script question

[Scott David Daniels]

gitulyar wrote:
> On Apr 17, 5:23 pm, Scott David Daniels <Scott.Dani... at Acm.Org> wrote:
>> Marco Mariani wrote:
>>> Piet van Oostrum wrote:
>>>> funclist = [func01, func02, func03, ... ]
>>>> for i in range(1,n):
>>>>     funclist[i]()
>>>> Or myscript.funclist[i]() from another module.
>> ...
>> For example, you could do it like:
>>      funclist = []
>>      def _included(function):
>>          funclist.append(function)
>>          return function
>>
>>      @_included
>>      def func01():
>>      ...
>>      def call_all():
>>          for function in funclist:
>>              function()
>>
>> --Scott David Daniels
>> Scott.Dani... at Acm.Org
> 
> This code does the same as '__all__' should, but anyway it's
> the ugly way, because you have to edit '__all__' every time
> you add/remove new functions to your module and if '__all__'
> is absent your code won't work.
> 
> As for me the best solution is to use 'dir()':
> 
>     for func in dir(myscript):
>         if match(func):
>             getattr(myscript, func)()

Well, I prefer that the names of the functions not matter, (that
renaming a function not change the behavior of a program).  And if
you are not going by name, how do you refactor the functions that
you do want to call (and lift common behavior to a function).
It really does come down to style, and I like to avoid using
function names as the way to discover functions.  I put up with
the naming convention for testing, but I don't like it.  The only
excuse I see is that there are often a huge number of tests, and
there _is_ a convention for such names as defined by the xUnit style.
If we were starting from scratch now, I'd argue for a @test decorator.

--Scott David Daniels
Scott.Daniels at Acm.Org