On Fri, 31 Mar 2000, Klaus Weide wrote:
> On Fri, 31 Mar 2000, Vlad Harchev wrote:
> > On Fri, 31 Mar 2000, Klaus Weide wrote:
> > > I don't think I have seen this, so please give a concrete example.
> >
> > Enable source cache. Go to www.yandex.ru (search engine). Press '\' - new
> > page will be fetched. Press '\' again - new page will be fetched again. I
> > assume most commercial sites are good testsuit for it.
>
> I tried, with my current code (still based on dev.18). It doesn't
> act as you describe, it gets a copy from the source cache as expected.
> That's the intended behavior (maybe not the best one, but I didn't
> change it).
>
> But, your example doesn't really demonstrate what you think it does -
> the headers I get are like this
>
> HTTP/1.1 200 OK
> Date: Fri, 31 Mar 2000 16:13:59 GMT
> Server: Apache/1.3.6 (Unix) mod_perl/1.19 rus/PL28.15
> Cache-Control: max-age=3600
> Expires: Fri, 31 Mar 2000 17:13:59 GMT
> Last-Modified: Thu, 30 Mar 2000 13:06:23 GMT
> Connection: close
> Content-Type: text/html; charset=windows-1251
>
> In other words, the expiration time is one hour in the future.
> And the '=' page doesn't show any 'no-cache', either.
> Maybe _your_ lynx comes to a different conclusion because your
> timezone is set up the wrong way...
>
> But even with a document that does have a 'no-cache' recognized
> by lynx, I still don't get what you describe.
>
> And I don't see how I possibly could - there is no code in the
> HTreparse_document path to check for it. (Not in my code, and
> I doubt it has been added.)
Could you please try dev23 to see what I described (i.e. document is
refetched each time you press '\')? I was playing with lynx dev23 I got in bz2
format from lynx.isc.org (i.e. it's the most clean lynx you can get) - there
are no modifications by me in it.
I spent a couple of hours with gdb attaching to running lynx and inspecting
- LYSourceCache is indeed 2 (SOURCE_CACHE_MEMORY) in HTreparse_document, but
source_cache_chunk is always NULL - so it reloads doc from the net (while
there are pages like www.samara.ru - that seem don't have 'expires' header
or pragma - pressing '\' on them will get document from the cache).
I'm very sorry if it's somebody else's changes that cause such behaviour
(I probably mistakenly conclued that they are yours) - but let's fix it
anyway.
> Klaus
>
Best regards,
-Vlad
