Re: pkg/51266: devel/gobject-introspection build fails on i386-7.99.32

["John D. Baker" <jdbaker%mylinuxisp.com@localhost>]

The following reply was made to PR pkg/51266; it has been noted by GNATS.

From: "John D. Baker" <jdbaker%mylinuxisp.com@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: pkg/51266: devel/gobject-introspection build fails on i386-7.99.32
Date: Sun, 7 Aug 2016 16:52:52 -0500 (CDT)

 On Sun, 7 Aug 2016, David Holland wrote:
 
 >  On Tue, Aug 02, 2016 at 10:05:01PM +0000, John D. Baker wrote:
 >   >  Such a build from the tertiary build host (quite a bit faster machine
 >   >  so it finished first) results in the same failure.  The last difference
 >   >  between my builds and the TNF builds is that I've been using
 >   >  "-V USE_PIGZGZIP=yes" while the TNF builds don't.
 >   >  
 >   >  I've been using that option for quite some time and I don't recall having
 >   >  any problem with -current until the switch to GCC 5.x.
 >  
 >  This is ... not good since the TNF builds use pigz.
 
 Oh?  I didn't see that in the "build.sh" command line in a recent snapshot
 build log.  Is it in the "mk.conf" file that's being used?  I recall
 someone blaming my use of "USE_PIGZGZIP=yes" for situations where the
 system build had stalled on something--usually 'cc1' spinning on some
 source file and not finishing.  They declared that the snapshot build
 hosts don't use it.  I haven't had that problem for a while and only
 recently disabled "USE_PIGZGZIP" (undefined) trying to make my local
 builds look more like those of the TNF build hosts.
 
 >  Thanks for tracking it down! Was clearly a major pain...
 
 To what do you refer?  So far, none of my changes have produced local
 builds that allow me to build "gobject-introspection".  If that does
 build and install (only on a pristine TNF-snapshot install with no
 existing packages), I still cannot build "multimedia/gstreamer0.10",
 "security/libsecret", "net/libsoup", "chat/telepathy-glib", or
 "multimedia/gstreamer1".  Nor can I restore that ability on a system
 which previously built these packages, but which subsequently began
 failing to to so.
 
 
 Since the last posting to this PR, all on-site hosts have passed at
 least 3 passes of "memtestplus" and the semi-diskless amd64 test host
 finished building my usual complement of packages while running only
 TNF snapshots and following the update to 201608060920Z is still able
 to build all of the problem packages.
 
 Meanwhile, the i386 test host, running only TNF snapshots from local disk,
 successfully built my usual complement of packages, except "libreoffice"
 which is blocked by pkg/51221.  Following an attempt to build "openjdk8"
 anyway and incurring the error in pkg/51221, the system subsequently
 lost the ability to build any of the problem packages.
 
 The amd64 test host, running a local build from disk, posted the following
 summary from 'sudo atf-run | atf-report':
 
 Summary for 657 test programs:
     4544 passed test cases.
     38 failed test cases.
     38 expected failed test cases.
     78 skipped test cases.
 
 -- 
 |/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
 |\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
 | X  No HTML/proprietary data in email.   BSD just sits there and works!
 |/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645