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

Re: Bug#807019: tracking bin-num - broken unison due to binnmu upload



Le 06/12/2015 12:15, Norbert Preining a écrit :
> * is there a way to track down who uploaded -3+b1?

For "who", I don't know. But for "why", cf
/usr/share/doc/unison2.40.102/changelog.Debian.amd64.gz:
> unison2.40.102 (2.40.102-3+b1) sid; urgency=low, binary-only=yes
> 
>   * Binary-only non-maintainer upload for amd64; no source changes.
>   * Rebuild against ncurses 6.0.
> 
>  -- amd64 / i386 Build Daemon (babin) <buildd-babin@buildd.debian.org>  Fri, 31 Jul 2015 09:50:21 +0200

...which is strange, because unison doesn't use ncurses AFAICT. Also,
the date is misleading; it corresponds to the last sourceful upload, not
the binNMU.

But I now understand the problem: unison2.40.102 uses Obj.magic (i.e. an
unsafe coercion) to cast a format string into a string. The previous
unison version was compiled with OCaml 4.01.0, where format strings were
indeed strings. The new version was compiled with OCaml 4.02.3, where it
is no longer the case. unison2.32.52 should suffer from the same problem.

The change done in unison 2.48 to overcome this looks pretty big... I'm
not sure I'll be able/willing to provide a unison2.40.102 any more.
Moreover, this package was created to provide compatibility with
previous Debian releases, but another change in OCaml 4.02 makes it
incompatible anyway (both communicating unisons need to be compiled with
the same version of OCaml in practice, which won't be the case any more
when one side is Debian stable, and the other Debian testing). IMHO,
that's a design flaw in Unison that cannot be easily fixed.

-- 
Stéphane


Reply to: