gcl-devel
[Top][All Lists]
Advanced

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

[Gcl-devel] Re: recompilation


From: Camm Maguire
Subject: [Gcl-devel] Re: recompilation
Date: 19 Jul 2006 19:48:33 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2

Greetings!

Here is my suggestion - I don't know how to accomplish this without
user support, i.e. automatically.

(let ((si::*disable-recompile* t))
        (compile-nqthm)
        (load-nqthm))
;either
(si::do-recompile "new_file.lsp")
(load "new_file.o");optional
;or
(si::do-recompile)

(save-nqthm)

Separately, with the gcl of the moment, I am not getting the unused
variable warnings in nqthm.

Separately, have you had a chance to look at the ret mismatch warnings
in nqthm?  Is gcl making the right corrections, is it weak, or is it
plain incorrect?  This is with the default source, with nqthm
asserting its own procls too.

Take care,

Robert Boyer <address@hidden> writes:

> It used to take about 1 minute to compile and build nqthm.  Now it takes
> about four minutes.
> 
> It used to be that we compiled nqthm in one gcl, and then in the second,
> we just loaded up the compiled code (in an instant) and saved.  That's
> what we're still trying to do.
> 
> But now there are 2199 "recompiling" messages in the total build and save
> process, and 741 of them in the save phase alone.  There are only about
> 900 functions in the entire nqthm system, so one could roughly guess that
> everything is being compiled three times.
> 
> This is with nqthm being specially told to do no proclaiming at all.
> 
> Something is a bit odd about this.  But I can't quite put my finger on it.
> 
> It is fantastic what a great job GCL is doing about figuring out the right
> proclaims.  But I somehow feel that things might be still improved,
> user-interface-wise.  It might require some changes by me (the user).
> 
> For example, maybe it would be best if I loaded all the interpreted code
> at once, so the compiler could do an analysis on all the interpreted code
> once and for all.  Just wild speculation.  Maybe DEFUN could do complete
> macroexpansion, and at the same time some proclamation inference.  Or
> maybe I should just shut up.  Maybe compilation time&noise is something I
> should just not think about in the grand scheme of things.
> 
> Bob
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah




reply via email to

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