This is the mail archive of the cygwin@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Potential problems with cygwin GCC and -mno-cygwin switch


On Tue, Jan 08, 2002 at 05:29:14PM -0500, Soren Andersen wrote:
>This lack of sponsorship maybe is also part of the noted tendency for
>minGW priciple persons to manifest some, uhh, let's say testiness.

I've been reading the mingw mailing lists for a while and I really don't
see anything like this.  Most of the replies are very courteous.

They don't seem to have anyone like me, for instance.  :-)

>I've seen all this from certain people involved in minGW.  Overall,
>though, its an amazing thing that minGW even exists, and has
>accomplished as much as it as.

I really don't see very much of this at all.  I'm surprised to see this
observation.

>One thing that is pretty clear to me is that there is no one person,
>aside maybe from Mumit Khan, who can legitimately present him/herself
>as "speaking for" minGW.  I think that needs to be acknowledged if
>there's been some impression that "minGW is criticizing cygwin".  minGW
>is first and foremost a free-for-all, a collaborative exercize that
>moves forward by fits and starts.  In any such assemblage of
>personalities there are bound to be some outspoken individuals (no
>sh__:-) who express frustrations they are having in a way that isn't
>echoed by more silent participants.

There is a group of core MingGW maintainers, or at least that's what I
understand.  A couple of the MinGW maintainers have actually indicated
that they still use cygwin for building their compiler tools.

And, as may have been noted, the mingw web page is really not wrong.
MinGW support in cygwin *is* flaky and we *have* talked about
deprecating it.

>>From what I've seen, it looks like MinGW support in Cygwin GCC is
>>up-to-date and better than ever before.  So, I have no idea what the
>>MinGW web site is referring to.  Does anyone from Cygwin agree that
>>MinGW support will be deprecated?
>
>I hope not.  I am going to be studying the responses to this msg for
>the next several days in an attempt to understand WHAT they are talking
>about (argh).  I gather that it is mostly about linker scripts which i
>have never understood very well (and to tell the truth, hope i don't
>have to).

Off the top of my head, there are a few issues with mingw support in
cygwin gcc (I think most if not all have already been mentioned):

1) It's supported by me currently.  While I have no problem with mingw as
   an entity, it's not my project, and maintaining the gcc/ld aspects do
   not thrill me.  I have a few patches in my tree that are not part of
   the standard gcc offering.  That's one reason why gcc 3.x built from the
   official gcc release will behave differently from the cygwin gcc 2.95.3.

2) While I have gone to some pains to isolate header files in the -mno-cygwin
   case, I didn't do the same thing for libraries.  That means if you do
   'gcc -mno-cygwin foo.c -lncurses' ld will attempt (and fail) to link the
   cygwin version of ncurses into your program in some cases.

3) c++ support doesn't work since we don't provide a mingw version of
   libstdc++.a.  This has been on my tuit list for a while, though.

4) 'gcc -mno-cygwin -print-some-gcc-thing' doesn't work right.

>>While I do think Cygwin GCC currently does a great job of supporting
>>MinGW, I do have a few issues with it:
>{snip details}
>
>Hopefully this can all get resolved peacefully and harmoniously.  The
>one thing I hope is that the collective attitudes at minGW never get to
>the point where people "over there" (some of whom are also "people over
>here") have forgotten the debt of appreciation they owe to cygwin, for
>being the historical predecessor and "host" that allowed them to come
>into existence, if for nothing else.

There is no "over there".  The MinGW maintainers are a friendly bunch.
I scan the MinGW lists for cygwin issues and a number of them read the
cygwin list as well.

So, please don't invent any antipathy between our two groups.  I have
never seen anyone badmouth cygwin in the mingw mailing lists.  In my
opinion MinGW is a sister project and should be treated as such.

gcc -mno-cygwin isn't going anywhere.  I somtimes speculate that we will
be deprecating it but since this switch is required to build some things
in the 'winsup hierarchy' we really can't do that.

I've also speculated that gcc -mno-cygwin should just run a mingw cross
compiler but that is rather infeasible, too.  It means that if you
are building the winsup hierarchy from scratch you have to somehow
also build a completely separate compiler and linker.  There is no way
that I even want to imagine the Makefile nightmare necessary to accomplish
that.

So, what is needed is someone to fix gcc, ld, and whatever to do the right
thing.  Barring that, gcc -mno-cygwin will remain relatively stagnant.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]