Porting Cocoafibs

[padski]

see CocoaFIBS is now opensource for the preceding discussion.

[padski]

burper wrote:

Thanks.
I'm past at least some of the package headaches. I'm on FC4, but it sounds like ubuntu is the place to be these days. Can't keep changing horses though.

yeah, I'm on debian myself. haven't had much chance to look at ubuntu.  It's much more important to know the system you're on.

In any case, gnustep-base seems to be making alright now. Will forge ahead with it.
Are you saying that you are building cocoafibs on linux/gnustep when you mention the 472 compiler warnings? Or is that on mac?

yes I'm building on macosx with xcode 2.2.  Haven't seriously looked at building on linux, for the reasons I gave before: it's likely to be more than just building, quite a bit of porting effort.  It would be fantastic to get past that though!

My energy may not last long, if history is any judge. Many other things to care for.
I'll let you know how far I get.

more:
gnustep-make, base, gui and back built and installed ok I think.
Trying to build cocoafibs now. It at least starts, but there is no GNUmakefile, so I started my own. Any idea what it should look like, at least wrt source file ordering?

dunno. Don't depend on the #includes to get a full dependency graph but I wouldn't worry about that to start with anyway. don't know if there's anything special in the link order, but I doubt it.  The compiler in xcode is gnu, so I'll try to dig deeper and get some proper output from a build, instead of all this pretty windowing nonsense

maybe this helps
CocoaDev Wiki: GNUStep

wandering around xcode I saw an option to output some of the xcode binary files in a text format if that helps.  I'll take a closer look and get back to you.

[burper]

Well, for one thing, the nibs files need to be converted to gmodel (apparently),
and that can only be done on a mac.
http://freshmeat.net/projects/nib2gmodel/

I just sent my shiny new MacBook off to school with my daughter, who is not a techie, so I won't ask her to do it.

Even then, whatever I do on linux, I would want to test out on Mac, so what I have here is a sub-optimal developemnt environment

I see what you mean about the licensing as well, it just doesn't appear anywhere in the code or anywhere on sf.net. I presume that legally puts it in the public domain, strictly speaking. True, his intentions to OSS it are clear by his fb posting and the fact that he put it up on sf, but which license? And what about fcm, or isn't that actually used?

In any case, I'll pick away at it. I actually have less warning messages than you! Of course, that's because I run into some errors that you don't, but about half the code actually compiles.

I'll take whatever patches you have that make warning messages go away, but I have bigger problems anyway, so no hurry.

[padski]

Well, for one thing, the nibs files need to be converted to gmodel (apparently)

not if you're using the 0.11 according to this chap

Even then, whatever I do on linux, I would want to test out on Mac, so what I have here is a sub-optimal developemnt environment

well, I'll cheerfully test anything you like.  I may even be able to sort you out remote access to a mac.

I see what you mean about the licensing as well, it just doesn't appear anywhere in the code or anywhere on sf.net. I presume that legally puts it in the public domain, strictly speaking. True, his intentions to OSS it are clear by his fb posting and the fact that he put it up on sf, but which license? And what about fcm, or isn't that actually used?

It's worse than I realised. I recalled seeing a BSD style license text, but that is just in FIBSCookieMonster (which I'm pretty sure is used). Agreed the general intention is clear, and obviously we should be grateful to agerson for his generosity in releasing the code, but I think we would have a hard time getting this distributed by, for example, debian or fedora. And I'm also concerned at the missing copyright notices and the content not authored by agerson.  It's the possibilty that content may have mistakenly been included for which the author does not have the proper license which worries me in this situation, and the hassle of chasing down the details.

I don't mean to sound petty. You know how it is: sorting that out, getting a cleaner handover of licensing and sourceforge access would make it a lot easier to move forward.

In any case, I'll pick away at it. I actually have less warning messages than you! Of course, that's because I run into some errors that you don't, but about half the code actually compiles. I'll take whatever patches you have that make warning messages go away

that's a pretty good result.  I had hoped from the stuff I'd seen on porting that most of it would port easily.

the vast majority of warnings I see are trivial:

* unused variables
* methods not found (missing #include, or prototype missing from .h)

I have most of these done.  I'll try to crank out that patch and the xproj thing tonight

but I have bigger problems anyway, so no hurry.

[padski]

burper,

sent some bits and pieces to your email, and I submitted a registration for a sourceforge project in the name stepfibs [ which I thought was rather witty ]