Re: Data --> object store --> client

On Fri, 2002-08-09 at 23:54, Adrian Chadd wrote:
> On Fri, Aug 09, 2002, Joey Coco wrote:
> > > I'm going to address 1) now, 2) is much harder (although some work is
> > > underway now that will make developing 2) patchs much easier in the
> > > future).
> > >
> >
> > It's #2.. #1 we got figured out. What we want to do is if the content
> > matches a certain "criteria" either blank out the content, or display a
> > squid access denied page..
>
> Ah. yes. Well, if you wanna do html filtering, you'll need to
> code your filter in such a way that if you discover a partial match
> at the end of a buffer, to keep the buffer around somewhere until
> the next buffer comes along so you can do a 'complete match'.

Huh? There's no need to do that, any decent stream parser will not need
buffered data to complete matches (well, with the single exception of
regex's, and I suspect that with some of the *good* regex libraries
around today, even that is possible).

The only data that needs buffering is the current partial match - and
only because it can't be sent until it's known to be a miss and
therefore allowed
 
> You _COULD_ also, with some of the stuff that'll go into squid-HEAD
> soonish, write a filter module that buffers all text/html content
> until the complete amount has been read in, and _THEN_ send it to
> the store/client, without affecting other stuff.

Won't that still cause timeout issues for large documents? I didn't
think we supported CONTINUE or whatever it is yet.

Rob