[Webkit-unassigned] [Bug 85028] [BlackBerry] Cookies should be checked during parsing to improve performance.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=85028





--- Comment #5 from Jason <jason.liu at torchmobile.com.cn>  2012-05-06 18:52:40 PST ---
(In reply to comment #3)
> The approach looks good.
> 
> size_t dotPosition = parsedValue.find(".", 1);
> 
> That looks wrong, though. The original code skipped this check if the domain did not start with a dot.  So originally we had (where x is any character that isn't a dot):
> 
> .xxxxx - accepted
> .xx.xx - accepted
> xxxxxx - REJECTED
> xxx.xx - accepted
> 
> This is the same as searching for the dot from position 0, not 1.

We check if the domain contains an embedded dot. 
And originally we had:
 .xxxxx - rejected   not accepted
 .xx.xx - accepted
 xxxxxx - REJECTED
 xxx.xx - accepted
So this is the same as searching for the dot from 1, not 0.

> Also we have lost this check:
> 
> 162  // ignore domain security if protocol doesn't have domains
> 163  if (shouldIgnoreDomain(cookie->protocol()))
> 164      return false;
> 165
If the protocol doesn't have domains, the code path won't contain domain parsing. So drop this "if" check in parsing.

> 166  // Reject Cookie if domain is empty
> 167  if (!cookie->domain().length())
> 168      return true; 
> 

size_t dotPosition = realDomain.find(".", 1); This is including length check.

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