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

Bug#42634: PROPOSAL] Automatic migration to /usr/share/doc



On Tue, Aug 17, 1999 at 08:39:37PM +0100, Ian Jackson wrote:
> Instead, I propose that:
> * Any dpkg bug in this area be fixed.  If I can figure out what people
> claim the bug is I'll fix it.  (I won't build an NMU, but we seem to
> have no shortage of people willing to do dpkg NMUs.)

Please fix it first. Show us the code, or get out of the way, so to speak.

We've had "dpkg will be fixed RSN" for at least the last two years. It's
getting old.

> * Existing packages which use /usr/share/doc, and all new packages
> which do so, should be made to contain a predepends on the base-files
> I mention below.

That is, every package from now to enternity should have a versioned
pre-depends on base-files. How strikingly elegant.

> * Base-files should be changed.

  * Base-files must require that the fixed dpkg is installed.

If base-files doesn't depend/pre-depend on dpkg, downgrading dpkg and
downgrading any other deb that uses /usr/share/doc will lose all its
documentation.

Hrm. This doesn't allow you to ever downgrade either base-files or
(transitively) dpkg (at least without downgrading every other package
on your system).

> The preinst should check to see if
> both [/usr/doc and /usr/share/doc --aj] exist, and:
>  - if they do and have different contents it could print a warning.

ie, it will fail for everyone who's working from potato right now, and
continue to do so until woody when everything uses /usr/share/doc. But
hey, that's the fault of those horrible people like the libc6 maintainer
who actually follow policy. How inconsiderate of them.

>  - if only /usr/doc exists it should be renamed to /usr/share/doc and
>    a symlink left behind
>  - if only /usr/doc exists but /usr/share is a different fs, it
>    should make the reverse symlink.

That is, anyone who already shares /usr/share amongst different machines
will specifically have their documentation put in the wrong place.

> The effect would be that people who installed preemptive packages from
> unstable would have both a /usr/share/doc and a /usr/doc until all the
> packages were migrated (or they moved them themselves); other people
> would get a nice transition if they had different filesystems, or
> could move it themselves if a mount point is involved.

We already have a proposal that works for everyone (even the people who've
installed "preemptive" packages [0]) *now*, without modifications to
dpkg, without pre-dependencies, without limiting how you can downgrade
packages, and that doesn't require packaging related cruft for the
remaining lifetime of Debian.

Against, is that it uses symlinks, and symlinks are bad, mmmkay? [1]

Cheers,
aj

[0] The ones that followed policy after it changed. Kind of a reverse
    preemption. Postemption, maybe. Well-maintained, or up-to-date,
    might be other words.

[1] Or do we actually have a demonstration of how using symlinks might
    fail? Or, for that matter, how /usr/doc -> /usr/share/doc is
    any different to /usr/doc/<foo> -> /usr/share/doc/<foo> wrt this
    bug.

-- 
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. PGP encrypted mail preferred.

 ``The thing is: trying to be too generic is EVIL. It's stupid, it 
        results in slower code, and it results in more bugs.''
                                        -- Linus Torvalds

Attachment: pgpvETRboRSfj.pgp
Description: PGP signature


Reply to: