Aleksej Saushev wrote:
> Raymond Toy <raymond.toy at ericsson.com> writes:
>
>
>> Aleksej Saushev wrote:
>>
>>> Raymond Toy <toy.raymond at gmail.com> writes:
>>>
>>>> However, there is no guarantee that different systems will always have
>>>> different filenames, so they'll overwrite each other. Also,
>>>> *maxima-tempdir* defaults to the user's home directory, so fasls from
>>>> maxima subsystems will clutter the users home directory.
>>>>
>>>>
>>> Maybe it is better to default to /tmp or ${TMPDIR}?
>>>
>>>
>> Do you mean *maxima-tempdir* should default to /tmp? I assume there's a
>> good reason for making it the home directory. Perhaps to support windows?
>>
>> Or do you mean put fasls into /tmp/binary-<lisp>? I don't think that's
>> a good idea. We still have the problem with non-unique names, and I
>> certainly don't want to be loading up someone else's fasls.
>>
>> We could do /tmp/<user>/<stuff>, but I think I would want my fasls to
>> survive a reboot.
>>
>
> No, I mean that _temp_-dir should be temporary (in accordance to POLA),
>
POLA?
>
>> (We do have an issue with gcl in that gcl doesn't have
>> ensure-directories-exist, so mk:defsystem can't create the directories
>> if they don't already exist. One possible workaround is to make
>> ensure-directories-exist call mkdir -p to create the directories. This
>> will probably work on all unix systems we have today, but I don't know
>> how to do that on windows.)
>>
I'm mistaken. mk:defsys will create the directories with gcl. (I see
that I added that.)
>
>
>>> Specifically, note the first rule for .OBJDIR
>>>
>>>
>> I suppose we could do that. I don't want to make it overly complicated.
>>
>
> That wasn't meant as instruction, it is information about "prior art."
>
> I'm not sure about precise usage pattern for generated files, but if
> they are meant completely intermediate without own value, quite like
> object files, you may follow this way.
>
Unlike .o files from C, fasl files are important in Lisp.
Ray