RE: One character per brigade?

[Tess Flynn <tf...@mqsoftware.com>]

Thanks to everyone that helped me with this. It works now!

-Tess

-----Original Message-----
From: Nick Kew [mailto:nick@webthing.com]
Sent: Wednesday, November 29, 2006 4:14 PM
To: modules-dev@httpd.apache.org
Subject: Re: One character per brigade?


On Wed, 29 Nov 2006 16:04:21 -0600
"Tess Flynn" <tf...@mqsoftware.com> wrote:

> > If you're just examining the data but not modifying it, you don't
> > want to mess about with new brigades.
> 
> I was pointed in that direction by this post:
> http://marc.theaimsgroup.com/?l=apache-modules&m=108029424904988&w=2
> 
> I was using the new brigade only as a buffer.

Your original post didn't make it clear (to me) you had any good
reason to buffer the data.

>	 I could have done this
> as a memory block within my ctx, but it didn't seem as efficient
> given I didn't know how many buckets the request was broken across.
> After I reached the end of the input, my plan was to run
> apr_brigade_flatten() against the brigade buffer to get the final
> string.

If you need the complete data as a string, that's right.  If you
can avoid buffering the data, it's better.

>	 I need to run an XPath against the content of both the
> request and the reply, so I can't have an incomplete document.

You don't run an XPath on a string; you run it on a DOM tree.
Why not feed the incoming data directly into your XML parser
rather than buffer a string?  See for example mod_transform.

> > Take a look at mod_diagnostics, which reports on exactly the data
> > you're looking for, and might be a good startingpoint for your
> > module.
> 
> Maybe I'm missing the point, but I don't quite see how
> mod_diagnostics will help me. It doesn't seem to gather the entire
> request or reply, but operates on each bucket individually.

That suggestion was for your task as it appeared before you
mentioned buffering and XPath.

-- 
Nick Kew

Application Development with Apache - the Apache Modules Book
http://www.apachetutor.org/