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

Re: Bug#231235: arabtex font installation is broken



Hi,

frank@kuesterei.ch (Frank Küster) wrote:

> In fact it's even more complicated:
>
> The filename clash makes it necessary for tetex-*base* to conflict with
> arabtex<=3.10-5 (and it was a mistake to put the Conflicts: to
> tetex-*bin*. Let's investigate:

Right. I didn't pay attention to the fact that the Conflicts was added
to tetex-bin. As I wrote in:

  http://lists.debian.org/debian-tetex-maint/2004/debian-tetex-maint-200402/msg00102.html

tetex-*base* must hold the Conflicts.

> arabtex: Depends tetex-bin >=2
> tetex-base: Conflicts arabtex <=3.10-5
> tetex-bin: Depends tetex-base=$nearly-same-version
>
> Thus, we have a circular situation, if all three packages are to be
> upgraded from woody: arabtex cannot be upgraded unless tetex-bin is
> upgraded, this needs tetex-base, but this cannot be upgraded because
> arabtex is still installed.

No, you are replaying what I wrote in the message that confused
everyone. Please reread the message pointed above, and tell me if it
doesn't solve the problem you think you are seeing (the key is to think
of {unpacking, configuration} and not only of {upgrading} of packages).

> There's no problem for a testing/unstable system which has tetex-bin>=2
> yet, but from woody its weird. I think it is really necessary that
> arabtex installs with both tetex_1.x and tetex_2.x. However, I think it
          ^^^^^^^^
           unpacks. Yes.

> does not really need to _work_ with tetex_1.x, I mean it is not
> necessary that all fonts are correctly registered even to tetex_1.x.

But having arabtex "Depends: tetex-bin (>= 2), tetex-base (>= 2)"
ensures that it will not be marked as installed in the configuration
where it doesn't work (ie, with tetex 1), because it will not be
configured unless tetex 2 is already configured. This does not change
anything to the unpacking order.

> If it doesn't work well, this is bad for anyone installing arabtex from
> unstable on his woody system. But this is tolerable, if Clint doesn't
> manage (or want) to handle both situations. You should document in
> README.Debian that the missing dependency on tetex>=2 is only to ensure
> a smooth upgrade.

IMHO, the dependency is safe and good to have. It will not break the
upgrade, and it will ensure that the new arabex works correctly once
configured.

-- 
Florent



Reply to: