Re: write-file could carry permissions

Re: write-file could carry permissions

[Richard Stallman]

    $ chmod +x something
    $ emacs something
    C-x C-w somethingelse
    $ ls -l #shows that the executable permissions were not carried along,
    as a convenience...

If you write the buffer into a different file, I don't see why the
permissions would be the same.  However, this suggested to me that
maybe Emacs should always set the executable bit when you edit a file
in Shell Script mode.  Is there any reason that would be bad?

Re: write-file could carry permissions

[Thien-Thi Nguyen]

Richard Stallman <rms@gnu.org> writes:

   maybe Emacs should always set the executable bit when you edit
   a file in Shell Script mode.  Is there any reason that would be
   bad?

if you write a file in a directory where some cron-driven script
walker sees it, it is better if that file is not executable unless
you explicitly make it so.  (principle of least surprise according
to the sysadmin.)

on the other hand, there's no problem w/ emacs providing a
convenient way to effect this decision (e.g., a write-file hook
subject to y-or-n-p).

thi

Re: write-file could carry permissions

[Ehud Karni]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 13 May 2003 00:39:27 -0400, Thien-Thi Nguyen <ttn@glug.org> wrote:
> 
>    maybe Emacs should always set the executable bit when you edit
>    a file in Shell Script mode.  Is there any reason that would be
>    bad?
> 
> if you write a file in a directory where some cron-driven script
> walker sees it, it is better if that file is not executable unless
> you explicitly make it so.  (principle of least surprise according
> to the sysadmin.)
> 
> on the other hand, there's no problem w/ emacs providing a
> convenient way to effect this decision (e.g., a write-file hook
> subject to y-or-n-p).

There is an autoloaded function supplied in current emacs (21.3) that
does that - `executable-make-buffer-file-executable-if-script-p' in 
progmode/executable.el. Just add it to `after-save-hook'.

BTW. Beside Thien-Thi reservation about cron confusion, there is also
the problem of trying to change file mode on remote system (ange-ftp)
that usually fails (remote chmod not supported).

Ehud.


- -- 
 Ehud Karni           Tel: +972-3-7966-561  /"\
 Mivtach - Simon      Fax: +972-3-7966-667  \ /  ASCII Ribbon Campaign
 Insurance agencies   (USA) voice mail and   X   Against   HTML   Mail
 http://www.mvs.co.il  FAX:  1-815-5509341  / \
 mailto:ehud@unix.mvs.co.il                  Better  Safe  Than  Sorry
-----BEGIN PGP SIGNATURE-----
Comment: use http://www.keyserver.net/ to get my key (and others)

iD8DBQE+wMjHLFvTvpjqOY0RApHKAJ9wk3cZPHx5R8/73iXEf4NFwg+pxQCfX9Ed
RN/VHFQueZXC6pXBz05xtRQ=
=2x+H
-----END PGP SIGNATURE-----

Re: write-file could carry permissions

[Dan Jacobson]

>>>>> "RMS" == Richard Stallman <rms@gnu.org> writes:

RMS>     $ chmod +x something
RMS>     $ emacs something
RMS>     C-x C-w somethingelse
RMS>     $ ls -l #shows that the executable permissions were not carried along,
RMS>     as a convenience...

RMS> If you write the buffer into a different file, I don't see why the
RMS> permissions would be the same.  However, this suggested to me that
RMS> maybe Emacs should always set the executable bit when you edit a file
RMS> in Shell Script mode.  Is there any reason that would be bad?

Well, I wasn't asking for it to go that far.  Now look what I've
started. OK, never mind my request.
-- 
http://jidanni.org/ Taiwan(04)25854780

Re: write-file could carry permissions

[Stefan Monnier]

>     $ chmod +x something
>     $ emacs something
>     C-x C-w somethingelse
>     $ ls -l #shows that the executable permissions were not carried along,
>     as a convenience...
> 
> If you write the buffer into a different file, I don't see why the
> permissions would be the same.  However, this suggested to me that
> maybe Emacs should always set the executable bit when you edit a file
> in Shell Script mode.  Is there any reason that would be bad?

As someone already pointed out,
we have make-buffer-file-executable-if-script-p and I recommend everybody
add it to his after-save-hook.

As for making all sh-script mode buffer executable, I strongly
oppose this idea, since many script files are never meant to be executed,
but only sourced (this is the case for .bashrc kind of things for example,
but also for many other files when you write an extensive script and
split it into modules/libraries/...).

Another reason why I oppose this idea is because the after-save-hook
setting above is a much better solution which also covers awk scripts,
prl scripts, foo scripts, bar scripts, ...


	Stefan