On Thu, 22 Apr 2004, [ISO-8859-2] Micha� Matusiak wrote:
> In handleIMSGiveClientNewEntry() control flows to processMiss()
> (this is done because of httpReplyValidatorsMatch()).
And is what it should not.
The httpReplyValidatorsMatch() function simply does not work on a 304
reply, and even on a full reply I am very doubtful it should be used like
it is here.
> In my opinion this is a bad behaviour.
And I fully agree.
> let's assume that server responds with 304 and there is no
> Last-Modified header in server reply (for example apache doesn't
> send Last-Modified header with 304 response).
> Squid considers new response headers as 'inconsistent' with his old
> ones. Clearly we can't say this is wrong response.
Yes.
> We need to know what is the intended semantics of
> httpReplyValidatorsMatch()?
Agreed. I don't understand why this function is used like it is here.
> This function has also bug (in our opinion) which seems to be easy
> to correct:
> ...
> one = httpHeaderGetStrOrList(&rep->header, HDR_CONTENT_MD5);
> ...
> if (strcasecmp(one.buf(), two.buf())) {
> ...
> }
> ...
> strcasecmp() seems cause to SEG FAULT when one of the parameters is
> NULL. This is the case.
I doubt this function ever gets this far.
> Could you give me access to CVS? My account name is mmatusiak.
Done.
Now please read
http://devel.squid-cache.org/howto.html
and the documents linked from there.
http://devel.squid-cache.org/rules.html
http://devel.squid-cache.org/CVS.html
Regards
Henrik
