a long shot -> sending output to a browser question

[Andreas Kostyrka]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Montag, 3. Dezember 2001 10:25 schrieb Oleg Broytmann:
> On Sun, Dec 02, 2001 at 09:27:20PM -0600, Cameron Laird wrote:
> > "Multipart" is another answer.  Netscape browsers
> > (but no others, ever?) arguably conformed to standards
>
>    It is called "server push", and many browsers support it. Even M$IE had
> support for server push, but lost it not so long ago.
Well, basically I've discovered following classes (the association is out of 
my memory, meaning check it):
VERY BAD:
  Basically this crashes or buffers to much of the page (to make it useful 
for my chat client). Mac browsers tend to this, newer IEs seem also to buffer 
the page. These clients you should catch and use a "refresh" technique.
BAD:
  Cases where a browser that should be able to connect and stream text/html 
refuses to: Proxies, transparent proxies, firewalls, etc. on the way to the 
server. These usually can be caught by using a timeout for the stream. If the 
browser doesn't execute some Javascript in time, one basically forces a 
"refresh" mode.
GOOD:
  All is ok. Still some gotchya's: Depending upon browser you cannot have to 
many streams open, additionally the "spinning icon" of the browser usually 
doesn't stop spinning, which might confuse the user.

So it boils down to one trivial thing:
- -) Test it with all kinds of acceptable browsers.
- -) Provide a way to force "refresh" replacement mode.

Andreas
- -- 
Andreas Kostyrka; Raiffeisenstr. 16/9; 2320 Zwölfaxing
Tel: +43/676/4091256; Fax: +43/1/7065299
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE8C0ulHJdudm4KnO0RAmohAKChP8fNOFt7iLPV6+AGKa9MsguD8QCfQ4pZ
66xLsuuqO8j1RIbda5To/ME=
=3VU4
-----END PGP SIGNATURE-----