Bug#919951: ocaml builder must not be called `dune' or provide /usr/bin/dune
Le 20/01/2019 à 23:14, Ian Jackson a écrit :
> In #919622 and the associated debian-devel thread,
> "Conflict over /usr/bin/dune"
> https://lists.debian.org/debian-devel/2019/01/msg00227.html
> the file conflict over /usr/bin/dune was discussed.
>
> The rough consensus of the debian-devel thread was that /usr/bin/dune
> ought definitely not to be taken by the ocaml build system, and that
> the best claim on it was the C++ library which already provides a
> number of /usr/bin/dune?* binaries.
I do not agree there was such concensus.
You gave your opinion which seemed off-topic to me, Andy and Andreas
remotely agreed with you, Philipp had the same point of view as me.
Maintainers of whitedune did not object.
There is no conflict with the C++ library, I wonder why this was put on
the table.
> Instead, the maintainers of the ocaml package reassigned the bug
> against their `dune' package to the whitedune package, which
> previously provided /usr/bin/dune as a compat symlink.
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=919622
>
> They used the phrase
> "As discussed on debian-devel"
> which is very misleading because it makes it sounds like there was a
> consensus for this course of action, whereas the opposite is true.
Sorry but from my point of view, I proposed something and there was no
(relevant) objection. I thought you were giving your opinion in passing,
and didn't understand you would be so vocal about it.
> Apparently as a result of this there was an NMU of `whitedune' to drop
> the symlink /usr/bin/dune.
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=919622#58
>
> The maintainers of the ocaml `dune' have now uploaded `dune' (the
> ocaml package) with /usr/bin/dune and Breaks+Replaces to claim the
> file.
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=919622#99
You seem to imply that transgressions have been made... but as far as I
can tell, procedures were followed and the relevant people (should) have
been notified.
> Meanwhile there seems to have been no contact with the maintainers of
> the C++ library which is the only hit on Wikipedia for
> https://en.wikipedia.org/wiki/Dune_(software)
> (Amazingly, this is still true at the time of writing even though
> I referred to this fact in the debian-devel thread.)
I still fail to see how this is relevant.
> Note that this ocaml tool `dune' was previously known as `jbuilder'.
> It has nothing to do with Java AIUI. No-one has suggested a plausible
> charitable explanation for why the ocaml upstream made such
> egregiously bad naming mistakes twice in succession.
>
> Additionally the binary package name `dune' for the ocaml tool is bad,
> too.
Actually, I am not fond of the `dune' name either, but I think it's not
my job to judge and rename it. And I would feel ridiculous asking
upstream for a change (I am already ashamed of this situation).
> Please would the Technical Committee:
>
> * Declare that no-one is allowed the name /usr/bin/dune other than
> the C++ library dune-common or its friends.
>
> * Declare that no-one is allowed the binary package name
> /usr/bin/dune other than the C++ library dune-common
> or its friends.
>
> * Declare that the ocaml build system should choose a new source
> package name and use it henceforth.
>
> I am about to file an RC bug against the `dune' package, blocked by
> this one.
The tone of your mail looks like harassment to me.
Cheers,
--
Stéphane
Reply to: