[Bsd-sharp-list] Problem building gnome-sharp20 on -CURRENT

[Tom McLaughlin]

On Thu, 2007-07-05 at 14:46 +0000, tmclaugh at sdf.lonestar.org wrote:
> > On Tue, 2007-06-26 at 21:02 -0400, Phillip N. wrote:
> >> Hi Joe,
> >>
> >> El dom, 10-06-2007 a las 17:05 -0400, Joe Marcus Clarke escribió:
> >> > On Sun, 2007-06-10 at 16:02 -0400, Tom McLaughlin wrote:
> >> > > On Sun, 2007-06-10 at 13:49 -0400, Joe Marcus Clarke wrote:
> >> > > > While I'm sure Kris has reported this to you already,
> >> gnome-sharp20 does
> >> > > > not build on -CURRENT after the gcc4.2 import:
> >> > > >
> >> > > > http://www.marcuscom.com/tb/errors/7-CURRENT-FreeBSD/gnome-sharp-2.16.0_3.log
> >> > > >
> >> > > > The problem seems to be due to source file corruption.  The file
> >> in
> >> > > > question, Settings.cs, actually has one of those Thread error
> >> lines at
> >> > > > line 212:
> >> > > >
> >> > > > "Thread 285ded00 has exited with leftover thread-specific data
> >> after 4
> >> > > > destructor iterations"
> >> > > >
> >>
> >> Does this happend alwais? or "just" sometimes?
> >
> > Always on i386.  Tomboy isn't included in amd64 for some reason, so I
> > haven't seen it there.
> >
> 
> Last time I enabled amd64 support it hung while building on pointyhat. 
> Phillip says it works in his -CURRENT amd64 jail so I'll reenable it and
> send Pav/Kris a heads up when I do it.
> 
> >>
> >> I just setup a current (from today) tinderbox amd64 here:
> >>
> >> reeBSD siete.sofsis.cl 7.0-CURRENT FreeBSD 7.0-CURRENT #0: Tue Jun 26
> >> 16:23:59 UTC 2007
> >> killfill at siete.sofsis.cl:/usr/obj/usr/src/sys/GENERIC  amd64
> >>
> >> And it just compiled gnome-sharp20 just fine:
> >> http://sofsis.cl/gnome-sharp-2.16.0_3.log
> >>
> >> > > > I'm not sure how this line got in there, but if one manually
> >> removes it,
> >> > > > then resumes the port build, it will complete successfully.  It
> >> would be
> >> > > > great if you guys could take a look at this as it is preventing
> >> GNOME
> >> > > > from building on -CURRENT.  Thanks.
> >> > > >
> >> > > > Joe
> >> > > >
> >> > >
> >> > > What's the date of your -CURRENT?
> >> >
> >> > Host and TB are from yesterday's -CURRENT.
> >> >
> >> > >
> >> > > FreeBSD 7.0-CURRENT #0: Sun May 27 15:53:33 EDT 2007
> >> > >     root at releng-7:/usr/obj/usr/src/sys/GENERIC
> >> > >
> >> > > (host and tb chroot are kept in sync.)
> >> >
> >> > I've seen this now in TB, on my -CURRENT MacBook Pro, and in
> >> pointyhat's
> >> > logs.  There is definitely something going on here, but I haven't
> >> looked
> >> > to see how that Thread line is making it into the Settings.cs.
> >> >
> >>
> >> Your macbook has only 1 cpu, right?
> >
> > One dual core CPU, so effectively two.  So does my Tinderbox machine
> > (thanks to HTT).
> >
> >>  (i guess becouse you use generic.. :)
> >
> > And GENERIC with -CURRENT has SMP enabled.
> >
> >>
> >> What about the i386 pointy mashine, Is it SMP?
> >
> > Some of the nodes may be, yes.
> >
> > Joe
> >
> 
> My latest tinderbox builds are just finishing on a -CURRENT as of July
> 2nd.  The BSD# tree build with Mono 1.2.4 built all packages fine.  I am
> seeing the gnome-sharp fail on the FreeBSD tree build.  Once the latest
> run ends I'm going to delete mono and some other packages and build them
> again so everything is built with the current gcc and with the latest
> CFLAGS and see what happens.
> 
> Phillip, if every package succeeds to build in the 7-i386-BSDSharp build
> you need to tell me when to merge everything into the FreeBSD tree.

After some toiling a bit I've found that my results periodically change
when rebuilding gtk-sharp10 and gnome-sharp20 in both builds.  If I
repeatedly run each build enough times eventually all all packages will
build.  Below is the error from the gnome-sharp logs:

Settings.cs(212,8): error CS8025: Parsing error
Compilation failed: 1 error(s), 0 warnings
Thread 285ded00 has exited with leftover thread-specific gmake

That third line doesn't appear in every log I looked at but it does in
some.  Better yet, the ports build repeatedly without incident in my
-CURRENT jail.  I have a feeling this is the same pointyhat build
failure which has happened periodically which my tinderbox could never
reproduce.  It appears a change to -CURRENT has made Mono a little more
prone to error.  dmesg on the box shows no core having been dropped so
Mono didn't crash which makes this issue all the more  better to try and
diagnose.

Any Mono users up for being takers on this problem?

tom

-- 
| tmclaugh at sdf.lonestar.org             tmclaugh at FreeBSD.org |
| FreeBSD                                   http://www.FreeBSD.org |