Make the interface to be compatible with GCL-compiled fricas package #41231

cxzhong · 2025-11-25T06:07:03Z

Most people use the Ubuntu or Debian system. But they do not provide a compatible fricas compiled by sbcl. So I fix it with gcl's fricas compatibility.
Fix #40569

📝 Checklist

The title is concise and informative.
The description explains in detail what this PR is about.
I have linked a relevant issue or discussion.
I have created tests covering the changes.
I have updated the documentation and checked the documentation preview.

⌛ Dependencies

cxzhong · 2025-11-25T06:33:19Z

@whoami730 I fixed it now. Can you help me test with sbcl fricas or sage's fricas?

cxzhong · 2025-11-25T06:45:01Z

Also see https://ask.sagemath.org/question/81900/how-to-interact-with-fricas-when-installed-sage-via-conda/

dimpase · 2025-11-25T15:50:37Z

The PR title is misleading. It's clearly about GCL-compiled FriCAS, nothing to do with Ubuntu.
By the way, as usual, Ubuntu just uses the Debian package.

dimpase · 2025-11-25T16:17:23Z

There are much better ways to detect the underlying lisp, e.g. by looking at the output of fricas --version:

$ fricas --version
FriCAS 1.3.12
based on sbcl 2.5.8

I asked on https://groups.google.com/g/fricas-devel/c/a2aqmO58pA4/m/hDgfRlqVBgAJ whether there are better ways.

cxzhong · 2025-11-25T16:18:21Z

The PR title is misleading. It's clearly about GCL-compiled FriCAS, nothing to do with Ubuntu. By the way, as usual, Ubuntu just uses the Debian package.

I just ask them. They said I should raise an issue in Debian community

dimpase · 2025-11-25T16:29:19Z

To get the Lisp type, use Lisp!

sage: fricas.eval(")lisp (lisp-implementation-type)")
'\nValue = "SBCL"'

dimpase

simplify Lisp detection

cxzhong · 2025-11-25T16:31:24Z

To get the Lisp type, use Lisp!

sage: fricas.eval(")lisp (lisp-implementation-type)")
'\nValue = "SBCL"'

This is my implementation. and for gcl compiled fricas, it also should let the buffer clear before we get this

cxzhong · 2025-11-25T16:32:20Z

src/sage/interfaces/fricas.py

+                E.expect(self._prompt, timeout=0.1)
+            except pexpect.TIMEOUT:
+                break
+        E.sendline(')lisp (lisp-implementation-type)')


@dimpase in this line I ask that lisp-implementation-type

And there are other message like

this Function declaration sageprint : InputForm -> String has been added to workspace.

When the fricas running.
So we need to clean this.

# GCL-based FriCAS may output function declaration messages # asynchronously after the prompt appears.

Maybe there is a FriCAS way to suppress these?
Please ask on the FriCAS forum, before starting to mess around with all these timeouts, which are really a weapon of last resort.

Yes, we need to ask this

This function is added when init fricas. But sbcl version of fricas does not output that.

dimpase · 2025-11-25T16:34:53Z

src/sage/interfaces/fricas.py

+        E = self._expect
+        # First, consume any buffered output from previous commands
+        # to ensure we get a clean response to the lisp command
+        while True:


What's going on in this loop then?

I do that for all other messages is clean. So we can run this list-umplementation-type safely

Updated command execution for FriCAS to use stdin for quitting instead of -eval to avoid misinterpretation by GCL-compiled FriCAS.

cxzhong · 2025-11-25T17:45:28Z

@dimpase Can you help me test the new change can work on sbcl fricas?

Reduce timeout for subprocess call to fricas from 30 to 1 second.

Refactor subprocess call to avoid timeout handling.

cxzhong · 2025-11-25T18:12:04Z

I have completed test, it is OK for sbcl version

dimpase · 2025-11-25T21:42:22Z

I don't understand where the types messages are coming from. FRICAS_INIT_CODE is a variable which contains the init code for the Sage FriCAS session, which has )set messages type off.

If you see types printed it is a bug

cxzhong · 2025-11-26T08:50:19Z

I think it is a problem caused in GCL's fricas

(sage-dev) zhongcx@LAPTOP-A6042VHQ:~/sage$ cat << 'EOF' | fricas -nosman 2>&1
> )set functions compile on
> )set message any          on
> )set message autoload     off
> )set message bottomup     off
> )set message dropmap      off
> )set message expose       off
> )set message file         off
> )set message frame        off
> )set message highlighting off
> )set message instant      off
> )set message insteach     off
> )set message interponly   off
> )set message prompt       step
> )set message selection    off
> )set message set          off
> )set message startup      on
> )set message storage      off
> )set message testing      off
> )set message time         off
> )set message type         off
> )set message void         off
> )set output length 80
> )lisp (setf |$ioHook| (lambda (x &optional args) (when (member x '(|startAlgebraOutput| |endOfAlgebraOutput| |startKeyedMsg| |endOfKeyedMsg|)) (prin1 x) (princ #\Newline))))
> )lisp (setf |$IOindex| NIL)
> sageprint(x:InputForm):String == (atom? x => (float? x => return float(x)::String; integer? x => return integer(x)::String; string? x => return concat(["_"", string(x)::String, "_""])$String; symbol? x => return string(symbol(x))); S: List String := [sageprint y for y in destruct x]; R: String := new(1 + reduce(_+, [1 + #(s)$String for s in S], 0), space()$Character); copyInto!(R, "(", 1); i := 2; for s in S repeat(copyInto!(R, s, i); i := i + 1 + #(s)$String); copyInto!(R, ")", i-1); return R)
> x^2+1
> EOF
openServer result -2
                       FriCAS Computer Algebra System 
                 Version: FriCAS 1.3.11 built with gcl 2.7.1
                   Timestamp: Wed May 14 17:20:17 UTC 2025
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------
 
(1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> 
Value = #<function 0000000001d06990>
(1) -> 
Value = NIL
(NIL) -> |startKeyedMsg|
   Function declaration sageprint : InputForm -> String has been added to 
      workspace.
|endOfKeyedMsg|
(NIL) -> 
|startAlgebraOutput|
    2
   x  + 1
|endOfAlgebraOutput|

As you see, I turn off this. But it still raise a NIL block output and value = NIL block. And for gcl compiled FriCAS messages can not close. I think it is a bug for upstream fix. we can use this fix it. And it will cause 0.2 s slow down when init the FriCAS

cxzhong · 2025-11-26T08:54:52Z

I also test the set can proper deal with fricas

(sage-dev) zhongcx@LAPTOP-A6042VHQ:~/sage$ cat << 'EOF' | fricas -nosman 2>&1
> )set message startup off
> )set message prompt silent
> )lisp (setf |$ioHook| (lambda (x &optional args) (when (member x '(|startAlgebraOutput| |endOfAlgebraOutput| |startKeyedMsg| |endOfKeyedMsg|)) (prin1 x) (princ #\Newline))))
> )lisp (setf |$IOindex| NIL)
> sageprint(x:InputForm):String == (atom? x => (float? x => return float(x)::String; integer? x => return integer(x)::String; string? x => return concat(["_"", string(x)::String, "_""])$String; symbol? x => return string(symbol(x))); S: List String := [sageprint y for y in destruct x]; R: String := new(1 + reduce(_+, [1 + #(s)$String for s in S], 0), space()$Character); copyInto!(R, "(", 1); i := 2; for s in S repeat(copyInto!(R, s, i); i := i + 1 + #(s)$String); copyInto!(R, ")", i-1); return R)
> x^2+1
> EOF
openServer result -2
                       FriCAS Computer Algebra System 
                 Version: FriCAS 1.3.11 built with gcl 2.7.1
                   Timestamp: Wed May 14 17:20:17 UTC 2025
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------
 
(1) -> (1) -> ---------------------------- The prompt Option ----------------------------

 Description: set type of input prompt to display

 The prompt option may be followed by any one of the following:

    none
    frame
    plain
 -> step 
    verbose

 The current setting is indicated within the list.


 Your value silent is not among the valid choices.

(1) -> 
Value = #<function 0000000001d06960>
(1) -> 
Value = NIL
(NIL) -> |startKeyedMsg|
   Function declaration sageprint : InputForm -> String has been added 
      to workspace.
|endOfKeyedMsg|
|startKeyedMsg|
                                                                   Type: Void
|endOfKeyedMsg|
(NIL) -> 
|startAlgebraOutput|
    2
   x  + 1
|endOfAlgebraOutput|
|startKeyedMsg|
                                                    Type: Polynomial(Integer)
|endOfKeyedMsg|

mantepse · 2025-11-26T09:11:47Z

In my opinion it is an absolute no-go to special case gcl.

The interface to fricas is fragile, mostly because we use communication via strings. Apparently this is hard to change, nobody has done it in many years. Passing very long input and receiving long output may lead to failure, which can be a real pain, because you would usually use FriCAS for computationally expensive stuff.

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

cxzhong · 2025-11-26T10:04:37Z

In my opinion it is an absolute no-go to special case gcl.

The interface to fricas is fragile, mostly because we use communication via strings. Apparently this is hard to change, nobody has done it in many years. Passing very long input and receiving long output may lead to failure, which can be a real pain, because you would usually use FriCAS for computationally expensive stuff.

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

I think it is no side effects. And it is used in Ubuntu and Debian as system's packages. I think many users may meet this problem

cxzhong · 2025-11-26T10:24:07Z

It seems fricas in sbcl can also have the same input but it works well with sage, so what exactly the root cause?

zhongcx@LAPTOP-A6042VHQ:~/sage$ cat << 'EOF' | fricas -nosman 2>&1
> )set functions compile on
> )set message any          on
> )set message autoload     off
> )set message bottomup     off
> )set message dropmap      off
> )set message expose       off
> )set message file         off
> )set message frame        off
> )set message highlighting off
> )set message instant      off
> )set message insteach     off
> )set message interponly   off
> )set message prompt       step
> )set message selection    off
> )set message set          off
> )set message startup      on
> )set message storage      off
> )set message testing      off
> )set message time         off
> )set message type         off
> )set message void         off
> )set output length 80
> )lisp (setf |$ioHook| (lambda (x &optional args) (when (member x '(|startAlgebraOutput| |endOfAlgebraOutput| |startKeyedMsg| |endOfKeyedMsg|)) (prin1 x) (princ #\Newline))))
> )lisp (setf |$IOindex| NIL)
> sageprint(x:InputForm):String == (atom? x => (float? x => return float(x)::String; integer? x => return integer(x)::String; string? x => return concat(["_"", string(x)::String, "_""])$String; symbol? x => return string(symbol(x))); S: List String := [sageprint y for y in destruct x]; R: String := new(1 + reduce(_+, [1 + #(s)$String for s in S], 0), space()$Character); copyInto!(R, "(", 1); i := 2; for s in S repeat(copyInto!(R, s, i); i := i + 1 + #(s)$String); copyInto!(R, ")", i-1); return R)
> x^2+1
> EOF
Checking for foreign routines
FRICAS="/usr/local/lib/fricas/target/x86_64-conda-linux-gnu"
spad-lib="/usr/local/lib/fricas/target/x86_64-conda-linux-gnu/lib/libspad.so"
foreign routines found
openServer result -2
                       FriCAS Computer Algebra System 
           Version: FriCAS 2025-04-14 built with sbcl 2.5.2.debian
                   Timestamp: Wed Nov 26 02:01:29 CST 2025
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------
 
(1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> (1) -> 
Value = #<INTERPRETED-FUNCTION NIL {100110023B}>
(1) -> 
Value = NIL
(NIL) -> |startKeyedMsg|
   Function declaration sageprint : InputForm -> String has been added to 
      workspace.
|endOfKeyedMsg|
(NIL) -> 
|startAlgebraOutput|
    2
   x  + 1
|endOfAlgebraOutput|

cxzhong · 2025-11-26T12:39:56Z

If you add -eval it like

zhongcx@LAPTOP-A6042VHQ:~/sage$ fricas -nosman -eval ')set functions compile on' -eval ')set message any on' -eval ')set message autoload off' -eval ')set message bottomup off' -eval ')set message dropmap off' -eval ')set message expose off' -eval ')set message file off' -eval ')set message frame off' -eval ')set message highlighting off' -eval ')set message instant off' -eval ')set message insteach off' -eval ')set message interponly off' -eval ')set message prompt step' -eval ')set message selection off' -eval ')set message set off' -eval ')set message startup on' -eval ')set message storage off' -eval ')set message testing off' -eval ')set message time off' -eval ')set message type off' -eval ')set message void off' -eval ')set output length 80' -eval ')lisp (setf |$ioHook| (lambda (x &optional args) (when (member x (quote (|startAlgebraOutput| |endOfAlgebraOutput| |startKeyedMsg| |endOfKeyedMsg|))) (prin1 x) (princ #\Newline))))' -eval ')lisp (setf |$IOindex| NIL)' -eval 'sageprint(x:InputForm):String == (atom? x => (float? x => return float(x)::String; integer? x => return integer(x)::String; string? x => return concat(["_"", string(x)::String, "_""])$String; symbol? x => return string(symbol(x))); S: List String := [sageprint y for y in destruct x]; R: String := new(1 + reduce(_+, [1 + #(s)$String for s in S], 0), space()$Character); copyInto!(R, "(", 1); i := 2; for s in S repeat(copyInto!(R, s, i); i := i + 1 + #(s)$String); copyInto!(R, ")", i-1); return R)' -eval 'x^2+1' -eval ')quit'
Checking for foreign routines
FRICAS="/usr/local/lib/fricas/target/x86_64-conda-linux-gnu"
spad-lib="/usr/local/lib/fricas/target/x86_64-conda-linux-gnu/lib/libspad.so"
foreign routines found
openServer result -2
                       FriCAS Computer Algebra System 
           Version: FriCAS 2025-04-14 built with sbcl 2.5.2.debian
                   Timestamp: Wed Nov 26 02:01:29 CST 2025
-----------------------------------------------------------------------------
   Issue )copyright to view copyright notices.
   Issue )summary for a summary of useful system commands.
   Issue )quit to leave FriCAS and return to shell.
-----------------------------------------------------------------------------
 
Value = #<INTERPRETED-FUNCTION NIL {100110023B}>
Value = NIL
|startKeyedMsg|
   Function declaration sageprint : InputForm -> String has been added to 
      workspace.
|endOfKeyedMsg|

|startAlgebraOutput|
    2
   x  + 1
|endOfAlgebraOutput|

cxzhong · 2025-11-26T13:11:32Z

In my opinion it is an absolute no-go to special case gcl.

The interface to fricas is fragile, mostly because we use communication via strings. Apparently this is hard to change, nobody has done it in many years. Passing very long input and receiving long output may lead to failure, which can be a real pain, because you would usually use FriCAS for computationally expensive stuff.

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

The problem looks like in gcl different behavior. not fricas.

dimpase · 2025-11-26T20:20:20Z

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

Can one monkey-patch FriCAS in a way one can do with Python?

We can disable the use of system - wide GCL Fricas until we have a fix.

cxzhong · 2025-11-26T20:32:09Z

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

Can one monkey-patch FriCAS in a way one can do with Python?

We can disable the use of system - wide GCL Fricas until we have a fix.

But fricas developer also do not know what is the exact issue.

dimpase · 2025-11-26T20:38:15Z

They haven't really looked yet, and it's not something that looks hard to debug.

cxzhong · 2025-11-26T21:17:26Z

It is an upstream bug that gcl prints messages after )set message off, and should be fixed there. I suspect that this should actually be easy to do.

Can one monkey-patch FriCAS in a way one can do with Python?

We can disable the use of system - wide GCL Fricas until we have a fix.

May be I change it to directly throw an error if gcl is detected, then modify the is_present method. Can you describe it to fricas developer? I think they do not understand. In fact, it do not affect the computing part.

Help the interface to be compatible with Ubuntu's fricas

1f63090

github-actions bot added the s: needs review label Nov 25, 2025

cxzhong added 2 commits November 25, 2025 14:18

Update fricas.py

95c7297

fix lint

aef3b32

cxzhong added the c: packages: optional label Nov 25, 2025

cxzhong requested a review from dimpase November 25, 2025 06:33

Improve speed in ecl's fricas

8b9c590

cxzhong changed the title ~~Help the interface to be compatible with Ubuntu's fricas~~ Make the interface to be compatible with Ubuntu's fricas package Nov 25, 2025

Remove unnecessary error check

fc859ef

cxzhong changed the title ~~Make the interface to be compatible with Ubuntu's fricas package~~ Make the interface to be compatible with GCL-compiled fricas package Nov 25, 2025

dimpase suggested changes Nov 25, 2025

View reviewed changes

cxzhong commented Nov 25, 2025

View reviewed changes

dimpase reviewed Nov 25, 2025

View reviewed changes

cxzhong added 3 commits November 26, 2025 01:36

fix a synatx issue in gcl fricas

c639418

revert

1a501f7

Fix FriCAS command execution for proper quitting

5fb0b8e

Updated command execution for FriCAS to use stdin for quitting instead of -eval to avoid misinterpretation by GCL-compiled FriCAS.

cxzhong added 2 commits November 26, 2025 01:53

Decrease timeout for fricas subprocess command

396ce08

Reduce timeout for subprocess call to fricas from 30 to 1 second.

Refactor subprocess command execution in fricas.py

a95b5f8

Refactor subprocess call to avoid timeout handling.

cxzhong mentioned this pull request Nov 26, 2025

Can not close the function declare message with )set message in gcl's fricas fricas/fricas#206

Closed

Uh oh!

Make the interface to be compatible with GCL-compiled fricas package #41231

Are you sure you want to change the base?

Make the interface to be compatible with GCL-compiled fricas package #41231

Conversation

cxzhong commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📝 Checklist

⌛ Dependencies

Uh oh!

cxzhong commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cxzhong commented Nov 25, 2025

Uh oh!

dimpase commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dimpase commented Nov 25, 2025

Uh oh!

cxzhong commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dimpase commented Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dimpase left a comment

Choose a reason for hiding this comment

Uh oh!

cxzhong commented Nov 25, 2025

Uh oh!

cxzhong Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

cxzhong Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dimpase Nov 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cxzhong Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

cxzhong Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

dimpase Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

cxzhong Nov 25, 2025

Choose a reason for hiding this comment

Uh oh!

cxzhong commented Nov 25, 2025

Uh oh!

cxzhong commented Nov 25, 2025

Uh oh!

dimpase commented Nov 25, 2025

Uh oh!

cxzhong commented Nov 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cxzhong commented Nov 26, 2025

Uh oh!

mantepse commented Nov 26, 2025

Uh oh!

cxzhong commented Nov 26, 2025

Uh oh!

cxzhong commented Nov 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cxzhong commented Nov 26, 2025

Uh oh!

cxzhong commented Nov 26, 2025

Uh oh!

dimpase commented Nov 26, 2025

Uh oh!

cxzhong commented Nov 26, 2025

Uh oh!

cxzhong commented Nov 25, 2025 •

edited

Loading

cxzhong commented Nov 25, 2025 •

edited

Loading

dimpase commented Nov 25, 2025 •

edited

Loading

cxzhong commented Nov 25, 2025 •

edited

Loading

dimpase commented Nov 25, 2025 •

edited

Loading

cxzhong Nov 25, 2025 •

edited

Loading

dimpase Nov 25, 2025 •

edited

Loading

cxzhong commented Nov 26, 2025 •

edited

Loading

cxzhong commented Nov 26, 2025 •

edited

Loading

cxzhong commented Nov 26, 2025 •

edited

Loading