This is the mail archive of the
mailing list for the Cygwin project.
Re: nano - packaged and ready for some criticism :)
Christopher Faylor wrote:
> On Fri, Nov 16, 2001 at 04:27:38PM +1100, Robert Collins wrote:
> >> --> The info package should have this script -- either in
> >> or /etc/postinstall/info.sh (or both).
> >Who is the info maintainer? Chris? Can said maintainer please put a
> >postinstall script to regen info.dir completely in the next release....
> I guess I'm maintaining texinfo.
> I can do this, sure.
> I wonder if all of the packages correctly rely on info, though.
Good point. If they install info files, then they need to call
install-info during the setup/install process -- which means they need
to depend on texinfo. (I betcha my own packages don't follow this
> And, this just causes info to regen the 'dir' file if it is installed,
> right? It won't solve the problem of adding a new .info file.
You are correct. If texinfo has an /etc/postinstall/info.sh script,
then it will regen the dir file when *texinfo* is reinstalled. If
texinfo has a /usr/bin/fix-info script, then we can instruct new
(confused) users to run that script if/when their info stops working.
Neither solves the problem of what setup should do with a binary package
that contains info files. Possibilities:
1) all maintainers must include a postinstall script that calls
install-info for each of their own info files.
2) setup magic:
(in C++) for each fn in (`tar tvjf foo.tar.bz2 | grep
'^usr/info/*.info'`) ; do
(shell out:) system("install-info --info-file=$fn
(back in C++) done
(Yeah, I know my "pseudo-code" looks a lot like sh.)