Le mar. 1 août 2023, 22:00, WaldekHebisch <de...@fricas.math.uni.wroc.pl> a écrit :
> On Tue, Aug 01, 2023 at 08:38:35PM +0200, Grégory Vanuxem wrote: > > Hello, > > > > >From commit f5fafaa on Feb. 11, "Remove old GCL specific code", > > > > > https://github.com/fricas/fricas/commit/f5fafaa29785fb0fd83bd974096e4e316762f31b > > > > src/interp/utils.lisp was modified as this: > > 319-326 > > ============================================== > > (setq |$formattedOutputStream| (|mkOutputConsoleStream|)) > > (fricas-init) > > - #+(or :GCL :poplog) > > + #+:poplog > > (|spad|) > > - #-(or :GCL :poplog) > > + #-:poplog > > (let ((*debugger-hook* > > (lambda (condition previous-handler) > > (spad-system-error-handler condition)) > > (setq |$texOutputStream| (|mkOutputConsoleStream|)) > > ============================================== > > and after, in this code branch (handler-bind ...) is called with > (|spad|). > > > > The problem I would like to highlight here is that, since this change, > > GCL is no longer able to compile this file. Namely, the CLtL1 image > > that, by default GCL use, does not support (handle-bind ...) and > > probably more ((handle-case ..) also for example). > > Yes, currently only ANSI GCL can build FriCAS. We need to handle > erros and ANSI gives a way. In the past we used GCL-specific > method for "traditional GCL", but it no longer works. > Ok. > > So to have this routine defined you have to switch to the ANSI image. > > For that, a GCL_ANSI environnent variable needs to be set to > > something. On Debian-like distributions, at least, this is necessary. > > See below an excerpt of the README.Debian file. > > > > So I would suggest to explain this in the INSTALL.rst file or do > > something about that. > > The best we can do is to reject non-ANSI gcl at configure stage. > I agree, that's better. With a non technical explanation in the INSTALL file I would say. Testing the Common Lisp implementation (config.lisp) can be of help. ATM our gcl support is at experimental stage, things generaly > work but are less polished than other parts of FriCAS. For > example configure accepts gcl versions which have no chance > of working. > > Yes, we should explain that FriCAS build only with ANSI GCL. > Concerning GCL_ANSI environnent variable, that is specific > to GCL distibuted with Debian. If you build GCL from > sources, then you need to configure ANSI image. > I didn't know, I only noticed someone who encountered this issue on the web. > > By the way, I have 'make check'ed FriCAS against SBCL, GNU Common > > Lisp, CLISP, ECL and no problem occurs. For CMU Common Lisp, I need > > more time, I have used a version with a modified startup and I do not > > know for now, it was late. For Clozure CL some issues remain but > > integration was improved and it remains only one issue.Thanks for > > this! > > Concerning errors with GCL: probably the most serious issue > is need to artifically limit memory used by GCL to 2G. > No errors with the latest GCL from Debian (testing I guess) but I saw that in the INSTALL file. __ Greg -- You received this message because you are subscribed to the Google Groups "FriCAS - computer algebra system" group. To unsubscribe from this group and stop receiving emails from it, send an email to fricas-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/fricas-devel/CAHnU2dYpHLO9Gfm0pH2ZEg_PLMwV4H_NkpVGzKGt2FQFCAqs1g%40mail.gmail.com.
