Re: Bug#404733: Mozilla-based and related packages status
On Sun, Jan 21, 2007 at 09:56:40AM +0100, Alexander Sack wrote:
> On Sat, Jan 20, 2007 at 08:06:08PM -0800, Steve Langasek wrote:
> > > The codebase is the same as the one for iceape and icedove currently in
> > > unstable... the only difference is the way it is linked altogether in a
> > > single big dynamic library instead of a single big binary program.
> > Yeah, I've done some test builds now of xulrunner on my alpha, and this
> > isn't a regression in xulrunner; it's either a regression in the toolchain,
> > or a change in some underlying lib that trips the already known bug in
> > binutils on alpha causing link times to scale poorly with the number of
> > symbols.
> ... what uploads might have caused this regression?
Looking at the build-deps, gnome libs seem the likely candidates.
> > So the only fix needed here is to raise the timeouts on the buildd for
> > iceweasel, iceape, icedove, and xulrunner to an obscene number.
> Sorry for my ignorance, but would this timeout-fix mean that builds
> for alpha will take ages in future (e.g. when building security
My xulrunner link test on a 533MHz EV56 machine took 753 minutes, and
accounts for over half the total build time. The alpha autobuilder is an
ev6 machine, which seems in general to be about 2-3x as fast as my box.
So that build time is pretty damning indictment of current binutils on
alpha, but still wouldn't make alpha the slowest arch for building
xulrunner (arm took 21h to build it last).
Steve Langasek Give me a lever long enough and a Free OS
Debian Developer to set it on, and I can move the world.