Re: Kickstarter was not successful... but it did help things...

[Dr . H . Nikolaus Schaller]

Am 12.09.2013 um 18:30 schrieb Doug Simons:

> On Sep 12, 2013, at 4:03 AM, David Chisnall wrote:
> 
>> We could almost certainly provide them with an automated tool that they can 
>> run on their codebase that would give them a pretty clear idea of the OS X 
>> APIs that they use.  Actually, providing such a tool with the ability to 
>> produce a report against the current version of GNUstep showing what is 
>> missing would be very helpful for a lot of projects.
> 
> If such a tool could be produced, I think it would be a tremendous asset for 
> GNUstep. I suspect there are plenty of developers of applications for OS X 
> who would be interested in porting to GNUstep but are put off by the daunting 
> task involved in making the effort, especially knowing that there are almost 
> certainly some things missing but not knowing how extensive those gaps might 
> be.
> 
> If there was an easy way for a developer to analyze their code and compare 
> its requirements to what's already working in GNUstep, it would help 
> enormously. First, they might see that there are only a few methods or small 
> classes missing in GNUstep that they require. If it's a small number they 
> might decide to implement those pieces themselves. Or someone in the GNUstep 
> community might decide to implement them. The OS X developer might also check 
> back occasionally to see how things are progressing relative to their 
> specific needs, if it was easy to do.

Let me share some experience of porting some Apple Sample Code. It turned out 
in many cases to be just some removal of the latest fancy stuff from OSX 10.6 
or later, because Apple wants to push their most recent additions...

This result may sound strange, but 10.4 and Obj-C 1.0 was almost feature 
complete. Everything introduced since then were rather marginal improvements, 
but breaking backwards-compatibility. E.g. you can always call getters and 
setters with the [] notation, but if someone decides to use the dot notation 
(or even mix both), it needs a newer SDK, but can be backported manually. And 
many methods and classes introduced later just simplify what could have been 
done before with a little more code.

But, this is experience done *after* deciding to port something to GNUstep...

I.e. it does not really help or convince any OS X developer to start to port 
their app to GNUstep. They will doubt and will simply hesitate to 'backport' 
something and modify/expand their code, because they have learned how easy to 
use e.g. array and dictionary constants are.

This is IMHO the real reason why people are dreaming of 10.9 compatibility...

> 
> If the tool

I am not sure if that can really be automated well enough. At least a simple 
thing would be to scan the application bundle and its binaries which frameworks 
they reference - and check with the GNUstep list.

> were implemented in such a way that the information could be shared with the 
> GNUstep developers (such as making it submit each application profile to an 
> online app for comparison against the current code) it would provide some 
> incredibly valuable information about which API's are used most. This would 
> not only help focus development efforts on the missing pieces, but also 
> possibly on optimizing the existing parts based on their level of use.
> 
> This would primarily help to target the "Commercial developers, who have an 
> Apple-only app" users. But I think if that group is helped it will likely go 
> a long way toward what other users need as well. And it may help draw in more 
> commercial developers to contribute to GNUstep.
> 
> On a related note, I think the idea of taking a pragmatic approach to 
> implementation (focusing on the methods and classes that are used and needed, 
> rather than striving for a complete implementation of any particular 10.x 
> release) makes good sense, to produce the most valuable results with the 
> effort available.
> 
> Cheers,
> 
> Doug
> 
> 
> _______________________________________________
> Discuss-gnustep mailing list
> Discuss-gnustep@gnu.org
> https://lists.gnu.org/mailman/listinfo/discuss-gnustep