[Scilab-users] unloader.sci & module.quit

Samuel Gougeon sgougeon at free.fr
Tue Feb 15 20:24:03 CET 2022


Le 15/02/2022 à 19:03, Jean-Yves Baudais a écrit :
>> --> clear(string(mylib)(2:$))
> Thanks! It does what I expected.
>
>> Should we actually clear also macros belonging to the library, in the default
>> unloaded.sce script?
>> There is currently an explicit warning in comment: " This does not unregister
>> its macros"
>> I think it would be cleaner. But there may be some reasons to not having done
>> it...
> For my Scilab understanding, what are the reasons?

I don't know. May be none.

To me, there is a bunch of questions / issues about this unloader.sce 
and the ~/etc/module.quit scripts.

The default unloader.sce is generated for instance by tbx_make(..)

  * If the module to unload is an autoloaded ATOMS module, its library
    is protected. Then trying to clear it yields an expected error.
    The default unloader.sce does not take this into account. Yet, it
    should avoid or handle the error without stopping.

  * unloader.sce should clear loaded functions of the library, as you
    suggest.
    Indeed, if it's not wished, it is simple to comment the line in the
    final version.
    Otherwise, if you have asked how to do, it's that the syntax is not
    trivial to find.

  * atomsQuit() : the documentation (and code) of this function are
    prone to discussion:
     1. its page says that it executes .quit files of all loaded ATOMS
        modules, while it calls their unloader.sce script
     2. its page says "/This function is called by
        SCI/etc/scilab.quit/". If it is its only usage, then it's an
        internal, and it should not be documented.
     3. According to its code: if the autoloading mode of the ATOMS
        system is OFF, then atomsQuit() does nothing! That's strange.
        This could mean that it is definitely not aimed to be used out
        of the automatic Scilab startup.
     4. It can't be used to quit some given modules. Only to quit all
        loaded modules. So..
     5. Proposals:
          + Either we keep atomsQuit() as is, and undocument it to make
            it as an internal function
          + Or we remove the condition (3), and we add an input argument
            = vector of modules name to quit.

Any comments?

Samuel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.scilab.org/pipermail/users/attachments/20220215/07f1513d/attachment.htm>


More information about the users mailing list