claw Installation via Quicklisp for SBCL causes `SB-KERNEL::CONTROL-STACK-EXHAUSTED`

Hello, I was attempting to install your other package, bodge-nuklear and I ran into this interesting issue.

INFO: Control stack guard page unprotected
[1/3] System 'bodge-nuklear' found. Loading the system..Control stack guard page temporarily disabled: proceed with caution
Unhandled SB-KERNEL::CONTROL-STACK-EXHAUSTED in thread #<SB-THREAD:THREAD "main thread" RUNNING
                                                          {1001548173}>:
  Control stack exhausted (no more space for function call frames).
This is probably due to heavily nested or infinitely recursive function
calls, or a tail call that SBCL cannot or has not optimized away.

PROCEED WITH CAUTION.

Backtrace for: #<SB-THREAD:THREAD "main thread" RUNNING {1001548173}>
0: (SB-DEBUG::DEBUGGER-DISABLED-HOOK #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10056A6253}> #<unused argument> :QUIT T)
1: (SB-DEBUG::RUN-HOOK SB-EXT:*INVOKE-DEBUGGER-HOOK* #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10056A6253}>)
2: (INVOKE-DEBUGGER #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10056A6253}>)
3: (ERROR SB-KERNEL::CONTROL-STACK-EXHAUSTED)
4: (SB-KERNEL::CONTROL-STACK-EXHAUSTED-ERROR)
5: ("foreign function: call_into_lisp")
6: ("foreign function: post_signal_tramp")
7: ("foreign function: #x7FCAA500D161")
8: (SB-UNIX:UNIX-REALPATH "/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd")
9: (SB-IMPL::%QUERY-FILE-SYSTEM #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd" :TRUENAME NIL)
10: (SB-IMPL::QUERY-FILE-SYSTEM #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd" :TRUENAME NIL)
11: (PROBE-FILE #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd")
12: (UIOP/FILESYSTEM:PROBE-FILE* #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd" :TRUENAME T)
13: (UIOP/FILESYSTEM:TRUENAMIZE #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd")
14: (ASDF/SESSION:NORMALIZE-NAMESTRING #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd")
15: (ASDF/SESSION:GET-FILE-STAMP #P"/home/my_user/.roswell/lisp/quicklisp/dists/quicklisp/software/cffi_0.24.1/cffi.asd")
16: ((:METHOD ASDF/ACTION:COMPUTE-ACTION-STAMP (T ASDF/OPERATION:OPERATION ASDF/COMPONENT:COMPONENT)) #<ASDF/PLAN:SEQUENTIAL-PLAN {10056A5CF3}> #<ASDF/FIND-SYSTEM:DEFINE-OP > #<ASDF/SYSTEM:SYSTEM "cffi"> :JUST-DONE NIL) [fast-method]
17: (ASDF/FIND-SYSTEM::DEFINITION-DEPENDENCIES-UP-TO-DATE-P #<ASDF/SYSTEM:SYSTEM "cffi">)
18: ((LAMBDA NIL :IN ASDF/SYSTEM:FIND-SYSTEM))
19: (ASDF/SESSION:CONSULT-ASDF-CACHE (ASDF/SYSTEM:FIND-SYSTEM "cffi") #<FUNCTION (LAMBDA NIL :IN ASDF/SYSTEM:FIND-SYSTEM) {100569D64B}>)
20: (ASDF/SESSION:CALL-WITH-ASDF-SESSION #<FUNCTION (LAMBDA NIL :IN ASDF/SYSTEM:FIND-SYSTEM) {100569D64B}> :OVERRIDE NIL :KEY (ASDF/SYSTEM:FIND-SYSTEM "cffi") :OVERRIDE-CACHE NIL :OVERRIDE-FORCING NIL)
21: ((LABELS QUICKLISP-CLIENT::RECURSE :IN QUICKLISP-CLIENT::COMPUTE-LOAD-STRATEGY) "cffi")
22: ((LABELS QUICKLISP-CLIENT::RECURSE :IN QUICKLISP-CLIENT::COMPUTE-LOAD-STRATEGY) "claw")
23: ((LABELS QUICKLISP-CLIENT::RECURSE :IN QUICKLISP-CLIENT::COMPUTE-LOAD-STRATEGY) "claw-utils")
24: ((LABELS QUICKLISP-CLIENT::RECURSE :IN QUICKLISP-CLIENT::COMPUTE-LOAD-STRATEGY) "claw/util")
25: ((LABELS QUICKLISP-CLIENT::RECURSE :IN QUICKLISP-CLIENT::COMPUTE-LOAD-STRATEGY) "claw/wrapper")

It seems there's something in relation to claw. Or maybe CFFI.

I'm using SBCL, version 2.1.9.

Oct 03 '21 02:10 alecStewart1

Maybe this is a bit more helpful.

INFO: Control stack guard page unprotected
[1/3] System 'bodge-nuklear' found. Loading the system..Control stack guard page temporarily disabled: proceed with caution
Unhandled SB-KERNEL::CONTROL-STACK-EXHAUSTED in thread #<SB-THREAD:THREAD "main thread" RUNNING
                                                          {1001548183}>:
  Control stack exhausted (no more space for function call frames).
This is probably due to heavily nested or infinitely recursive function
calls, or a tail call that SBCL cannot or has not optimized away.

PROCEED WITH CAUTION.

Backtrace for: #<SB-THREAD:THREAD "main thread" RUNNING {1001548183}>
0: (SB-DEBUG::DEBUGGER-DISABLED-HOOK #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10042DAD73}> #<unused argument> :QUIT T)
1: (SB-DEBUG::RUN-HOOK SB-EXT:*INVOKE-DEBUGGER-HOOK* #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10042DAD73}>)
2: (INVOKE-DEBUGGER #<SB-KERNEL::CONTROL-STACK-EXHAUSTED {10042DAD73}>)
3: (ERROR SB-KERNEL::CONTROL-STACK-EXHAUSTED)
4: (SB-KERNEL::CONTROL-STACK-EXHAUSTED-ERROR)
5: ("foreign function: call_into_lisp")
6: ("foreign function: post_signal_tramp")
7: ("foreign function: #x7F502875043D")

Er...maybe. I don't think it'd be sane to try and function the foreign function #x7F502875043D.

Oct 03 '21 03:10 alecStewart1

Can you load CFFI only? Does it errors out in the same way?

(asdf:load-system :cffi)

Oct 03 '21 08:10 borodust

I can in fact. The bit you gave returns T.

Oct 11 '21 02:10 alecStewart1

Just as a comment, I was able to install it correctly. On Windows 10, SBCL 2.1.9

Oct 11 '21 07:10 apemangr

Still having issues. I'm using sbcl with roswell, would it have something to do with that? I image not, but I'm just spitballing at this point.

Still getting

INFO: Control stack guard page reprotected
INFO: Control stack guard page unprotected
Control stack guard page temporarily disabled: proceed with caution

debugger invoked on a SB-KERNEL::CONTROL-STACK-EXHAUSTED in thread
#<THREAD "main thread" RUNNING {1001BB0183}>:
  Control stack exhausted (no more space for function call frames).
This is probably due to heavily nested or infinitely recursive function
calls, or a tail call that SBCL cannot or has not optimized away.

PROCEED WITH CAUTION.

Oct 27 '21 20:10 alecStewart1

I really can't guess what is wrong here. Just in case, can you load bodge-nuklear from bare SBCL? Like, start SBCL from terminal directly and try quickloading from there. Also, what does (ql:where-is-system :bodge-nuklear) report? Does (ql:update-all-dists) help?

Oct 27 '21 23:10 borodust

Apologies for just getting back to this. I'll check sometime later this week if I have any issues on newer versions of SBCL.

Oct 10 '22 17:10 alecStewart1

Having this problem for some time (year or two; Win10 / SBCL x64 2.2.x - 2.3.x via Roswell). Did not went away even after (unrelated) full Roswell(etc.) wipe and reinstall.

Only occurs for bodge AFAIK
update-all-dists doesn't help
I have quicklisp, ultralisp, borodust & shirakumo (the last one for just a couple of months) dists "enabled" for quicklisp (IDK whether this is relevant).
plain sbcl doesn't even know about quicklisp (I'd have to set it up, and probably tolerate or delete the second set of locations it would create)
- but if it would really help, I'd bite the bullet
but roswell has everything under \Users\<username>\.roswell
while fasl cache is in APPDATA (long paths have to work, since it's pretty deep on its own & then it mirrors the long paths under . roswell, never was a problem)

EDIT: But quickloading :nuklear-blob went fine. Should I try to trace some package or something to see on what it loops ?

Sep 05 '23 09:09 Mr-Dispatch

Yup, please, check if it works w/o roswell being involved. I don't use roswell and I don't explicitly maintain compatibility with it. But if you will be able to pinpoint the issue, depending on what it is, I might incorporate required fixes.

Sep 05 '23 11:09 borodust

Ok, installing fresh quicklisp under SBCL 2.3.6 (compiled - but not executed - through Roswell), the following almost works:

* (ql-dist:install-dist "http://bodge.borodust.org/dist/org.borodust.bodge.txt")
...
* (ql:quickload '(nuklear-blob bodge-nuklear))
...
* (nuklear.example:run)
#<NUKLEAR.EXAMPLE::NUKLEAR-APP {1007FF00D3}>
* Assertion failed: atlas->font_num, file nuklear/nuklear.h, line 13741

What can I do to help debug Roswell -- my precioussss, glum :)

Sep 05 '23 11:09 Mr-Dispatch

@Mr-Dispatch This new issue you've got is unrelated to claw or roswell. This is an assertion failing inside Nuklear native code. Nuklear is notoriously hard to debug - I suggest trying some other library. Maybe Shinmera's Alloy.

As for roswell, I'm no help here, unfortunately. No idea what's up with it.

Sep 05 '23 12:09 borodust

DUH it's unrelated, just wanted to share (thought maybe the *-blob package is outdated).

As for figuring out what's wrong under roswell, could you give me some clues as to how to try to debug it ? I thought maybe tracing quicklisp or some other package might give some insight. Sysinternals tools provide something like STRACE, but that's even more verbose and low-level a.f. (not much of a CL guru, I'm afraid :-/ )

EDIT: I'm creating Roswell issue for this, they should know best.

Sep 05 '23 12:09 Mr-Dispatch