[Webkit-unassigned] [Bug 107655] [v8] prepare SerializedScriptValue for transition to Latin-1

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=107655





--- Comment #6 from Joshua Bell <jsbell at chromium.org>  2013-01-23 09:29:47 PST ---
(In reply to comment #5)
> jsbell, alecflett: just to clarify what will happen.  Once V8_ONE_BYTE_STRINGS_ENABLED is defined in v8, one byte strings will now potentially have payloads with bytes > 0xf.  This was previously impossible, so a version 2 SSV should not be allowed to read from a version 3 value.  Version 3 can read version 2 without issue.

Thanks. Conceptually, this looks correct.

There is a problem, though - this conceptually *removes* test coverage for parsing v2 SSVs, since with this patch and once the new flag is defined, the test will only exercise parsing v3 SSVs.

You'll need to add tests that use the (optional) oldFormat param of _testSerialization and explicitly test 0x02ff, or expand testSerialization() to insert the version prefix since most v2 and v3 SSVs will serialize identically. I'm not sure which is better. Or perhaps oldFormat should turn into a map of version/serializations? Not sure what is best here.

(alecflett's not going to be able to look at this until tomorrow, unfortunately.)

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.