Summary: | clear databases before running read-transactions-running-concurrently test | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | jochen | ||||
Component: | New Bugs | Assignee: | jochen | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | thakis, webkit.review.bot | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
jochen
2013-02-18 11:50:37 PST
Created attachment 188926 [details]
Patch
Comment on attachment 188926 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=188926&action=review > LayoutTests/ChangeLog:9 > + of databases from previous tests are hanging around. Shouldn't the test runner do this before running a test? (In reply to comment #2) > (From update of attachment 188926 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=188926&action=review > > > LayoutTests/ChangeLog:9 > > + of databases from previous tests are hanging around. > > Shouldn't the test runner do this before running a test? it's a rather expensive operation, as it hits the filesystem. (In reply to comment #3) > (In reply to comment #2) > > (From update of attachment 188926 [details] [details]) > > View in context: https://bugs.webkit.org/attachment.cgi?id=188926&action=review > > > > > LayoutTests/ChangeLog:9 > > > + of databases from previous tests are hanging around. > > > > Shouldn't the test runner do this before running a test? > > it's a rather expensive operation, as it hits the filesystem. How often? A single stat is relatively cheap since every test stats at least the html file and often a js and a css file. And I'd guess in most cases there's nothing to clean up. (stating a .d file in ninja takes ~6µs on a medium-warm cache.) (In reply to comment #4) > (In reply to comment #3) > > (In reply to comment #2) > > > (From update of attachment 188926 [details] [details] [details]) > > > View in context: https://bugs.webkit.org/attachment.cgi?id=188926&action=review > > > > > > > LayoutTests/ChangeLog:9 > > > > + of databases from previous tests are hanging around. > > > > > > Shouldn't the test runner do this before running a test? > > > > it's a rather expensive operation, as it hits the filesystem. > > How often? A single stat is relatively cheap since every test stats at least the html file and often a js and a css file. And I'd guess in most cases there's nothing to clean up. > > (stating a .d file in ninja takes ~6µs on a medium-warm cache.) dunno, is that also true on e.g. android (or content shell, where we have to do this in another process)? Anyway, this seems to be the common pattern, e.g. cookies tests also clean up after themselves, instead of DRT resetting the cookie store after every test (although you could argue that this would make sense as well) Comment on attachment 188926 [details]
Patch
ok ok
Comment on attachment 188926 [details] Patch Clearing flags on attachment: 188926 Committed r143261: <http://trac.webkit.org/changeset/143261> All reviewed patches have been landed. Closing bug. |