emacs-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: address@hidden: rcs2log]


From: Stefan Monnier
Subject: Re: address@hidden: rcs2log]
Date: Tue, 29 Oct 2002 14:22:13 -0500

> I see the way that vc uses rcs2log as being the same as the way that hexl
> mode uses an external helper program.

There's a subtle difference: hexl-mode is completely unusable without the
`hexl' program, whereas VC is fully functional even without rcs2log.
I.e. support for rcs2log is an added feature to VC, but it is not the
main purpose of VC.

> One is Eli's suggestion, namely a more sophisticated replacement for my
> rcs2log.bat file that deduces the correct search path to use for sort.
> Clearly this should not use any non-standard Windows technology other than
> what is necessary anyway to run rcs2log.  (So, for example, a perl script
> would not be not a good solution, although using Windows Script Host might
> be.)  A good solution would be to pull the appropriate Cygwin path out of
> the Registry, but I'm not sure how feasible that is.
> 
> The other is to modify vc.el to allow the user to specify a directory
> containing the correct sort program and to prepend that to exec-path before
> trying to execute rcs2log.  This could be made to happen only on Windows
> platforms.

I think modifying vc.el is completely wrong.  Right now vc.el doesn't know
anything about rcs2log apart from some arguments it can take and what
things it might output.  It doesn't know it's written in `sh', it doesn't
know it uses `sort'.  And indeed, rcs2log might be rewritten in Perl
some day and stop using `sort' and `sh'.

Also rcs2log is actually not particularly linked to Emacs (which
is why it's also distributed along with CVS), so any fix for w32
should allow it to work correctly even when it's not run from Emacs.

> I'll think about these two options and see whether I can produce a solution
> that people find more acceptable than my original rcs2log.bat solution.

I think an rcs2log.bat wrapper is the right approach, I believe.
I actually haven't seen any disagreement with it.  The only problem
mentioned so far has to do with `sort' which is basically unrelated
to the problem at hand (except for the fact that it is related to the
same script and to the same platform).


        Stefan





reply via email to

[Prev in Thread] Current Thread [Next in Thread]