discuss-gnustep
[Top][All Lists]
Advanced

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

Re: API for next stable release of base library


From: Richard Frith-Macdonald
Subject: Re: API for next stable release of base library
Date: Wed, 18 Oct 2006 14:11:46 +0100


On 18 Oct 2006, at 13:04, Marko Riedel wrote:


Hello there,

I hope you do not plan to remove your web proxy code from NSURL*, as that
will break Ticker.app.

What's your take on this?

Judging by the weblogs on GNUstep.it, Ticker.app seems to do its share of
drawing people to GNUstep.

Best regards,

I use Ticker.app all the time ... it's a great little utility.

Specifically, my take on the proxy code in NSURL is that the existing mechanism should probably go away some day ...
but not soon, and not before we have a replacement mechanism.

My reasoning for this is that the existing proxy support was written as a necessary add-on before Apple/MacOS-X had any proxy support at all, but now Apple has a completely rewritten and changed version of the URL handling classes, and their new implementation does provide proxy support.

Since we want to maintain MacOS-X compatibility, we need to implement their new URL system ... and there is a task registered in the GNUstep project in savannah to do this implementation, waiting for volunteers (I've done some of the work but it's far from complete).

Once that is done, it would seem to make sense to migrate existing applications over to use the new system ... so apps would be more portable.

On a more general point, I certainly do not 'plan' to remove it (except in the very vague sense mentioned above), and in fact do not 'plan' to remove any public API specifically at this point (apart perhaps from the rather trivial GSFindFiles()). What I was hoping to achieve with the email starting this thread was to establish a feel for what functionality people use or don't use and an idea of how(/if) we want to clean up the API.

I think a guiding principles should be to consider removing extensions which are trivially implemented (eg less than one screen (25 lines) of simple code) using standard API, and certainly to avoid adding any such extensions, but I'd like to know what other people think. I'd also like to know about ideas for reorganising poorly designed API and/or adding methods which complement existing functionality and are not easy for a developer to add in their own applications.






reply via email to

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