Re: [MinGW-dvlpr] [Mingw-users] fstat stub in libmoldname90.a

[Keith M. <kei...@us...>]

On Monday 15 February 2010 04:29:58 Chris Sutcliffe wrote:
> > Eventually I found:
> >
> > http://sourceforge.net/tracker/index.php?func=detail&aid=2134161
> >&group_id=2435&atid=302435
> >
> > Apparently there's been a number of attempts to fix this,
> > but none merged.
>
> I've applied the patch (with some minor modifications due to the
> age of the patch) to CVS head.  This issue seems to be a
> re-occurring one, and I think it reasonable to apply the patch
> (the information is publicly available on MSDN).
>
> Are there any objections to me committing the patch?

I can't see any obvious reason not to.

I believe the reason this has languished so long is, at least partly 
due to the sheer size of the patch, coupled with its applicability 
to so many disparate functional groups; I, for one, would have much 
preferred to see it submitted in a number of separate, and logically 
cohesive chunks.

Some points to consider:

1) Purely cosmetic, but I would favour _CRT_ALIAS over _CRTALIAS, as 
the newly defined declarator macro.  However, you are the maintainer,
Chris, so it's your choice.

2) Are we confident that the use of __MSVCRT_VERSION__ >= xxx is a 
satisfactory selector within .def.in files?  We've previously always 
explicitly identified each specific library individually, where 
symbols are not universal.

3) I probably don't need to say it, but take care when updating the 
ChangeLog; it may be worth reminding the poster that ChangeLog text 
should *not* be included within the scope of the diff.

4) Not a reason to delay applying the patch, but we may wish to 
follow up with redirector stubs for the removed symbols, as suggested 
by Daniel Jacobowitz; it may not be strictly MSVC compatible, but if 
it makes end users lives easier, I'm in favour.

-- 

Regards,
Keith.