I always thought this is a bad idea that invites to produce a lot of non-tested packages. Why not just make two clean info files?
OK, let's look at these arguments:
Because it is a matter of scalability. When you have a piece of software that supports many variants like
with ssl, without ssl, with blah, without blah. with x , without x and so on, a maintainer cannot be expected to take care of 3 separate files just to take care of one software piece.
It still has to be seen whether this is easier with the quite complicated syntax of the variants system. The examples we have seen so far are not encouraging. Michèle's example shows that for beginners at least it is not easier.
Furthermore it is a matter of bandwidth and being considerate toward our user community. To some it does make a difference how much data they blast over the wire and not duplicating such fields as Description or the like also keeps the database smaller.
Info files are really not big. Downloading 4 separate info files for the 4 gimp2 variants would take how much on a slow phone connection? 2 seconds? The argument with the Description field is actually an argument against variants, IMHO: One should really have different descriptions for different variants. Or at least have a common DescDetail that describes the differences betweeen the variants. In the case of gimp2 this was solved by not having a DescDetail field at all. Not a good example. (Actually, gimp2 is not a good example at all, the DescPackaging field is abused for DescUsage, so that people don't see the good advice given there, unless they read the info file.)
Furthermore the amount of data that gets read during a fink index or a fink info is a lot smaller.
Did you test wether this is really faster? I doubt that it's true. The parsing of variants and the creation of the virtual multiple packages is probably slower than the parsing of multiple but simpler info files.
All in all I can understand your concerns, but they have nothing to do with variants themselves, it is only a matter of introducing a strict and proper package testing procedure.
This is a little naive. People's behavior is strongly influenced by the tools they are using. Do you really expect that the maintainer of gimp2 has tested all 16 packages created from the gimp2.info file?
My general advise would be to acquire a G5 or similar machine where selected people can have shell access. This machine would do nothing else but compile packages and their variants.
Nice dream :-)
-- Martin
------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click _______________________________________________ Fink-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/fink-devel
