Re: [patch] refresh bugs

tis 2006-05-02 klockan 11:08 +1200 skrev Doug Dixon:

> Not sure I follow... unless I'm mistaken, I think this fixes Bug #7

Bug #7 is about updating HTTP headers on a 304. This is mainly

  Date
  Expires
  Cache-Control

but there may also be other headers which needs to be refreshed with
content from the new response.

> I guess the question is whether the call to storeTimestampsSet
> (old_entry) sets the timestamp correctly or not in this instance, and
> therefore whether we need to overwrite this immediately afterwards
> (as happens at present).

Currently the timestamps are never set 100% correct according to specs.
And it's worse on refreshed objects as then only the internal timestamp
is updated, not the HTTP headers..

Some of the most apparent bad effects of this can be seen in cache
hierarchies, where child caches sometimes always consider objects stale
due to old Date & Expires headers. It's also seen by changes in
Cache-Control never becoming effective at all until a forced reload of
the whole object.

Regards
Henrik