Tramp and rcompile

Tramp and rcompile

[Stefan Monnier]

The `rcompile' code calls a Tramp macro via `funcall'.
Obviously, this can't work.  This code was introduced by Michael Albinus
<michael.albinus@gmx.de> back in 2006, so I presume that the macro was
a function back then (or he had had too much fun the previous night).

I'm tempted to just throw out most of that code since M-x compile should
work at least as well for that purpose nowadays.  But if someone who
actually knows about `rcompile' and/or uses the remote compilation
feature of `compile' could DTRT, it'd be better.


        Stefan

Re: Tramp and rcompile

[Michael Albinus]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> The `rcompile' code calls a Tramp macro via `funcall'.
> Obviously, this can't work.  This code was introduced by Michael Albinus
> <michael.albinus@gmx.de> back in 2006, so I presume that the macro was
> a function back then (or he had had too much fun the previous night).

Likely the latter one, I don't remember. But it looks like I've tested
only the not byte-compiled version ...

> I'm tempted to just throw out most of that code since M-x compile should
> work at least as well for that purpose nowadays.

I second this. I have had always a bad feeling, using Tramp *internal*
functions somewhere else.

>         Stefan

Best regards, Michael.

Re: Tramp and rcompile

[Stefan Monnier]

>> The `rcompile' code calls a Tramp macro via `funcall'.
>> Obviously, this can't work.  This code was introduced by Michael Albinus
>> <michael.albinus@gmx.de> back in 2006, so I presume that the macro was
>> a function back then (or he had had too much fun the previous night).

> Likely the latter one, I don't remember.  But it looks like I've tested
> only the not byte-compiled version ...

Going through an explicit `funcall' should ensure that it will fail even
if you don't byte-compile the file.

>> I'm tempted to just throw out most of that code since M-x compile should
>> work at least as well for that purpose nowadays.

> I second this.  I have had always a bad feeling, using Tramp *internal*
> functions somewhere else.

OK, so I'll remove the obviously broken code (it will still use an
internal Tramp function, tho).
I also suggest we move rcompile.el to the `obsolete' directory.


        Stefan