Summary: | [EFL][GTK] Fix build break since r185262 | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Gyuyoung Kim <gyuyoung.kim> | ||||||||||||
Component: | WebKit EFL | Assignee: | Gyuyoung Kim <gyuyoung.kim> | ||||||||||||
Status: | RESOLVED FIXED | ||||||||||||||
Severity: | Normal | CC: | cgarcia, commit-queue, lucas.de.marchi, mcatanzaro | ||||||||||||
Priority: | P2 | ||||||||||||||
Version: | 528+ (Nightly build) | ||||||||||||||
Hardware: | Unspecified | ||||||||||||||
OS: | Unspecified | ||||||||||||||
Bug Depends on: | 145711 | ||||||||||||||
Bug Blocks: | |||||||||||||||
Attachments: |
|
Description
Gyuyoung Kim
2015-06-06 01:46:39 PDT
Created attachment 254410 [details]
Patch
Created attachment 254411 [details]
Patch
I'm not sure if this fix is correct. Anyone give me comment ? Can we just get rid of those legacy directory functions? It's not clear to me why we need two different places for these. Created attachment 254421 [details]
Patch
Created attachment 254422 [details]
Patch
(In reply to comment #4) > Can we just get rid of those legacy directory functions? It's not clear to > me why we need two different places for these. I'm not sure what is correct fix now. However I want to fix EFL and GTK as soon as possible. I'd like to land this patch, then how about fix correctly for each port ? Comment on attachment 254422 [details]
Patch
To make EFL and GTK bot to green for now, I land this patch though this may not correct fix. If there is wrong fix, please fix it on following patch.
Created attachment 254432 [details]
Patch
Comment on attachment 254432 [details] Patch Clearing flags on attachment: 254432 Committed r185300: <http://trac.webkit.org/changeset/185300> All reviewed patches have been landed. Closing bug. Yes, this was indeed wrong, at least for GTK+ port that has API to se the disk cache directory. There are actually two problems here. One is that we need to override the disk cache directory after the WebProcessPool has been created, and the other problem is that disk cache directory has been removed from web process pool configuration. We need to bring it back while we have network support in the web process. (In reply to comment #12) > Yes, this was indeed wrong, at least for GTK+ port that has API to se the > disk cache directory. There are actually two problems here. One is that we > need to override the disk cache directory after the WebProcessPool has been > created, and the other problem is that disk cache directory has been removed > from web process pool configuration. We need to bring it back while we have > network support in the web process. Thank you for your explanation. I will also follow this problem for EFL port too. Anyway now EFL and GTK bots are fine. (In reply to comment #13) > (In reply to comment #12) > > Yes, this was indeed wrong, at least for GTK+ port that has API to se the > > disk cache directory. There are actually two problems here. One is that we > > need to override the disk cache directory after the WebProcessPool has been > > created, and the other problem is that disk cache directory has been removed > > from web process pool configuration. We need to bring it back while we have > > network support in the web process. > > Thank you for your explanation. I will also follow this problem for EFL port > too. Anyway now EFL and GTK bots are fine. Oh, right, I guess you need to append "webkit" to the disk cache directory for EFL in the UI process too. |