> libcurl already separates the body from the header so I don't
> understand why
> you would do this as well.

Basically,
- beacuse in non-curl mode I save http headers in my client object, and wanted to have consistent behaviour when curl is enabled
- because looking at http headers I can check for proper content-type, content-lenght, charset encoding

if there is a better way in curl-php to have curl feed the http headers to my app, it would be a welcome enhancement...

> And if you do it yourself anyway, I suggest you
> read RFC2616 on how to do it and not just assume that proxies
> will use specific strings such as "Connection established".

Well, duh. I had read both RFC 1945 and 2616 up to now, and nowhere found mentioned this particular "behaviour", e.g. getting an HTTP/1.0 200 response header followed by two CRLF followed by the real HTTP/1.0 response.

Since the only spec I could find for this particular case is the one "luotonen 01" you also mention on http://curl.haxx.se/rfc/, where this particular 'Reason-Phrase' is mentioned and I have precious few https tunnelling proxies to test against, I just tought I might ask to the powers-that-be (of http ;)

Afaict looking for a particular reason-phrase is dead wrong, but then how am I supposed to know I should expect two seaparate sets of HTTP 200 status lines?
Maybe I would be better off only/always doing this check when using proxy AND https?

Bye
Gaetano

_______________________________________________
http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-and-php
Received on 2006-09-05