kdelibs/khtml/xml

[Tobias Anton]

On Monday 12 August 2002 13:07, Tobias Anton wrote:
> On Monday 12 August 2002 12:40, Tobias Anton wrote:
> > On Thursday 08 August 2002 22:20, Waldo Bastian wrote:
> > > On Thursday 08 August 2002 07:10 am, Tobias Anton wrote:
> > > > On Wednesday 07 August 2002 19:03, Waldo Bastian wrote:
> > > > > On Wednesday 07 August 2002 05:57 am, Tobias Anton wrote:
> > > > > > The correct approach would be to pass the creation time
> > > > > > of the cache entry up to KHTML. This would also solve some other
> > > > > > glitches, i.e. we could address the problem that images in the
> > > > > > loader cache don't expire at all. (at least if this has not been
> > > > > > changed in the loader)
> > > > >
> > > > > The loader cache _does_ expire images.
> > > >
> > > > You're right; that was new to me.
> > > > But still, it doesn't pick up the expire dates from the metatags.
> > > > In order to correctly calculate the moment when to re-request an
> > > > image, KHTML must know about the creation date of the cache entry
> > > > fille, because expiredate = creationtime + expiretimeout.
> > > >
> > > > That's why I have made a small patch that passes the cache entry file
> > > > creation time in the metadata.
> > > >
> > > > Would you have a look at it?
> > >
> > > I have commited code for the http slave already.
> > >
> > > Can you test whether the attached khtml patch does what it should?
> >
> > I tested a bit and found out that docLoader->setExpireDate() is called
> > with a relative expire date while the expire date is zero.
s/expire date is zero/creation date is zero/

> > This leads to wrong absolute expire dates both in the document and in the
> > embedded images.
>
> Oh silly me!
> I failed to realize that my freshly compiled kio_http was not in use yet.
> I need to recompile & install some more stuff to do further testing. You'll
> hear from me after that.
Your two patches together do not work when a page is loaded for the first 
time. The "cache-creation-date" metadata is not set if the respective cache 
entry is just about to be created (m_bCachedWrite being true in the slave).
Shouldn't we add the "cache-creation-date"-metadata for that case, too?

Tobias
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kio-creationdate2.diff
Type: text/x-diff
Size: 606 bytes
Desc: not available
URL: <https://mail.kde.org/mailman/private/kfm-devel/attachments/20020814/8bc96656/attachment.diff>