For comparison, running the layout tests in a release Purify build took about 20 hours last time we did that. We now split it into 1-hour chunks for convenience and to divide the eggs into multiple baskets.[1] A debug Purify build is too slow to be worth running.
- Pam [1] http://idioms.thefreedictionary.com/put+all+eggs+in+one+basket On Sat, Mar 28, 2009 at 3:21 PM, Dan Kegel <daniel.r.ke...@gmail.com> wrote: > > The change that adds support for valgrinding layout tests is > http://codereview.chromium.org/55034 > > I'm about about 10% of the way through valgrinding the layout tests > on a debug build, and about 20% of the way through on a release build. > With valgrind set to ignore "reachable" and "possible" leaks, and using > the valgrind/suppressions.txt files in the tree to suppress previously > known problems, and not paying much attention to whether > tests are actually passing, > the vast majority of new valgrind warnings I see are the two memory leaks: > http://code.google.com/p/chromium/issues/detail?id=9475 > http://code.google.com/p/chromium/issues/detail?id=9458 > I've also seen two invalid read errors: > http://code.google.com/p/chromium/issues/detail?id=9486 > http://code.google.com/p/chromium/issues/detail?id=9488 > > All in all, it looks pretty promising, except for how darn long > it takes to do a complete purify or valgrind run. It would take > about ten machine-days to do a complete run with a debug build, > and about half that with a release buld. To get to a one-hour > cycle time with a release build would take about 120 bots. > - Dan > > > > --~--~---------~--~----~------------~-------~--~----~ Chromium Developers mailing list: chromium-dev@googlegroups.com View archives, change email options, or unsubscribe: http://groups.google.com/group/chromium-dev -~----------~----~----~----~------~----~------~--~---
