Re: Debian's idiosyncratic complexification of Emacs

[Stefan Monnier]

>>>>> "Manoj" == Manoj Srivastava <address@hidden> writes:

> On Wed, 23 Jul 2008 15:49:59 -0400, Stefan Monnier
> <address@hidden> said:  

>> It strikes me that Debian's Emacsen seem to not be plain enough.  I
>> mean, Debian seems to change Emacs's startup.el even tho there's no
>> need for it.  Instead of changing startup.el to (load
>> "debian-startup") and call some magic function in it, it'd be much
>> better to leave Emacs's own startup code unchanged and simply provide
>> a site-start.el that loads debian-startup as well as
>> /etc/emacs/site-start.el and all the rest.

>         Well, site-start.el was deemed to be for site specific stuff,
>  and is shipped empty by the vendor (i.e., Debian).

AFAICT, Debian's site-specific stuff is in /etc/debian/site-start.el, so
they could use /usr/share/emacs/<revnum>/site-lisp/start-start.el for
the Debian-specific changes.

>  The Studd in startup.el is only to cater to the vendor changes for
>  third party Elisp packages, and is not really a site specific
>  change. This means that the load-path is customized even
>  when --no-site-file is specified.

I don't understand what you mean here.  AFAICT, the Debian changes are
executed iff the site-start.el file is loaded, so they could do it in
the site-start.el file just as well.

>         Isn't there some issue with order of loading there? By modifying
>  startup.el, changes are made that allow setting load-path before _any_
>  action is taken, namely, language setting, window system
>  initialization, and option processing.  So the two things are not
>  equivalent.

There could be.  But as the patch you sent shows, it doesn't apply here:
the debian-specific changes are placed right before loading
site-start.el, so they could just as well be placed directly in
site-start.el.


        Stefan