Re: [Springframework-developer] Failed Builds

[<da...@da...>]

On Tue, Jan 10, 2006 at 02:27:07AM +0100, Alef Arendsen wrote:
> On the subject of automatic building, it seems the builds running at=20
> SF.net fail quite often (if not always). I haven't looked at all the=20
> error messages, but some come from binding errors, some I can't seem to=
=20
> find the error message for. Darren, are x86-linux and sparc-solaris the=
=20
> only platforms the builds are running on, or is it an issue related to=20
> those platforms only?

they do fail a lot, but certainly not always.  You can track their progress
at http://springframework.sourceforge.net/test/ and the page also shows
which platforms the builds are occurring on.  Remember too, that the SF
machines are continuously building and running the test suite, so they show
up more intra-day errors that may not be caught by the nightly build.  Often
when one or more fail during the day, it's confirmed with a similar failed
build messsage from your nightly build too - but I don't know if it's always
for the same reasons.

The SF machines are overloaded and underspec'ed, which accounts for some of
the errors (esp. "Proxy creation too slow" and some of the ones with no
stack traces or error messages).  I have offered, on behalf of UPCO, to host
a proper continuous build environment on dedicated hardware for Spring.  The
intention was to use one Linux and one Windows machine to do a similar job
to the SF compile farm stuff I run, but based on Luntbuild instead.  Had no
response on this.. probably slipped under the radar as I sent it to Rod and
Juergen early Dec.  (preparing for TSE and JavaPolis).  So, if it's
something you guys think the project would benefit from (I do!) the offer's
still open :)

Cheers,
--=20
Darren Davison
Public Key: 0xDD356B0D