On Jan 6, 2010, at 9:29 AM, Michael Roberts wrote:
stef: great. Yanni: I agree. I used to run the tests myself and publish them to the google wiki for the end of 1.0 cycle. it did not scale ;-)
I see the two efforts as complimentary. I am really working from the inside out. Your approach is outside in. For example I have refactored the image update code so that it is a bit more manageable. Detecting the build failures is important, for example as I wrote my post, an update stalled with EOCD error in MC code.
this is cool!
hudson looks good. all the dashboard and easy install stuff for free. do you regularly use & maintain it?
cheers, Mike
On Wed, Jan 6, 2010 at 7:08 AM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
mike
I will ask in my team if our infrastructure can be used. Damien, marcus do you know the servers we have?
Stef
On Jan 5, 2010, at 10:27 PM, Michael Roberts wrote:
Hi, following on from the timely Hudson post ;-) I too have been working on continuous integration. I am more interested in identifying & integrating the various bits of plumbing first rather than the build solution itself. I figured the infrastructure needed to be understood first, and there could be value in such an interim solution
To this end I have: -written a simple image builder that updates the image in place. -running remotely on a Linux host I have borrowed access -building PharoCore 1.1 from #11110 onwards -started to publish images at http://pharo.gforge.inria.fr/experimental/
remaining is to rename the core images to unstable and then look at running some tests. This has exposed various issues, mainly image timeout caused by the various dialogs that can popup when filing in code.
so 1) if anyone would like to test the integrity of these experimental auto-build images, please do 2) discuss if something in this namespace http://pharo.gforge.inria.fr/<> is suitable for automatic publishing - i am not sure how you automatically create gforge "releases" but for core images perhaps this is not needed 3) find a real server we can run this on, (if this is considered worthwhile). My requirements are Linux + daemontools + vnc server - i can host my solution for a while, we need something strategic
then once we have a server I'm happy to look at Hudson or whatever we consider the right tool for doing this. I have also been playing with Mason and sent some feedback to Colin. Like i say, for me the real issues are running a Pharo image completely automated.
Cheers, Mike
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project