Re: cygwin breakage

[Gary V. Vaughan]

Charles Wilson wrote:
> Gary V. Vaughan wrote:
>
> > No, no!  Let's not revert it.  Otherwise we will hit exactly the same
> > problem in HEAD when we first try to use it on Cygwin.
>
>
> Then what is the purpose of having a branch-2-0 vs. HEAD? We finally got 
> branch-2-0 working on cygwin; yet one destabilizing change and poof! 
> three failures and eight skips.
>
> I've no problems with leaving Peter's stuff in HEAD; at least there we 
> have some time to debug it.  But I'm fearful of an impending 2.0 release 
> (that's why there's a release branch, right?) where any random "cleanup" 
> or "cosmetic" patch can break existing functionality.

I agree with *almost* entirely here.  The reason I'm keen to fix this 
quoting problem on the branch is that I am planning on replacing all the
$show/$run cruft with a new function based on func_quote_for_eval (on 
HEAD after the release), and it will be a lot easier to discover what is
wrong with func_quote_for_eval on cygwin with Peter's patch than when 
libtool uses a similar trick in dozens of places and basically falls to
pieces on cygwin.

> If it's a release branch, treat it like a release branch.  If it's not, 
> the make it HEAD, move HEAD to "garys-favorite-playground" and be done 
> with it.

I have libtool--release--2.0 (mirroring branch-2-0), libtool--devo--1.0 
(mirroring HEAD) and libtool--gary--1.0 (garys-favourite-playground) :-)

> > In a moment of madness (it must this flu I have), I'm trying to 
> > install virtual PC on my mac, so I can install windows 98 on it, so I 
> > can install cygwin on that, so I can install libtool branch-2-0 on 
> > that. I'll see if I can figure out what is going on.
>
>
> That's good new.  INSANE.  But good.
> It takes me about an hour to bootstrap, build, and run the test suite on 
> libtool/cygwin, on an ath64 3000 (~ 1.8GHz clock speed).  I tremble in 
> fear: virtual PC.  <shudder>

I believe my Virtual PC is about equivalent to Duron-700, though it 
feels a lot slower than my old Duron 700 hardware... The bootstrap and
build took about 30 minutes, so I'm just gonna run the failed bits of
the testsuite.

> I don't have cygwin's tetex installed, but the texmf postinstall script 
> does this:
>
> texconfig rehash
> texconfig confall
> texconfig rehash
> texconfig init
>
> so I can see why it might take a while...

Yeah.  I probably didn't need it either, but I can't run distcheck on
my powerbook unless I have fink's tetex installed, so I put it in out
of habit really.

Cheers,
        Gary.
--
Gary V. Vaughan      ())_.  address@hidden,gnu.org}
Research Scientist   ( '/   http://tkd.kicks-ass.net
GNU Hacker           / )=   http://www.gnu.org/software/libtool
Technical Author   `(_~)_   http://sources.redhat.com/autobook

signature.asc
Description: OpenPGP digital signature