Re: Bug#746761: libgtk2-perl: Sometimes FTBFS (t/GtkCellRenderer.t fails) on mips, mipsel and armhf

On Thu, May 15, 2014 at 11:42:42AM +0200, intrigeri wrote:
> Hi, dear mips and arm porters,
> 
> [email protected] wrote (03 May 2014 11:05:09 GMT) :
> > since April 2010 (1.1.221-6), libgtk2-perl sometimes fails to build on
> > the armhf, mips and mipsel architectures, due to a failing test
> > (t/GtkCellRenderer.t). I did not find any similar failure on other
> > architectures. I am going to report this upstream.
> 
> > Dear porters, any idea what could be wrong with this package, or
> > its dependencies?
> 
> Sorry to bother you again, but here's just a gentle ping: this FTBFS
> has now been blocking the migration of libgtk2-perl to testing for
> two weeks.
> 
> Note that there's been no progress yet on the corresponding upstream
> bug I filed:  https://bugzilla.gnome.org/show_bug.cgi?id=729453

I ran a loop building it overnight.  17 builds and no failure on my
armhf system.  How I wish people could make their builds use multiple
cores when -j is passed, then I could have run 4 times that number
of passes.  Seems adding '--parallel' to the end of the 'dh $@' line in
debian/rules solves it and makes it build a lot faster.

It sure is interesting that the failure it the same in all 3 cases,
but the only way that error could happen seems to be if $hits{render}++
is never called, and I don't see any obvious reason for that to happen,
especially only some of the time.

-- 
Len Sorensen