On 7 January 2011 20:07, Sven Van Caekenberghe <[email protected]> wrote: > I was in meeting all day, hence the late reply. > > On 07 Jan 2011, at 14:08, Stéphane Ducasse wrote: > >> Sven >> >> two questions: >> - if we want to use zn for pharo to replace HTTPSocket, don't you >> think that Zinc-HTTP-Client-Server should be split into pieces. > > I don't think that would make a lot of sense: HTTP is a client-server > protocol, the client and server sides have a lot in common; a split in three > pieces would be possible but each piece would be rather small. Having the > necessary pieces for a simple HTTP server present will lower the bar for > others, so that makes sense too. (For example, the Seaside adaptor is really > very little code, no more need for Kom unless you really want it). > >> - do you think that some parts of zinc could be reused to implement >> other protocols such as POP, SMTP, IMAP? > > Maybe, finding shared code can only be done by doing both implementations I'm > afraid ;-) > > Another, related point is the following: maybe you noticed browsing the code > that I have a number of classes in Zn that redo some standard functionality > (ZnMimeType, ZnUrl, ZnCharacterEncoder, and some of the code in the utils > classes), a bit like Seaside does. The reason is that the existing classes > were broken beyond repair (IMHO), or that I was not confortable extending > them. Now, if the existing classes could be improved/replaced, it would make > sense to replace the Zn variant with a good system wide one. This is a search > for the best solution. >
well it would be good to have these classes fixed, and make zinc using them. because these ones are needed thoughout entire system, and certainty, for things like mime-type , there could be code which can handle certain type(s), but know nothing about Zinc. > Sven > -- Best regards, Igor Stasenko AKA sig.
