Error creating temp file

[Kevinpgalligan]

Hi,

I tried to run one of the example commands in SLIME -- (solve '(and (or a b) (or a !b c)) :minisat) -- after successfully installing cl-sat.minisat, but encountered the following error:

Subprocess with command "mktemp --tmpdir='/tmp/'  cnf.XXXXXXX"
 exited with error code 1
   [Condition of type UIOP/RUN-PROGRAM:SUBPROCESS-ERROR]

Restarts:
 0: [CONTINUE] IGNORE-ERROR-STATUS
 1: [RETRY] Retry SLIME REPL evaluation request.
 2: [*ABORT] Return to SLIME's top level.
 3: [ABORT] abort thread (#<THREAD "repl-thread" RUNNING {1002F283E3}>)

Backtrace:
  0: (UIOP/RUN-PROGRAM::%CHECK-RESULT 1 :COMMAND "mktemp --tmpdir='/tmp/'  cnf.XXXXXXX" :PROCESS NIL :IGNORE-ERROR-STATUS NIL)
  1: (UIOP/RUN-PROGRAM::%USE-SYSTEM "mktemp --tmpdir='/tmp/'  cnf.XXXXXXX" :OUTPUT (:STRING :STRIPPED T))
  2: ((:METHOD SOLVE (SAT-INSTANCE T)) #<SAT-INSTANCE {1003DF3603}> :MINISAT) [fast-method]
  3: (SB-INT:SIMPLE-EVAL-IN-LEXENV (SOLVE (QUOTE (AND # #)) :MINISAT) #<NULL-LEXENV>)
  4: (EVAL (SOLVE (QUOTE (AND # #)) :MINISAT))

Same error if I run the command through the REPL: (uiop:run-program "mktemp --tmpdir='/tmp/' cnf.XXXXXXX"), which indicates it's not a problem with cl-sat specifically.

If I try to run the command myself from the terminal -- mktemp --tmpdir='/tmp/' cnf.XXXXXXX -- it works fine.

If I run the command through the SBCL interpreter, it succeeds: (uiop:run-program "mktemp --tmpdir='/tmp/' cnf.XXXXXXX")

Any idea what might be the issue? Thanks!

[Kevinpgalligan]

Aaaand if I run the command through the SBCL interpreter, it works just fine. So this seems to be an issue with my REPL / Portacle. Closing this.

[guicho271828]

Still curious to see where it's happening. I don't know what portacle is actually doing, but is it perhaps using a container based infrastructure like docker? Then it's possible that /tmp/ is not mounted in that container

[Kevinpgalligan]

Didn't think of that, good shout. I'll experiment when I'm at home again. If it turns out to be a missing tmp folder in the Portacle environment, I might send a pull request here to quietly create a tmp folder in the with-tmp utility function, if that makes sense. It would make my life easier, at least.

[Kevinpgalligan]

Right, I figured out what's wrong. Portacle is overriding the mktemp command. See here for details: portacle/portacle#129

I suggest changing the call to the mktemp command to avoid the --tmpdir argument, since this approach seems to be more portable. Pull request incoming, let me know if you agree.

[Kevinpgalligan]

I've prepared a fix (see: https://github.com/Kevinpgalligan/cl-sat), and the new version of with-temp works in Portacle, but I've run into some issues while trying to test cl-sat:solve.

Here's what I run, with output:

CL-USER> (asdf:system-source-directory :cl-sat)
#P"/home/kevin/portacle/all/quicklisp/local-projects/cl-sat/"
CL-USER> (asdf:system-source-directory :cl-sat.minisat)
#P"/home/kevin/portacle/all/quicklisp/dists/quicklisp/software/cl-sat.minisat-20190202-git/"
CL-USER> (cl-sat:solve '(and (or a b) (or a !b c)) :minisat)
; cd /tmp/minisat.XXj2S3KK; /home/kevin/portacle/all/quicklisp/dists/quicklisp/software/cl-sat.minisat-20190202-git/minisat/build/release/bin/minisat  /tmp/cnf.Xtb8vHH resultWARNING: for repeatability, setting FPU to use double precision
============================[ Problem Statistics ]=============================
|                                                                             |
|  Number of variables:             6                                         |
|  Number of clauses:               2                                         |
|  Parse time:                   0.00 s                                       |
|  Eliminated clauses:           0.00 Mb                                      |
|  Simplification time:          0.00 s                                       |
|                                                                             |
============================[ Search Statistics ]==============================
| Conflicts |          ORIGINAL         |          LEARNT          | Progress |
|           |    Vars  Clauses Literals |    Limit  Clauses Lit/Cl |          |
===============================================================================
===============================================================================
restarts              : 1
conflicts             : 0              (0 /sec)
decisions             : 1              (0.00 % random) (3425 /sec)
propagations          : 3              (10274 /sec)
conflict literals     : 0              (-nan % deleted)
Memory used           : 6.02 MB
CPU time              : 0.000292 s

SATISFIABLE
[this is where the output stops]

It errors out before printing the variables:

The function CL-SAT.MINISAT::SAT-INSTANCE-VARIABLES is undefined.
   [Condition of type UNDEFINED-FUNCTION]

Restarts:
 0: [CONTINUE] Retry calling CL-SAT.MINISAT::SAT-INSTANCE-VARIABLES.
 1: [USE-VALUE] Call specified function.
 2: [RETURN-VALUE] Return specified values.
 3: [RETURN-NOTHING] Return zero values.
 4: [RETRY] Retry SLIME REPL evaluation request.
 5: [*ABORT] Return to SLIME's top level.
 --more--

Backtrace:
  0: ("undefined function" #<CL-SAT:SAT-INSTANCE {10020A0B63}>)
  1: ((:METHOD CL-SAT:SOLVE (PATHNAME (EQL :MINISAT))) #P"/tmp/cnf.Xtb8vHH" #<unused argument>) [fast-method]
  2: ((:METHOD CL-SAT:SOLVE (CL-SAT:SAT-INSTANCE T)) #<CL-SAT:SAT-INSTANCE {10020A0B63}> :MINISAT) [fast-method]
  3: (SB-INT:SIMPLE-EVAL-IN-LEXENV (CL-SAT:SOLVE (QUOTE (AND # #)) :MINISAT) #<NULL-LEXENV>)
  4: (EVAL (CL-SAT:SOLVE (QUOTE (AND # #)) :MINISAT))
 --more--

Any idea what's wrong? Also, are there any other ways that I should test my change, besides running solve within Portacle? I'm rather new to Common Lisp, so help would be appreciated.

Thanks!

[Kevinpgalligan]

It just occurred to me that the version of cl-sat.mini I've pulled using quicklisp may be incompatible with the most recent version of cl-sat, I'll try again tomorrow with the most up-to-date cl-sat.mini.

[guicho271828]

yes, sat-instance-variables function is gone in the latest. I guess you have an old version of the code somewhere in the system (e.g. local-projects/ ) and mixing them up.

[Kevinpgalligan]

Confirmed that it works in Portacle now. Sent a pull request (see above). I know it's just addressing my specific case and still won't work on all systems, so I understand if you want to do something more thorough here.