On Monday 26 November 2001 12:22 am, you wrote:
> On Sun, Nov 25, 2001 at 11:11:09PM -0500, Jason Boxman wrote:
>
>
> [...]
>
> > First, *HTTPD wants to tack on a duplicate protocol and version to each
> > outgoing stream:
>
> [...]
>
> Sorry about that.  I've committed a fix for it to CVS, and it'll be in
> the next release.

Not a problem as it was easy to work around. :)

> The next release is waiting for me to rewrite testomatic so I can
> resume the automated cross-platform tests.
>
> > Second, as you'll note above you cannot send a response header to a
> > browser without content, because the trailing network lines indicate to
> > the browser that the content will follow.
> >
> > In my application, I'm using PoCoCi::UserAgent and LWP::UserAgent's
> > callback for data chunks which means I'll be passing data back later on. 
> > Forcing the browser to think content follows the header as above thwarts
> > this.
>
> I don't understand this part.  Artur Bergman, the author of
> Filter::HTTPD, wrote a simple streaming mp3 server that sends headers
> first and then chunks of content later.  Here's the part that answers
> requests:

Yeah, I didn't either.  I was passing data back incorrectly to SysRw and that 
was messing things up.

<snip>
> The newline doesn't seem to hurt this application, but the remote
> client is an mp3 player.  Does it really break browsers?

Yeah, you're right.  I was just trying to grasp at something that could 
explain the weird behavior I was experiencing.

Now my callback for PoCo::Client::UserAgent by way of LWP::UserAgent is only 
geting called once, even for files larger than my specified blocksize, which 
is giving me more wicked results...  *sigh*

> -- Rocco Caputo / [EMAIL PROTECTED] / poe.perl.org / poe.sourceforge.net

Reply via email to