Re: [PATCH 2/2] gnu: lispf4: Fix the searchpath for SYSATOMS.

[ng0 <ngillmann@runbox.com>]

Ricardo Wurmus <rekado@elephly.net> writes:

> ng0 <ngillmann@runbox.com> writes:
>
>> ng0 <ngillmann@runbox.com> writes:
>>
>>> Hi,
>>>
>>> thanks for trying to help with this.
>>>
>>> Ricardo Wurmus <rekado@elephly.net> writes:
>>>
>>>> ng0 <ngillmann@runbox.com> writes:
>>>>
>>>>> * gnu/packages/lisp.scm (lispf4)[arguments]: Fix the searchpath for SYSATOMS.
>>>>> ---
>>>>>  gnu/packages/lisp.scm | 14 +++++++++++++-
>>>>>  1 file changed, 13 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/gnu/packages/lisp.scm b/gnu/packages/lisp.scm
>>>>> index 7836d48..d66fb8b 100644
>>>>> --- a/gnu/packages/lisp.scm
>>>>> +++ b/gnu/packages/lisp.scm
>>>>> @@ -474,8 +474,20 @@ interface.")
>>>>>           #:phases
>>>>>           (modify-phases %standard-phases
>>>>>             (delete 'configure)
>>>>> +           (add-before 'build 'fix-searchpath
>>>>> +             (lambda _
>>>>> +               (substitute* "Lispf42.f"
>>>>> +                 (("NAME='SYSATOMS'")
>>>>> +                  (string-append "NAME='" (assoc-ref %outputs "out")
>>>>> +                                 "/bin/SYSATOMS'")))
>>>>> +               (substitute* "lispf42.c"
>>>>> +                 (("SYSATOMS")
>>>>> +                  (string-append (assoc-ref %outputs "out") "/bin/SYSATOMS"))
>>>>> +                 ;;(("8+1") "72+8+1"))))
>>>>> +                 (("c_b98_st") "c_b98")
>>>>> +                 (("c_b98.val") ""))))
>>>>>             (replace 'install
>>>>> -            (lambda* (#:key outputs #:allow-other-keys)
>>>>> +             (lambda* (#:key outputs #:allow-other-keys)
>>>>>                (let* ((out (assoc-ref outputs "out"))
>>>>>                       (bin (string-append out "/bin"))
>>>>>                       (doc (string-append (assoc-ref outputs "doc")
>>>>
>>>>
>>>> Here’s an alternative:
>>>>
>>>>            (add-before 'build 'fix-reference-to-SYSATOMS
>>>>              (lambda* (#:key outputs #:allow-other-keys)
>>>>                (substitute* "lispf42.c"
>>>>                  (("#define c_b98.*")
>>>>                   (string-append "#define c_b98 \""
>>>>                                  (assoc-ref outputs "out")
>>>>                                  "/bin/SYSATOMS\"\n")))
>>>>                #t))
>>>
>>> Thanks, I'll try this in combination with the proposed move, but if I
>>> don't succeed with it today I will start working on my priority tasks
>>> and do this in november or whenever I get the chance.
>>>
>>>> This won’t work, however, because during the build phase the lispf4
>>>> executable is called to run a script.  At that point SYSATOMS only
>>>> exists in the current directory.  So with the above phase the build
>>>> would fail.
>>>>
>>>> A hack to solve this would be to first install “SYSATOMS” to the target
>>>> directory and then build.
>>
>> And this fixes at least running it :) New patch incoming soon. This
>> depends on the one which deletes the bundles, can you apply that when
>> you do the one which will I will send in later?
>
> Yes, but have you actually tried using it?  I have and while useless
> stuff like “(LIST 1 2 3)” works, creating own definitions does not.
>
> Here’s an example from the manual:
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~
> _(DE KAR(X) ((SUBR . CAR) X]
> --- Undefined function  
> APPLY
> DE
> --- Reset   
> _
> ~~~~~~~~~~~~~~~~~~~~~~~~~~
>
> Installing BASIC.IMG to $out/bin has no effect.  The thing behaves the
> same when the image is not installed at all.

Hm. Okay, I will look into this as soon as I can dedicate time to it.

> Currently, this package cannot be used and I don’t want to have
> a continuous stream of patches to develop the package in master.  Please
> test this thoroughly on your own and then submit a new patch to fix all
> the problems.

Ok

> Another thing I noticed is that this package comes with a full
> reference of Interlisp, which seems unrelated to this particular
> implementation.  The reference is a whole book and it is not clear under
> what license it is released.
>
> I’m currently tempted to remove the package because in its current state
> it isn’t usable and there doesn’t seem to be anyone using it to have
> noticed that it’s broken.  That’s somewhat unsatisfying.
>
> ~~ Ricardo
>
>

I fully agree, please remove it.
I will re-add it as soon as I have a working copy, stripped off the
bloat it currently carries and solving the non-functional state.
Maybe it's even easier to re-implement in Guile what SIEMENS-INTERLISP
did back then, and drop any work on this completely... For now I'll move
this back to my long work-in-progress bench.