Re: "Re: libbfd, libtool & Win32" and "Re: Building a MinGW GLib etc..."

[Guido Draheim]

Max Bowsher wrote:
> Robert Boehne wrote:
>
> > Max,
> >
> > I took a look at the patch, and I don't have any objections
> > to it being checked in to Libtool.  Can I get some comment
> > from the mingwin people before I take the leap and
> > check it in?
>
>
> Well, ummm... thankyou!
>
> But perhaps I should remove/fix some of the work-in-progress comments in there?
> Like "Let's try this...". I did say "work-in-progress", intending to tidy up and
> document reasons for the changes before submitting for checkin.
>
> Also, since that patch is the first ever time I have looked at the libtool
> source, I would appreciate scrutiny from the libtool side. Whilst almost
> everything I've touched is inside mingw/cygwin conditionals, I should mention
> that I do not know anything about writing portable Bourne shell.

well, you have me asking for scrutiny atleast ;-) I did not get the point
on the first coarse reading but I had a hard day anyway, and Robert does
already say it okay, and he's the best in reading libtool patches :-=)

as for writing portable bourneshell: crosscompiling is not a task that
runs on oldish hardware. don't worry, and if someone comes along later
telling of problem, it can be fixed then ;-) and ming/cygw native
compiling is not run in a problematic bourne shell, right...


> There is one major decision that needs to be made: Keep or remove support for
> old versions of the mingw toolchain.
> All the -mdll stuff is deprecated in favour of gcc -shared now.
>
> Option 1) Remove support pre "gcc -shared" toolchains
> Pro: simplifies maintenance.
> Con: might upset some people.

How old may a gcc/binutils pair be? My oldest crosscompilers
are gcc 2.95.3 and ld --version reports 2.11.90.8. And for
all I know, these are in fact the oldest versions around,
no one want to go back beyond, I guess.

Is that enough, Max?

> Option 2) Keep support for old syntax toolchains
> Pro: does anyone actually use non-current mingw?
> Con: Messy & probably not often tested now.

see other. If all breaks on that one, we might try with
subclassing build environment with another mingw*
extension for the star that can be tested for in
sensitive parts. Know what I mean?