develooper Front page | perl.perl5.porters | Postings from March 2000

RE: Disposition of Modules

Thread Previous
From:
Konovalov, Vadim
Date:
March 19, 2000 22:49
Subject:
RE: Disposition of Modules
Message ID:
402099F49BEED211999700805FC7359F46B2DF@ru0028exch01.spb.lucent.com
> From: nick@ing-simmons.net [mailto:nick@ing-simmons.net]
> Vadim Konovalov <vkonovalov@lucent.com> writes:
> >I compiled Tk800.019 with BC++5.2 with almost no problems .
> >(I did 3 small modifications to platform-specific source code to pass
> >compile, and that was a good exercise to the reader :) All 
> tests for Tk
> >passed ok. 
> 
> If you have not already (my mail is playing up) can you send me any 
> Tk fixes ?

Another problem with compiling of Tk using BC++5.2 is related to compiler.
It occures that when BC++5.2 sees in "perl.h" a line
#include "patchlevel.h"
it incorrectly includes "patchlevel.h" from Tk distribution instead of 
including the CORE perl's file with the same name, and PERL_VERSION-like
defines are unknown to preprocessor.
It seems a bug in compiler (preprocessor) to me, and there is no such
misbehaviour with BC++5.4 and BC++5.5 compilers. GCC and MSVC compilers
are free of this bug too.
I don't know how to fix that, and I'm not sure there is need to be a fix.
Renaming Tk's "patchlevel.h" will solve a problem, but it's hardly a good
solution.

The last thing I usually fix to compile Tk by BC++5.2 is commenting out
<zmouse.h> in pTk/tkWinX.c (line 25):
#ifndef __GNUC__       // not in Minw32 yet
#include <zmouse.h>
#endif

or just
#if !defined(__GNUC__) && !defined(__BORLANDC__)
I don't know a logic when zmouse.h exists, so I don't know how to fix that
correctly.

Good luck,
Vadim

Thread Previous


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About