Greetings, and thanks so much for the update! Glad to hear the fricas.input is working now. I was under the impression that the correct behavior was restored by Waldek some time ago in accounting for GCL's different pathname-type semantics. I would be surprised if the commit you reference were responsible, as it only affects the compiler.
Take care, Grégory Vanuxem <[email protected]> writes: > Hello, > > Le mer. 26 nov. 2025 à 17:58, Grégory Vanuxem <[email protected]> a écrit : >> >> Hello, >> >> Le mer. 26 nov. 2025 à 15:59, Dima Pasechnik <[email protected]> a écrit : >> > >> > This was only a tip of the problem. >> > >> > We would like to have Sage working with GCL-compiled FriCAS, but it leaks >> > messages on things added to the workspace asynchronously, without a way to >> > suppress them via ")set mess". >> > >> > sage: fricas('3') >> > >> > ends with >> > >> > TypeError: An error occurred when FriCAS evaluated '3': >> > Function declaration sageprint : InputForm -> String has been added to >> > workspace. >> > >> > see <https://github.com/sagemath/sage/issues/40569> >> > >> > Is this a FriCAS bug? >> >> I may be wrong but with quick tests on FriCAS built on top of GCL (the >> packaged FriCAS on Debian based system) and FriCAS built with Clozure >> CL or SBCL, the handling of default options at startup and via >> .fricas.input are different. Personally my .fricas.input contains >> ')set message time on', it is not taken into account at startup with >> GCL based FriCAS, contrary to the ones with SBCL or Clozure CL. So I >> would say it's an issue with FriCAS on top of GCL with startup >> options. >> >> Regards, >> >> Greg >> >> PS: I tested packaged FriCAS on a Debian-based system (1.3.11). Also >> 1.3.12 on SBCL or Clozure CL. And a recent version of FriCAS built on >> top of recent GCL. > > After some further investigations, what I pointed out is not related > to SAGE, sorry, but to the official Debian-based gcl27 package. > '.fricas.input' is not processed at startup with the actual official > fricas package. I tested the gcl package (2.6.14 - git: > Version_2_6_15pre18) to build FriCAS and the .fricas.input is > effectively processed. So since there was recently a discussion on > gcl-devel about pathname > (https://mail.gnu.org/archive/html/gcl-devel/2025-11/msg00001.html) I > gave a try to the GCL git version commit: HEAD master Version_2_7_2pre > (Account for possible #\. in pathname-type when compiling) and this > issue is solved, I don't know if this discussion is related to this > issue though. FriCAS built with this version of GCL no longer exhibits > this issue. > > For information. > > Greg -- Camm Maguire [email protected] ========================================================================== "The earth is but one country, and mankind its citizens." -- Baha'u'llah
