On Fri, 5 Mar 2004 09:30, Alexander Malmberg wrote: > It's late, and I haven't had time to look at the patches in detail yet, > but I'm still not really happy about the general idea. A configuration > file in /etc does make sense, though, if you're running on a "normal" > un*x system, so for those cases, I guess it's ok.
Could you clarify why you are unhappy with the idea? If I know what the concerns are then I might be able to address them. > (Would be nice if the configuration system was easily pluggable, so that > "unnormal" unix-like systems could have GNUstep configure itself in some > convenient way (eg. LinuxSTEP or GNUstep on Darwin), but that can come > later.) Well, a lot of that is already there. Perhaps if I explain the logic this way: 1) Look at environment variables and set root paths from them 2) Look at conf and set variables 3) Look at user conf and set variables 4) Fall back to compile time defaults All I've added is step 2, really. So, if you're running on "unnormal" unix-like systems things will work fine if your compile-time defaults are properly set. The docs are poor but the functionality is already there. One thing I did want to do is add compilation switches to disable parts so specific behaviour could be forced. For example, forcing compile time values instead. I've not put them into the current submission to avoid further complicating things. One of the issues being discussed is how to set up such compilation options (separate thread). Suffice to say that, in future, you'll be able to disable any of the first three steps. It makes no sense to turn off (4). The conf system is flexible enough that it can support both the GNUstep 'standard' paths as well as the Darwin standard and Darwin does have /etc support. The name and path to the conf file is a compile-time variable so it can differ by platform. Regards, Sheldon _______________________________________________ Bug-gnustep mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-gnustep
