Hi,
Many valid things have been said from both sides regarding WebClient.
As far as I understand it, there are 2 possible 'places' where code can reside: inside Pharo(Core) or outside. The inside code does not have to be compatible with other Squeak/Smalltalks (apart from maybe ANSI and other standards, I am not qualified here). The outside code can choose its compatibility target. Seaside goes very far here, but it hurts to think about the amount of effort that is being put in to achieve this (if it wasn't for the proof of its existence and portability, I would not believe this was even possible) (*) Some other packages might target just one Smalltalk. We cannot demand this from any other, even if it would be a good idea.
My question is, where does a (future) HTTP Client fit in ?
Both places are possible, one requiring more effort than the other.
If I look at what is inside PharoCore, there is an important dependency on HTTP client access at least (Monticello, Updating, Gofer, ...). From that standpoint such a client should be 'inside'.
Furthermore, many people have indicated that a good HTTP client (and/or server) is (are) very important today. Any Smalltalk looks silly without a proper HTTP implementation.
Even though an 'outside' client (based on Grease for example) would be doable and a good idea, the 'inside' route feels more natural.
BTW, In Java both exist (it is in the standard J2SE libs, and there are many clients, like
http://hc.apache.org/ ).
What do others think ?
Sven
(*) I respect this very much, it is a remarkable achievement.
_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Locked
