[Ecls-list] Error reporting to slime for c-inline/redirecting gcc output?

[Matthew Mondor]

On Fri, 19 Mar 2010 11:24:54 +0100
Juan Jose Garcia-Ripoll <juanjose.garciaripoll at googlemail.com> wrote:

> On Thu, Mar 18, 2010 at 7:33 PM, Tobias C. Rittweiler <tcr at freebits.de>wrote:
> 
> > I suggest you look into the ECL source how it spawns the gcc processes,
> > and try to redirect gcc's output into the Lisp world.
> >
> 
> One short term goal is to replace all calls to EXT:SYSTEM with calls to
> EXT:RUN-PROCESS in the compiler. This is tricky because both functions work
> totally different: one takes a single string, the other one needs a list of
> strings, each one with an argument for the compiler or linker. Splitting the
> list has to be done with great care because of potential spaces in
> arguments, etc, etc. Other than this, it would have the great benefit that
> we could filter out and pretty-print the C compiler error messages.

Another interesting thing I noticed is that although CC errors output
to the inferior-lisp emacs buffer, load/linking errors don't (I had to
use the ECL REPL directly when a missing symbol caused load-time
linking failure, and there it was clear).  Is this also a result of
EXT:SYSTEM use?

Thanks,
-- 
Matt