On Wed, Jul 07, 1999 at 11:02:26AM -0700, Joey Hess wrote: > If the upstream changelog file is HTML formatted, it must be accessible as > `/usr/doc/<package>/changelog.html.gz' with > `/usr/doc/<package>/changelog.gz' being a symbolic link to it. Er, uhh, I don't like that because I think that, in general, HTML documents should be identifiable as such by their filename. Ideally, they should have the ".html" suffix. How about mandating a non-HTML changelog.gz generated with lynx -dump -nolist | gzip? (It's important to use -nolist because lynx will the use absolute pathnames of file URLs, and these will almost always be totally wrong because at package creation time you're not working off the root, you're working in someplace like /mnt/this/is/where/I/build/my/package/foo-1.2.3/debian/tmp/usr/doc/... or something like that.) I guess I am proposing this as an amendment. I would not be comfortable with changelog.gz's that might be HTML or might not. I think they should always be (gzipped) plain text. Whether we should actually make all package maintainers do lynx dumps of available HTML changelogs, I don't actually feel that strongly about. I personally would have no problem complying, though. -- G. Branden Robinson | Psychology is really biology. Debian GNU/Linux | Biology is really chemistry. branden@ecn.purdue.edu | Chemistry is really physics. cartoon.ecn.purdue.edu/~branden/ | Physics is really math.
Attachment:
pgp0dwQEi2p_5.pgp
Description: PGP signature