Summary: | Regression(r217867): Legacy SessionHistoryEntryData format should not change | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Chris Dumez <cdumez> | ||||
Component: | WebKit2 | Assignee: | Chris Dumez <cdumez> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | andersca, beidson, ggaren, sam, simon.fraser, webkit-bug-importer | ||||
Priority: | P2 | Keywords: | InRadar | ||||
Version: | WebKit Nightly Build | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Bug Depends on: | |||||||
Bug Blocks: | 173033 | ||||||
Attachments: |
|
Description
Chris Dumez
2017-06-12 10:51:35 PDT
Created attachment 312678 [details]
Patch
Comment on attachment 312678 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=312678&action=review > Source/WebKit2/ChangeLog:10 > + as this apparently breaks compatiblity. Would be nice to explain a little more why. Comment on attachment 312678 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=312678&action=review >> Source/WebKit2/ChangeLog:10 >> + as this apparently breaks compatiblity. > > Would be nice to explain a little more why. I cc'd Sam and Anders to see if they can educate me. What I saw is that the legacy session state encoding / decoding code has quite a few "Encode / decode unused X field here to maintain format compatibility". So clearly, keeping the exact same format is important. Note that there is versioning but I tried bumping the version number and it did not help. My assumption is that the client (Safari?) relies on our current format somehow. Comment on attachment 312678 [details] Patch Clearing flags on attachment: 312678 Committed r218121: <http://trac.webkit.org/changeset/218121> All reviewed patches have been landed. Closing bug. |