| Summary: | [chromium] Updating chromium DEPS | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Greg Simon <gregsimon> | ||||
| Component: | New Bugs | Assignee: | Nobody <webkit-unassigned> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | commit-queue, ukai, vitalyr | ||||
| Priority: | P2 | ||||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Attachments: |
|
||||||
|
Description
Greg Simon
2011-05-20 22:00:32 PDT
Created attachment 94318 [details]
Patch
Rolling chromium DEPS to pick up the ScopedTempDir export from webkit_support so it can be used in chromium LayoutTests. Comment on attachment 94318 [details]
Patch
r=me
but we probably dont want code in the webkit repository to depend on ScopedTempDir.
it would make it very painful if someone ever changed that class in an incompatible
way. how would the average chromium hacker know to fixup webkit? and if they did
know they would then have to deal with the circular dependency. (hard to roll
webkit w/o things being temporarily broken .)
the contents of webkit_support.h mostly encapsulates the extent of the circular deps
which we have. if you dont want to copy code into webkit, then consider exposing
the "algorithm" of interest as new functions in webkit_support.h.
The current application is for a LayoutTest so the temp dir does not get filled up. There are not enough primitives in platform/FileSystem.h to do this sort of work entirely in WebCore. However, now that this API is set up in webkit_support its impl can be changed (only in chrome) so it does not use ScopedTempDir to avoid the problems you mention. Comment on attachment 94318 [details] Patch Clearing flags on attachment: 94318 Committed r87050: <http://trac.webkit.org/changeset/87050> All reviewed patches have been landed. Closing bug. |