On Wed, 24 Nov 2010, Michael Schnell wrote:
On 11/24/2010 11:42 AM, michael.vancann...@wisa.be wrote:
I tried making moderatly complicated production applications with it. It
cannot be done without redesigning ExtPascal from the ground up.
Same here, That is why I only recommended it as a way to try out ExtJS.
Comet is a rickety technology. Not robust at all, and creates more problems
than it solves. You can use polling, but that's just plain HTTP, not even
worth of putting a name on it.
I suppose a problem of Comet's is that it needs cooperation of the web
server, allowing for an appropriate number of simultaneous "hanging"
requests.
Not only that, but also switches, routers, firewalls and proxies.
In fact the necessity of using such questionable technology as a
workaround for the missing server event transport is a design flow of the
abysmal http protocol which never was intended for but needs to be abused for
this kind of applications due to the incompetence of most IT departments that
are not able to decently configure the firewalls.
I think any technology which relies on a persistent connection is bound to
fail. Not because the concept is inherently bad, but because in practice, a
connection is rarely persistent.
We've had this discussion before.
I wonder what technology Silverlight/Moonlight uses for this purpose. Could
this be a hint for an upcoming "HTML" Widget Type ?
I doubt that any technology which relies on browser plugins will make it.
HTTP/Html/Javascript are the only reliable standards (and even then) worth
working on.
All the rest is risky and thus - in my book - not even worth considering.
(not to mention the vendor lock-in they cause).
Michael.
--
_______________________________________________
Lazarus mailing list
Lazarus@lists.lazarus.freepascal.org
http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
