Steve wrote:
> On 1/6/05 00:42, in article
> 1117582976.124298.186480@[EMAIL PROTECTED]
"toby"
> <toby@[EMAIL PROTECTED]
> wrote:
>
> > I'd be happy to help.
>
> Thank you. If it's alright, I'll keep your email address on file just in
> case! :)
>
> >> The only thing that seems to be missing from the toolset is gdb. Have
you
> >> had any success in remote debugging OS X host -> Windows target? Is
it
> >> simple to do?
> >
> > I have not done that, as I rarely use symbolic debuggers, but it would
> > be an interesting thing to try...
>
> I'm just starting to read the gdb manual, to be honest I wouldn't have a
> clue where to start. I'll carry on reading for now... :)
Failing that, you can always run an IDE like Eclipse under Windows,
hooked up to MinGW toolset. It's probably easier to get started
debugging that way.
>
> > By the way I recently updated the build-cross script for more recent
> > MinGW release files:
> >
> > GCC_VERSION=3.4.2-20040916-1
> > BINUTILS=binutils-2.15.91-20040904-1
> > MINGW=mingw-runtime-3.7
> > W32API=w32api-3.2
>
> How do you know which versions to use? Are they the same as the
'official'
> MinGW download for Windows?
I probably wouldn't have updated beyond my last post on the subject,
except I needed to do a new build on a new system and found a couple of
those distribution files had gone missing. I take my cue from the
"Current" file list at
http://sourceforge.net/project/showfiles.php?group_id=2435
>
> > The gcc configure command now looks like:
> >
> > "$SRCDIR/$GCC/configure" --with-gcc --with-gnu-ld --with-gnu-as \
> > --with-sys-root="$PREFIX/$TARGET" \
> > --without-newlib --disable-multilib \
> > --target=$TARGET --prefix=$PREFIX \
> > --enable-threads --disable-nls --enable-languages=c,c++ \
> > --disable-win32-registry --disable-shared
> >
>
> Thanks. Out of interest, what difference do the new configure options
make?
Not much - mainly I decided to use the new --with-sys-root option.
>
>
> Using gcc it's interesting to watch it re****t more warnings and errors
where
> Codewarrior let it through without any problems. ... I was always under
the impression that Codewarrior was leading the way wrt.
> standards compliance?
My opinion is each compiler has different blind spots, so the best
thing is to show your code to as many of them as possible. Where I'm
working, we've found that gcc with all warnings (-Wall) spots a lot of
things that MSVC lets through. ****ting to gcc therefore has a generally
positive effect on code quality. In fact, the format-string checking of
gcc alone (an example of where it goes beyond the call of duty) has
pre-emptively picked up some serious ****tability-type issues.
--Toby
>
> --
> Steve.
|
