Re: [Caml-list] [ANNOUNCE] otags reloaded 3.12.1 for OCaml 3.12.1

[Hendrik Tews <tews@os.inf.tu-dresden.de>]

Stéphane Glondu writes:
   
   It looks like a hostile takeover, and you cannot do that without the
   maintainer's agreement unless the package is orphaned. 

No, no hostility intended. It was only that at first, I didn't
see your reply from January 19 (because it was debian-ocaml-maint
only). I therefore thought it would be best if I go ahead and
prepare a package. Only after I announced my package I saw your
replay in the debian-ocaml-maint archives.

   You obviously redid the packaging from scratch. 

Yes, because that was much simpler for me. I also thought that a
simple rules file has its benefits.

    If you don't use Git, I (or someone else) will update the repository

OK, I now understand that you want to keep otags in the git
repository. What should I do in order to obtain write access
there?

   Moreover, I saw that the debian/ repository is also part of the upstream
   tarball. 

I know that this is problematic. However, for the start it was
easiest for me to put debian/ into my own (upstream) repository.

   > I have not yet integrated the update-otags script for tagging
   > /usr/lib/ocaml. This has to wait until otags does not crash any
   > more because of bogus locations (see ocaml bug #5159).
   
   Even if you remove update-otags, you should keep the prerm script to
   clean up after an upgrade from Squeeze.
   
I have almost finished a workaround for #5159. This will be
released as 3.12.2. After that I would update the package and
incorporate update-otags.

   As Ralf said, "camlp4-extra (>= 3.12.1), camlp4-extra (< 3.13.)"... but
   why do you want the "< 3.13" part?
   
Otags works on the internal abstract syntax tree. It has to be
changed with every major release. My new otags contains the
camlp4 parsers internally. It is very likely that they will
produce syntax errors with 3.13 sources. Further, with
non-standard syntax extensions, otags uses an external camlp4 for
parsing, which sends the ast back as marshaled value (printer
DumpCamlp4). This cannot work with a different major version.

Bye,

Hendrik