[ocaml-infra] caml.inria.fr/pub will be down in Jan 2016 for 48 hours

[François Bobot]

On 02/10/2015 16:01, Ashish Agarwal wrote:
> On Fri, Oct 2, 2015 at 9:30 AM, François Bobot <francois.bobot at cea.fr
> <mailto:francois.bobot at cea.fr>> wrote:
>
>     I will ask on devel to Xavier and Damien to know where they stand.
>
>
> Sure, but what is the benefit? Do you see any problems with the current setup?
>

It is only of cosmetic benefit, yet by moving to ocaml.org:
  - we could always use this website instead of caml.inria.fr in the readme file, except for the 
bugtracker.
  - bts.ocaml.org could redirect to mantis. (NB: caml.inria.fr doesn't have https, unlike ocaml.org 
and mantis seems not uptodate "Copyright © 2000 - 2011 MantisBT Group")
  - we could make `http://caml.inria.fr/pub/docs/manual-ocaml/` share the nice style of ocaml.org


Xavier Leroy wrote:
> On 02/10/2015 15:51, François Bobot wrote:
>>   In your comment in a merge request[1] I understood that the contents
>> in caml.inria.fr/pub/* doesn't go to ocaml.org in order to keep the job
>> of Damien the release manager easy. Previously Fabrice and I though the
>> reason was to keep /pub/ on caml.inria.fr in order to authenticate that
>> a particular tarball is the official INRIA's version, not a fork
>> modified by the community.
>
> Yes, Fabrice has a point with the authenticity of the sources.  But
> note that this could be ensured by other means, e.g. GPG detached
> signatures on releases, like the Linux kernel does.
>
>> Damien, could infrastructure at lists.ocaml.org understand your workflow in
>> order to mimic it?
>
> Well, at some point of Damien's workflow, it boils down to "drop some
> files in an NFS-mounted partition so that they are served by
> http://caml.inria.fr/".  Hard to make it any simpler than that
>
> For those who join the discussion: a problem with caml.inria.fr is
> that it will be down for 2-3 days in January as it getes moved
> physically from Rocquencourt to the new offices in Paris.  This will
> annoy a number of users, esp. since OPAM fetches the OCaml core
> sources from caml.inria.fr and nowhere else AFAIK.
>
> - Xavier