Re: Use directory-local variables to establish some coding styles in Emacs (issue 6460109)

[John Mandereau]

Il giorno lun, 20/08/2012 alle 08.28 +0200, David Kastrup ha scritto:
> The possibly contentious case is when the user has put up an independent
> .dir-local.el file.  Git will complain before overwriting that without
> confirmation, so you'll get warning (not for git reset --hard or similar
> though).

IIRC Git also complains if you do "git checkout", and I don't think it's
good usage of Git to update branch and working tree with "git reset
--hard" without knowing what you may loose, so Git standard behaviour is
expected to make the user aware of such a contentious (look, it should
have happened recently with many developers here, because aclocal.m4 at
top of the source tree changed status, from a generated ignored file to
a file tracked by Git).  I don't think we should bother.


> Well, it is the choice between those 95% of developers that have not
> bothered to read and/or understand the CG unwittingly obeying the coding
> standards, or those 95% of developers unwittingly ignoring them.
> 
> Among those 5%, probably less than a fifth will want additional settings
> _right_ _there_ (and we don't currently have more than hundred active
> developers).  They will be inconvenienced, having to set a
> .dir-locals.el in subdirectories instead, or configuring per-file
> variables in the top directory.

IMHO, in case there is a developer inconvenienced by this
directory-local variable , she'll be enough pissed to speak up on this
list...

> One could let .dir-locals.el load an optional user-written file, but
> add-dir-local-variable would still pick the default file for placing its
> information.

... and that same developer will also likely be enough qualified to
provide a patch providing this feature.


>   I doubt we should bother unless somebody actually
> complains.

Agreed.

FWIW, I still have never set fill-column to 66 in Texinfo mode, at some
time I didn't know how to do it, now I can no longer pretend not to
know.  IIUC Graham, Jan and I approve adding this directory-local file
to Git, and this has raised much discussion enough so that it's
reasonable to expect that no developer following current development has
missed this, so I propose to keep the matching Google issue 2757 open
and let the patch associated 6450162 go for countdown.

John