From: <dan...@ya...> - 2001-02-26 10:10:29
|
--- Jose Fonseca <em...@fe...> wrote: > > > On Mon, 26 Feb 2001, [iso-8859-1] Danny Smith wrote: > > > --- Danny Smith <dan...@ya...> wrote: > Date: > Mon, 26 > > Feb 2001 12:51:27 +1300 (NZDT) > > > From: Danny Smith <dan...@ya...> > > > Subject: Re[2]: [Mingw-users] incompatible pointer type > > > To: Danny Smith <dan...@ya...>, > > > "José_Fonseca" <em...@fe...>, > > > min...@li... > > > > > > > > > --- Danny Smith <dan...@ya...> wrote: > --- > > > José_Fonseca <em...@fe...> wrote: > > > > > > ----- Original Message ----- > > > > > From: "Earnie Boyd" <ear...@ya...> > > > > > To: "José Fonseca" <em...@fe...> > > > > > Cc: <min...@li...> > > > > > Sent: Saturday, February 24, 2001 7:24 PM > > > > > Subject: Re: [Mingw-users] incompatible pointer type > > > > > > > > > <snip> > > > > > > > > > > > I've checked on Linux (replacing __stdcall by > > > > > __attribute__((stdcall)) > > > > > ofcourse) and the behaviour was the same - it's really a bug > in > > > the > > > > > standard > > > > > gcc... > > > > > I believe we should report it to the gcc development team, > don't > > > > you? > > > > > I believe this should be the course to go since there is > always > > > the > > > > > chance > > > > > of some originally vc++ code induce the same bug... > > > > > > > > > > While this isn't fixed on gcc, I'm going to make a awk script > to > > > > fix > > > > > automatically all headers of w32-api. This way there isn't > the > > > need > > > > > of > > > > > messing so much with the headers. > > > > > > > > > > José Fonseca > > > > > > > > > > > > > > > > > > A closely related bug was reported to GCC-bugs by Mumit over a > year > > > > ago: > > > > > > > > http://gcc.gnu.org/ml/gcc-bugs/1999-11n/msg00084.html. > > > > > > > > Danny > > > > > > > > > > And here is the patch: > > > http://gcc.gnu.org/ml/gcc-patches/2001-01/msg01751.html > > > > > > I'll try to apply tonight to 2.95.3-prerelease and see what > happens. > > > > > > Danny > > > > > > > Yes, applying Bob Wilson's patch (above) to gcc-2.95.3 fixes the > bug > > that Jose reported as well gives correct result when applied to > Mumit's > > testcase. > > > > IMO fixing gcc is preferable to changing the headers. Should this > be > > done for gcc-2.95.3 (due for release this week I think)? > > > > Danny > > > > That's great! Thanks! > > How stable is gcc-2.95.3 with the Mingw build environment? > I've successfully made a cross compiler with the patches in > sourceforge > project page, but the win32 native version makes memory access > violations... > > Jose Fonseca > I've been using 2.95.3 prelease test1 and then test2 (both with Mumit's Jan 2000 patches applied) since 05-01-2001 with no obvious probems. Make bootstrap used mingw gcc and binutils to compile and link, but cygwin tools for configure and and make environ. Using 2.95.3, have rebuilt several fairly large C++ libs --STLPort, WFC50 (Windows Foundation Classes which is a NT services lib, similar to MFC, but without the GUI baggage), FLTK, ODBC++, MetaKit-- their regression testsuites and projects which depend on these libs. But....I haven't been able to come to grips with gcc testsuite so there may be corner cases that will blow it out. Also I don't use GUI much so, apart from FLTK tests, really only tested console apps. Danny _____________________________________________________________________________ http://invites.yahoo.com.au - Yahoo! Invites - Organise your Mardi Gras party online! |