Bug#190753: About dropping the ‘should’ recommendation to rename binary programs using a suffix to indicate their programming language.
Le Mon, Oct 05, 2009 at 08:00:14PM +0200, Bill Allombert a écrit :
>
> The goal of removing the language suffix is precisely to avoid to have to
> edit your script when the program is rewritten in a different language.
Le Mon, Oct 05, 2009 at 11:10:24AM -0700, Don Armstrong a écrit :
>
> The practical advantages are that 1) you can reimplement scripts
> without renaming, 2) you don't make it more difficult on public users
> of the binary who have to remember both the name and the entirely
> arbitrary aspect of what language it was implemented in[1] and 3)
> things are consistently named, no matter what the package is.
>
> Furthermore, the policy is a *should* directive, not a *must*
> directive. There are many cases where there are things that are done
> wrongly, but fixing these historical mistakes are more costly than
> living with them. A wontfix bug with possible lintian overrides may be
> acceptable in such rare cases.
Le Mon, Oct 05, 2009 at 02:07:10PM -0700, Russ Allbery a écrit :
>
> Based on the debian-devel discussion and on past discussions in
> debian-policy, I'm fairly sure that we don't have consensus to remove this
> requirement. I suspect this is going to need to go to the Technical
> Committee unless the discussion here surprises me a lot.
Thank all for your fast answer.
I think that the core of the disagreement is on how frequent the
re-implementation in a different language happen. My experience is that in my
field, bioinformatics, it is close to zero. Moreover, when programs with
similar function and same basename are written, they are most often not
designed to be a ‘drop-in’ replacement. The filename extension is therefore
part of the name, and removing it only creates problems and difficulties. I
will stop to obey the the ‘should‘ Policy directive, because I think that this
is the right thing to do in my profesional environnement.
As noted by Steve M. Robbins (20090929062143.GA5792@sumost.ca), this will
formally mean that my packages have a bug from a Debian point of view, which I
constest, even at the wishlist level. This is why in addition to ignore the
directive, I propose to change it.
There is no consensus for the change, but I would like to underline that the
directive itself is not consensusual, as some other developpers supported me in
the thread on debian-devel. I think that this is a strong indication that the
directive must not be a should and that the final decision must be left to the
maintainer, without making his package buggy.
What is the timeline now? If the developers who think that everybody in Debian
should rename scripts with name suffix indicating a language have no other
arguments to add, then indeed it is the moment to call for the technical
commitee to take a final decision.
On my side, I am willing to make concessions, so a rewording of the should
directive to explain when exceptions are acceptable could be an alternative.
But my undertaning of a ‘should’ directive is that it would not make sense if
they do not apply in most of the cases, which here is my point of view: the
programs that the Policy wants me to rename do not in any way have the same
scope as /bin/ls, nor are they as standardised and widely used enough that they
could be called a part of a ‘public API’. They are utilities, of which we make
access more difficult for purely speculative reasons.
Have a nice day,
--
Charles Plessy
Debian Med packaging team,
http://www.debian.org/devel/debian-med
Tsurumi, Kanagawa, Japan
Reply to: