Summary: | REGRESSION (HTML5 tree builder): Text selection in a large text document is extremely slow | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Alexey Proskuryakov <ap> | ||||
Component: | DOM | Assignee: | Alexey Proskuryakov <ap> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | abarth, aestes, bzbarsky, commit-queue, eric, hsivonen, ian, koivisto, mjs | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | All | ||||||
OS: | All | ||||||
URL: | http://dscoder.com/test.txt | ||||||
Bug Depends on: | |||||||
Bug Blocks: | 41115 | ||||||
Attachments: |
|
Description
Alexey Proskuryakov
2011-03-07 13:08:45 PST
I gave ap some pointers on IRC to where in the code needs to be changed. Eric seems to think he did this already, but I can't find any evidence of that. For what it's worth, the most common compat issue we had with the old behavior in Gecko was sites using XHR and then extracting the text from a node in the responseXML. They _could_ use .textContent, but tended to use .firstChild.data, and this would break in random ways depending on the data... We did have to fix a few of the resulting performance issues in Gecko 2, yes. Wasn't that big a deal. I'll dig around. I swear I implemented text node splitting when doing the new parser. I have a patch that I'm testing now. Created attachment 84996 [details]
proposed fix
Even if we improve text selection performance, add a quirk for Mail, and decide to not split the nodes, this code will still be needed for the quirk.
Comment on attachment 84996 [details]
proposed fix
Looks good.
Comment on attachment 84996 [details]
proposed fix
OK.
Comment on attachment 84996 [details] proposed fix Clearing flags on attachment: 84996 Committed r80526: <http://trac.webkit.org/changeset/80526> All reviewed patches have been landed. Closing bug. |