This is the mail archive of the cygwin-apps@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: gcc package idea for consideration.


Okay, so rsyncing gcc-cvs for the first time took alot longer then I
anticipated.  So I did other things for most of today.  My closer look wasnt
so close, but i'd say things seem favourable at the moment.  I'm going to
attempt making a 3.2.3 cygwnized package on my own.  If I can do that by
next weekend, then you've got yourself an official stupid guy^W^W potential
replacement gcc maintainer.  (It looks easy enough, but proof in the pudding
and all. Not sure why proof, personally i'd usually suspect raisins.)

> >
> > I don't mind doing the merge to 3.3 if you're willing to be the
maintainer.
> > I've already started the dialog with Danny about how to go about this
anyway.
> >
> > cgf
>
> I haven't seen post Chris mentioned. I have been under heavy spam attack
lately
> and might have been overzealous with filters.
>
> This is my side of story:
> I have  3.2.3 and 3.3 CVS with cygming patchset and have been building and
> testing both regularly on cygwin and mingw.  I am about to release 3.2.3
> binariries for mingw and will upload the diff against CVS for that at same
> time.  So you could use that.  I could also upload a 3.3 diff.
Altenatively I
> could apply the diff's myself to a clean cygwin-mingw CVS branches

well, as a random nobody, I certainly wont be making any cygwin-mingw
branches at the moment.  I'd like to give my own hand at 3.2.3 - then i can
check against your diffs to see how clueless I am.  With 3.3 I'm not going
to say no to some assistance, as I must say I've payed more attention to
changes in the c++ front end then anything else over the last couple of
years.

>
> There are substantial structural changes between 3.2.x and 3.3 so be
warned to
> expect merge conflicts if doing the merge yourself.

And some minor ones between 3.2.1 and 3.2.3.
>
> 3.4 (FSF sources) is in relatively good shape.  The -mms-bitfield support
and
> fastcall support are official, as is the -mno-cygwin fudge, as indicated
by
> Chris. I haven't merged in the dll-exe exception hackery in my local tree
nor
> Mumit's old dllimport patches.  The former needs a lot of work to make it
> acceptable to FSF. Most of the latter are not really working anyway (they
don't
> do any harm either). I will try in near future to submit the working
pieces
> (like don't mark inline functions as dllimport) to GCC-patches.

I'm ashamed to say I hadnt noticed that -mno-cygwin had made it into 3.4
yet.  I do know that my source code is great for crashing the c++ front end
:)
That 3.4 is shaping up like this, makes the idea of being a maintainer much
more attractive.

>
> PCH (in 3.4) has problems, but still early days.
indeed.  I thought i had managed to use it a few weeks ago, but I suspect
I'm thinking of the wrong operating system.

>
> If your interested in Ada for cygwin, you will need a pre-existing Ada
compiler
> and runtime libs to bootstrap.

I dont think I would be interested in ada, but if I were to become gcc
maintainer, I'd be interested in coordinating with anyone who is.  I think
i'd like to see individual packages for each of the front ends.

>
> If you are interested in Java, Mohan Embar and Ranjit Matthews have
> mingw-specific patchsets for 3.3 (some may also apply to cygwin) that have
not
> yet made into official sources.

Although I've never used gcj - java is at least a language i have some (not
much) experience with, so I'd probably have a look at these.

Gareth - not staying up till 4am this night.


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