discuss-gnustep
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ANNOUNCE: skyrix-core-42 available


From: Richard Frith-Macdonald
Subject: Re: ANNOUNCE: skyrix-core-42 available
Date: Sun, 6 Apr 2003 11:00:16 +0100


On Sunday, April 6, 2003, at 08:44  am, Helge Hess wrote:

Richard Frith-Macdonald wrote:
Why do you need MIME for NSURL ?
Well, strictly speaking it was for HTTP rather than MIME, but they are so similar that I
implemented it as the GSMime class so it could be used generally too.

Hm.

Sounds a bit "over-engineered" to me ;-) HTTP is far simpler than MIME (more or less a small subset) even if this might not be apparent if you look at it. HTTP (as available through the NSURL API) has no duplicate headers, to line folding, a single charset, no line length limitations, no multiparts, ...

Sure ... That's because I wanted MIME support too ... and saw no reason to duplicate the code.

Why doing more than required in the very core library ? This way not only the code base is more complicated, but applications developed against GNUstep are not unlikely to be incompatible to other Foundations (pretty much the same way Cocoa apps are because of CF).

Only if they use the extensions ... and if they are using them it's because they want the capabilities, and would need to duplicate the functionality in some way if they wanted to port anyway. I know we have a basic difference of philosophy on this. I do not feel that people using the GSMime (or GXML) classes do so without knowing that they are extensions and get 'trapped' into non-portability without realising (and in any case they can use these things with other foundations if they want to). Actually, for compatibility it would probably be nice to write C wrappers for GSXML to emulate the CF xml parsing capabilites :-) If I wanted an XML based app to work on both GNUstep and MacOS-X I'd do that.

I suspect Pantomime didn't re-use it, because it had not been ported to MacOS-X

In my understanding Pantomime's primary goal is to implement the javax.mail API (which came out long after we started NGMime/NGImap4). IMHO this is a good choice (I also always wanted to make things more javax.mail like).

Not too bad, but like most Java APIs it is far too big. I prefer a smaller selection of more flexible methods and classes.

A MIME library cannot be implemented in a single class, MIME is a very complex thing in practice.

Yes ... my sloppiness in writing the last email ...GSMime is a small group of classes.





reply via email to

[Prev in Thread] Current Thread [Next in Thread]