On 6/8/07, Mat Schaffer <[EMAIL PROTECTED]> wrote:
Hi, folks. I read Assaf Arkin's June 6 post about Rake and circular task dependencies and felt like hacking ruby for a bit. So I put together a patch against rake's trunk for the code he worked out. Original post: http://blog.labnotes.org/2007/06/06/rake-hack-im-not-sleeping-just- caught-in-a-circular-dependency/ Patch on trunk is attached. I modified it slightly because (a) his code wasn't against trunk and (b) it looks like separate threads are already getting separate stacks without Asaf's modification of the MultiTask class.
Mat, Thanks. We have some test cases [1] (RSpec, only the first set is relevant). Each thread indeed gets a new stack, but without the change to MultiTask, each of the parallel tasks will start from a fresh stack, and won't be able to detect circular dependencies that started before the multitask. The last test in [1] checks that and will fail without this change to MultiTask. Assaf [1] http://www.intalio.org/buildr/trunk/test/rake_ext.rb Hope attaching the file is sufficient. Let me know if you'd prefer
and inline patch or some other kind of patch. This just 'svn diff'. -Mat
_______________________________________________ Rake-devel mailing list [email protected] http://rubyforge.org/mailman/listinfo/rake-devel
