> The failure is happening just because it can't possibly succeed, I set
> up the account for the buildbot in such a way that it has no access to
> a GUI context. I'm going to rectify that today so I can properly test
> TK.

Well, a nice thing would be for tests to be properly skipped in this
situation, rather than fail or crash :) Do you think you can try to
write a patch for this?

Absolutely, that's on my TODO list. First, figuring out the buildslave control process; then isolating where that crash is happening and reporting it to whomever (which, I think in a cursory look, is actually the TK folks); then figuring out how to convert no-GUI-context-possible situations into a skip.
">

(original) (raw)

On Fri, Oct 8, 2010 at 8:00 AM, Antoine Pitrou <solipsis@pitrou.net> wrote:


Hi,

> The failure is happening just because it can't possibly succeed, I set
> up the account for the buildbot in such a way that it has no access to
> a GUI context. I'm going to rectify that today so I can properly test
> TK.

Well, a nice thing would be for tests to be properly skipped in this
situation, rather than fail or crash :) Do you think you can try to
write a patch for this?

Absolutely, that's on my TODO list. First, figuring out the buildslave control process; then isolating where that crash is happening and reporting it to whomever (which, I think in a cursory look, is actually the TK folks); then figuring out how to convert no-GUI-context-possible situations into a skip.

--S