Hi On Mon, Jul 28, 2008 at 09:58:17AM +0200, Josip Rodin wrote: > On Sat, Jul 19, 2008 at 11:27:55AM +0900, Osamu Aoki wrote: > > Until someone stands up to set up such system, we should allow makefile > > to fetch and extract unstable package to create html. If we use some > > conditional in package's make file, it should not be too difficult to do > > so. It is each subtree owner's decision and you can do it now. > > That would possibly be fine only under the condition that it fetches the > package and copies the files into its own directory (in place of files which > would normally get built), leaving the publish rule intact.
That is what I meant. > But even that is > wasteful because it has to make e.g. a mirror selection, which evades the > existing www-master scripts that do the same more consistently. Hmmm... any pointer? I was going to pick one mirror and curl/wget/dget deb package(s). > Also, we definitely don't want novel makefile rules to be directly publishing > stuff on the web site, because all sorts of strange things can happen with > that. The existing 'publish' rules are risky enough. Which publish rule are you talking. No I do not want to touch main Makefile. Project makefile: what is the difference in risk to make html files directly from source or copy contents from package to its publishing area. I mean that we run "dpkg-deb -x package.deb /var/tmp/something/" and copy html to install directory. Then emove /var/tmp/something/. > Anyway, I think we've spent a lot of time musing about this alternative > solution, which may give people the wrong impression. The cleanest solution ??? clean ???? or you mean ***practical***. tex-live backport is non-trivial though. I agree it is practical for most cases. I should do one backport, I know. > is to have the necessary tool-chain packages backported - it's a useful > thing for everyone, and we are after all a community of packagers - it's > not that hard for us to make backports :) Joy, you know we (or I) make things in hard way. Now I am having fun making pbuilder to support unstable and backport with a simple command line option (Bug #398404). Then I had to lern git and its mail interface to contribute to the upstream. I have not made back port yet... Osamu -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
