On Mon, Jul 28, 2008 at 8:13 PM, James Keenan via RT <[EMAIL PROTECTED]> wrote: > On Mon Jul 28 10:48:21 2008, particle wrote: >> On Mon, Jul 28, 2008 at 10:27 AM, Eric Wilhelm >> <[EMAIL PROTECTED]> wrote: > >> > >> > Tests need to be written defensively for arbitrary parallelization to be >> > possible. > > The configuration step tests have been evolving for 13 months now, but > until today there was no requirement for defense against arbitrary > parallelization. While I think the goal of being able to run tests in > parallel is a good one, I also think that it was premature to apply this > patch so soon after its submission, as that did not leave us sufficient > time to evaluate its impact.
FYI, unless you set an environment variable and have Test Harness 3, there is no impact. No one's expecting the test suite to be immediately ready for parallel testing, but this gives us an easy way to kick the tires. > >> > >> i encourage you to try this out in the 'parallel' branch of the >> repository. > > This is a good suggestion, but I have to caution that the branch name > 'parallel' was *not* chosen to signify the type of testing which is the > objective of Eric's patch. > >> it's not synchronized with the latest changes to trunk, >> but you can do that locally (for a refresher, read >> docs/project/committer_guide.pod). work has been done to consolidate >> the test files in that branch, and although they may not address all >> the requirements for parallel testing to succeed, that is a related >> goal and seems like a good place to achieve them. >> ~jerry >> > > I'll be submitting a patch in RT 56716 tonight or tomorrow which > represents the merge of the 'parallel' branch into trunk. Normally, I > allow 2-3 days for inspection before I actually do the merge. > > Jerry: If you want me to merge on a faster schedule than that, please > let me know. But, in any event, since these tests (like, probably, all > our tests) were not necessarily developed with a requirement of parallel > testing in mind, I can't guarantee in the short run that they'll work > with 'make test' in parallel. > > Thank you very much. > > kid51 > > I agree with Jim here: the tests in this branch aren't more likely to pass the parallel testing. Given that updating a test to make it more conducive to testing in parallel won't affect running it by itself, I don't see why those patches (once they come in) can't just go straight to trunk. (Presuming they otherwise pass muster.) Regards. -- Will "Coke" Coleda
