I'm trying to find a version of CL to use under Linux so that i can stop
using Franz CL on Suns, which we will be retiring soon. I've been able
to get my current program (a recurrent neural net simulator) to run
under both GCL and CLISP, however both programs have problems. GCL
munges negative floating point numbers. For instance, (setf a -0.01)
will often result in -0.010000000003. This seems minor, but causes the
neural net simulator to produce very different answers when compared to
the Franz Cl version.
CLISP runs the program just fine, but its dribble does not appear to
work. It only captures commands typed to the console and the result of
those commands, not text that gets printed to the console while the
commands are running. This only happens in my program (which is using
*standard-output*). So far I have not been able to duplicate it using
simple test functions. This is an important feature becasue my batch
script runs 50-75 cases at a time and dribbles all the output to a file
so that it can be automatically analyzed.
A search of FAQS and newsgroups have turned up no mention of these
problems or any more recent fixes than the ones I already have. Has
anyone seen these problems or know where I might get a fix?
---Todd