/arrays". If, however, a schema type is involved (the error messages say something about these), it might help to see the declaration of ArgDefTable. Did you perhaps use "string" without a maximum length? -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm Thank you for your help with the C/external problem, I've removed 'external' and that part is now happy. Regarding the second problem, concerning a structured constant, or at least what seems to be the gpc technique to achieve the effect of a structured constant : The full chain of defs is as follows : const mArgName = 16; type ArgType = (OutputFileArg, InputFileArg, OptionsArg); ArgClass = (StringArg, FileArg, NullArg); ArgStatus = (RequiredArg, OptionalArg); ArgName = packed array[1..mArgName] of char; iArgName = 0..mArgName; ArgDef = record name: record txt: ArgName; amin: iArgName; len: iArgName; end; pos: integer; status: ArgStatus; aclass: ArgClass; end; ArgDefTable = array[ArgType] of ArgDef; type ArgDefs = ArgDefTable value ( ((' ', 1, 0), 0, OptionalArg, NullArg), (('Output_File ', 1, 11), 2, OptionalArg, FileArg), ... ); var kludge : ArgDefs ; I hope this is sufficient for you to sort out my problem. It compiled successfully under gpc-970824 but fails under 971001. Thank You, Richard Kerry.

1 0

← Newer
1
...
5
6
7
8
9
10
11
...
16
Older →

Jump to page:

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 October 1997