Subject: v29INF4: archive and submission format question for comp.sources.unix Newsgroups: comp.sources.unix,comp.sources.d Approved: paul@vix.com Submitted-by: paul@vix.com (Paul Vixie) Posting-number: Volume 29, Info 4 Archive-name: index29.4 I've been thinking for a long while that the format used by the comp.sources.unix archives is not as useful as it could be. What we do is create a directory for each volume, and under that we create a directory by the name of the package (perhaps qualified by its version number), and under that we place files like "part01" and "patch01", which are eventually compressed with "gzip." My first take on how to make it better was to change volumeNN/PKG-VER/{part,patch}XX to PKG/VER/{part,patch}XX since I have observed that most people don't care about the volume numbers once they're at the point of searching the archives. I have decided that this would be good as far as it goes, but does not go far enough. Though these sources are intended for UNIX systems, good software will run on many platforms and folks are often trying to run c.s.unix library code on non-UNIX systems. In one case I heard about, someone on a Mac wanted to use my old AVL tree functions and they picked apart the SHAR file with a text editor. This seems nonideal. Therefore I would like to start archiving *.tar.gz files containing the package applied, and I would like to do away with the SHAR files altogether. We would still keep the patch files separate unless an author wanted to submit an updated *.tar.gz file with all patches applied. There's more, though. Why are we sending SHAR files around, given that in this era it is possible for all Internet citizens to fetch (via ftp, www, or ftp-by-mail) any URL they want? If we're going to archive by *.tar.gz file, why not put the archives up on WWW (and FTP, of course) and just post the README files with URL's? (Note that I am willing to include the "master URL" for any given package as an additional data item, but a copy of the *.tar.gz file itself must be present in the c.s.unix archives before I'd post anything about it.) Another benefit of this scheme is that many authors do not know what a SHAR file is or how to make one -- but *.tar.gz files are easy to make and use, even on non-UNIX systems. If you agree with this proposal, please keep silent. I think it's a great idea and I expect that all right thinking :-) people will think so too. If you do not agree with this proposal, I'd like to hear from you. Paul Vixie