[Top][All Lists]

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

[Gcl-devel] experimental

From: Camm Maguire
Subject: [Gcl-devel] experimental
Date: Thu, 06 May 2010 12:55:23 -0400
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.2 (gnu/linux)

Greetings!  I've checked in an experimental branch.  Not for the faint
of heart, but gives some ideas on where I think things should be
going.  A lot of work on function calling and inlining, and a great
deal of simplification.  PCL is not yet ported, but this is not hard.  

As is hopefully clear by now, versions >=2.7 attempt to replace the
sys-proclaims facility with automatic function signature detection.
2.7 was based on the idea that loading functions with incompatible
signature calls could lead to segfaulting code -- therefore gcl would
recompile the offending code automatically if needed.  This proved
IMHO to be too susceptible to cascade.  The governing variable
si::*disable-recompile* is off by default now, and the experimental
branch enables all function calls to proceed correctly regardless of
signature assumptions when compiled, albeit perhaps slowly.
si::*fast-link-warnings* can show problems, and si::do-recompile or
less severe can be run by hand when needed.

To reduce image bloat, si::do-recompile accepts nil as its argument,
which means try to recompile the original source files with the new
signature information, instead of producing overwriting files.  This
runs afoul of pcl, which boots itself in some arcane way still not
completely clear to me.  One cannot recompile let alone load pcl
source files from an image with pcl already installed.  Anyone know a
way around this?

Take care,
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]