It's a bit of a hack, but you can see earlier posts in this list for how to force a build with no changes, but it won't honour your dependency. BTW it just uses curl and cron.
As for the bug in 1.0.3, I think only bugs in 1.1-alpha are being fixed by the commiters, so your best bet is to test on 1.1. Not great. BTW, you can do all this stuff in luntbuild. David On Wed, 4 Apr 2007 12:29:21 +0200, "Erik Ruisma" <[EMAIL PROTECTED]> said: > Indeed we use continuum 1.0.3. > > Ok we now got them running seperately, BUT only the first build is > actually > executed, the second build does not start as there are no changes in CVS. > Is there any way to force a build even if there are no changes? In fact > what > I really want is that the second build is only executed if the first > build > has been launched. > > So scheduling two builds for the same project at the same time does not > work, I believe that's a bug. > > On 4/3/07, David Roussel <[EMAIL PROTECTED]> wrote: > > > > I take it you've got them to run ok separately? Do then run ok when > > not scheduled together? > > > > Is this in continuum 1.0.3? > > > > > > On 3 Apr 2007, at 16:28, Erik Ruisma wrote: > > > > > Thanks for your proposal. That's what we did. > > > However this seems to create some new problems. > > > So what we have are two build definitions for one project: one > > > launching the > > > tests, one creating and deploying the site. > > > > > > We have scheduled both builds to run at the same moment. BUT only one > > > actually runs. > > > Why ? > > > I can find the following in our logs: > > > INFO | jvm 1 | 2007/04/02 20:00:00 | 2007-04-02 20:00:00,156 > > > [defaultScheduler_Worker-2] INFO Continuum - > > > Enqueuing > > > 'MyProject' (Build definition id=116). > > > INFO | jvm 1 | 2007/04/02 20:00:00 | 2007-04-02 20:00:00,156 > > > [defaultScheduler_Worker-2] INFO Continuum - > > > Enqueuing > > > 'MyProject' (Build definition id=118). > > > > > > Only the last one is executed. Is it possible that this is a bug in > > > continuum ? ie that MyProject is somehow used as the key and only > > > the last > > > element from the queue remains? > > > > > > What if we would use 2 different build times (to make things more > > > complex, I > > > don't prefer it): is it possible to force the last scheduled build > > > definition to run even if their are no modifications?? This seems > > > to be also > > > an issue...? > > > > > > > > > Any feedback welcome. > > > > > > > > > > > > On 3/15/07, David Roussel <[EMAIL PROTECTED]> wrote: > > >> > > >> Define two builds, one just to run the tests and report them. The > > >> second build to just do the site. > > >> > > >> On Wed, 14 Mar 2007 11:47:07 +0100, "Erik Ruisma" > > >> <[EMAIL PROTECTED]> said: > > >> > Hello all, > > >> > > > >> > I'm not quite sure if this is a continuum or a Maven question, > > >> but I > > >> > thought > > >> > to post first on the Continuum mailing list. > > >> > > > >> > In our company-wide settings I want that project sites are > > >> generated, > > >> and > > >> > artifacts get deployed to our internal repository even when > > >> there are > > >> > some > > >> > unit tests that fail. We also want that a mail is send when > > >> there are > > >> > test > > >> > failures, with some kind of indication that their was a problem > > >> during > > >> > the > > >> > build. > > >> > > > >> > How would you set this up ? > > >> > > > >> > > > >> > > > >
