[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Dpkg-shlibdeps and libc5/6



From: Philippe Troin <phil@fifi.org>

>It's fakeroot which breaks the thing.
>Transcript:
[...]
>
>Gasp. I either have to encode unset LD_PRELOAD in debian/rules or
>revert to the ol'su method ! (*or* as an exotic method, I could chmod
>u+sx my library, making it discard the LD_ environment variables... ???
>nah...)

FWIW, there's another, far simpler way (although it's quite pervasive
in the design of debian/rules); it's the way I've built all the
packages I've done since joining Debian (only a couple of months ago,
but I think I have something...)

- Put the stuff which would normally go in the "build" target of
  debian/rules into a "do-build" or "real-build" section.

- Have the "build" section depend on "do-build", and install the
  package into debian/tmp and do everything up to and including
  dpkg-shlibdeps and dpkg-gencontrol.

- Have the "binary-*" section do only this:

  chown -R 0.0 debian/tmp
  chmod -R go-ws debian/tmp
  # apply any special owner/group/mode settings here...
  dpkg-deb --build debian/tmp ..

This works for me, and has the advantages that (prior to fakeroot) it
didn't involve running "make install" as root, and (with fakeroot) that
dpkg-shlibdeps doesn't get run with LD_PRELOAD set.

If you want to see a generic debian/rules based aroud this model, look
in the virtual-dev package I've just uploaded.  The debian/rules is
completely generic (I hope); the real work is simpler than usual, and
happens in another executable makefile, debian/process.  I'm currently
poised to write some extra utilities around this to create YADR (yet
another debstd replacement(tm)).

Just my tuppence-ha'penny.

--Charles Briscoe-Smith
White pages entry, with PGP key: <URL:http://alethea.ukc.ac.uk/wp?95cpb4>
PGP public keyprint: 74 68 AB 2E 1C 60 22 94  B8 21 2D 01 DE 66 13 E2


--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .


Reply to: