Re: lynx-dev PATCH [dev22]: source caching fixes (was: patch to fix PSRC mode with SOURCE_CACHE!=NONE)

[Vlad Harchev]

On Thu, 15 Apr 1999, Leonid Pauzner wrote:

> 14-Apr-99 13:32 Vlad Harchev wrote:
> > On Wed, 14 Apr 1999, Leonid Pauzner wrote:
> 
> >> Probably the exit from options menu may now utilize the power of 
> >> SOURCE_CACHE,
> >> though it handles more parameters than HTdocument_settings_changed()
>                                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >>
> >> Leonid.
> >>
> >    I found another bug: if there is something on the cache stack (say doc A
> > -current doc and doc B), and we go to 'O' and change something, only
> > appearance of 'A' is changed, the appearance of B is not changed - the 
> > cahced
> > version is used. IMO it will be better to uncache everything in the stack.
> > This is best tested with 'links are/not numbered' - since rendered versions
> > differ a lot.
> 
> That was a very expensive solution before SOURCE_CACHE
> but now this can be done nicely by adding more parameters to the function
> above and avoid using of HTuncache_current_document() calls.
> Until we keep a compatibility with SOURCE_CACHE:NONE this is a little messy.

  But the way it's implemented now should be considered incorrect. IMO, user
should understand that changing options have a penalty and not to do this
very often. Also, IMO the use of caching proxies with SOURCE_CACHE:NONE won't 
make things messy/slow. IMO, we should implement the correct behaviour.

> 
> >  If the bug I described is fixed, then all documents will return to their
> > normal (not source) representnation (or we have to keep a stack of modes of
>                                                      ^^^^^^^^^^^^^^^^^^^^
> We already do that somehow in LYK_SOURCE - we trying to remember document
> owner and document charset to utilize them in source mode (parameters gotten
> for rendering HTML will unavailable in source mode until we store them
> somehow).  Probably, a proper way will be to split out the HTParentAnchor
> data and HText data so the first will not be lost when we uncache the second.
> And source mode flag may now belongs to HTParentAnchor. So the mode will
> always WWW_PRESENT until the *document* has its own value overwise.
> 

  Yes, I think it can be done that way.

> > the documents somehow - or probably set the pointer to HText to NULL in the
> > cache, but marking whether it was 'source' view or not)
> 

 Best regards,
  -Vlad