jed: loosen build-depend on libslang2-dev >= 2.0.7

[G. Milde]

On 24.05.07, Jörg Sommer wrote:
> Hello G.,

> G. Milde schrieb am Fri 18. May, 18:47 (+0200):
> > Forwarded message from "G. Milde" <milde at users.sourceforge.net>
> > > On 17.05.07, Rafael Laboissiere wrote:
> > > > * Jörg Sommer <joerg at alea.gnuu.de> [2007-05-16 11:42]:

> >   slang-2.0.6 had a bug in the handling of byte-compiled files
>     ^^^^^^^^^^^^^^^^^^^^^

> I don't know why you say ???slang had a bug??? and file a bug report to the
> jed package. 

1. I experienced a segmentation fault with Jed (and xjed) when trying
   to open a latex file. IMO, this is a serious bug.
   
2. No report about this problem was found in the Debian bug database.

3. Time-consuming experiments and discussion on the jed-users list
   revealed the reason of the segfaults, a bug in S-Lang <= 0.2.6.
   
4. Upgrading to libslang2 2.0.7 (then in unstable) solved the problem.
   I refrainded from reporting the bug.
   
5. Jed was requiring the bug-free slang version but this was changed
   due to lack of information, not as a well considered choice:   
   
> > > > > why do you force a build against libslang2-dev 2.0.7? What's the problem
> > > > > with prior versions? I don't see, why we need this.
> > > > I probably had a reason for that when I did the change in SVN revision 655
> > > > but I can not remember why.  I will revert this in the next release.

6. This triggered my bugreports to both jed and libslang2 to have the
   problem documented for other Debian users bitten by this bug.

> Why didn't you file the same bug report to the xjed, slrn, slsh and
> other packages they use SLang?

Because I found the bug during the daily use of jed for productive work.

I agree that for consistency I should have filed the bug against xjed as
well. I considered jed-common as this is used by both, jed and xjed, but
the libslang2 dependency is not in jed-common but in jed and xjed.

I do not use slrn and only occasionally slsh, so I did not encounter any
problems there. I would have bugged mc as well if it had segfaulted on
me.

> Would you file a bug report against all packages that use libc if you
> find a bug in libc?

No. But I would file a bug report against a package that segfaults
(even it this is because it uses a buggy version of a library).

> Should we also set a dependency on the newest version for all other
> dependend packages? 

No.

> There was a bug in libxf?-dev that generated to much dependencies which
> complicates library transitions from unstable to testing???at least an
> important bug. Should we set a version dependency for libxf?-dev, too?

Does this seriously affect Jed users?

> When can we drop such a dependency?

This is similar to the dependency on a new feature in some package:
When the debugged version of the required package is the default one,
the versionned dependency can be dropped.

> I suggest to reassign the bug to slang and merge it with the bug report
> there. 

Does this leave some trail for the Jed user trying to find out why Jed
crashed? If yes, this is the way to go.

> IMO it's wrong to add a dependency to work around???nothing else are
> you doing with adding the dependency???a bug in a broken library. 

???

> I wouldn't file a bug report against all or random packages, if I find
> a bug in fread()???that's stupid.

Be assured that I would not file bug reports against random packages
either. However, I would file a bug report against a program that
crashes whenever I want to open a file.

Guenter