Validate failures (again)

[Claus Reinke]

> Running validate before pushing changes is becoming somewhat annoying 
> because I have to run it twice in two different repos (one with and one 
> without my changes) and then compare the results. Why do people push 
> without validating? I thought we agreed that this shouldn't happen?

Given that the buildbots still haven't recovered (windows
was just dropped from the nearly empty daily summaries - not 
seen in 7 days..), any patches applied recently run the risk of 
breaking builds on other platforms. According to 

http://darcs.haskell.org/buildbot/head/one_line_per_build

successful builds have been rare recently.

Until that stabilizes again, perhaps you should count yourself 
lucky that your builds get to the tests? As I sent earlier, my 
windows head build simply fails (when updating my repo every 
other month or so, I usually have to wait for a window of 
opportunity for the windows build, but this time, the effect
isn't that platform specific).

I assumed the buildbot issues were related to the move,
and that buildsystem fixes would start when the buildbots
were back. Are there any estimates as to when the buildbot 
situation will get back to normal, so that build breaks and 
validate issues will be more immediately visible again?

Claus