rest2web-develop

[Rest2web-develop] ConfigObj Commit

[Michael F. <mi...@pc...>]

Hello Nicola,

Just committed more changes to ConfigObj.

Renamed private attributes with a single underscore prefix.
Changes to interpolation - exceeding recursion depth,
    or specifying a missing value, now raise errors.
Changes for Python 2.2 compatibility.
    (changed boolean tests - removed ``is True`` and ``is False``)
Added test for duplicate section and member (and fixed bug)


I *might* have some time during the day to look at the validate issues.
(type checking and type conversion).
First of all I'll add stringify and transforming validation.
This will do the type conversion when you write files and when you validate.

Can you think of any config information you might want to store where
the *type* of the value will be ambiguous ?
For example any config options where you might want a string or a number ?

The key thing is having a good standard set of validate functions that
take care of all the common cases .
Basic types :

    float
    decimal
    integer
    string
    date/time
    boolean   
    None
    lists of any of these

All these basic sorts it would be *very* simple to detect and
automatically convert.

Mark Andrews also favoured a regex approach as well - for things like
post codes (zip codes), phone numbers, IP addresses. Obviously this will
only check that the value matches the spec - not do any transformation.

Can you tell that I'm procrastinating writing the ``writein`` method...

Best Regards,

Fuzzy
http://www.voidspace.org.uk/python

Re: [Rest2web-develop] ConfigObj Commit

[Michael F. <mi...@pc...>]

Nicola Larosa wrote:
[snip..]
>>I *might* have some time during the day to look at the validate issues.
>>(type checking and type conversion).
>>First of all I'll add stringify and transforming validation.
>>This will do the type conversion when you write files and when you validate.
>>
>>Can you think of any config information you might want to store where
>>the *type* of the value will be ambiguous ?
>>For example any config options where you might want a string or a number ?
> 
> 
> A phone number, or zip code, could be ambiguous.
> 

But seeing as you *specify* what you want, in these cases you probably
want to leave it as a string. See below about the fact that I'm not
*really* suggesting *automatic* type conversion.

As you say in the other email - it's a bit like duck typing, the
conversion is only done if you actually need (specify) it.

> 
> 
>>The key thing is having a good standard set of validate functions that
>>take care of all the common cases .
>>Basic types :
>>
>>    float
>>    decimal
>>    integer
>>    string
>>    date/time
>>    boolean   
>>    None
>>    lists of any of these
>>
>>All these basic sorts it would be *very* simple to detect and
>>automatically convert.
> 
> 
> Simple, yes. Not *very* so in the date/time case, I fear.
> 

Sure - there are modules that will attempt to cover the most common ways
of specifying dates though.

I *don't* think I'm really suggesting *automatic* type conversion - you
still use the ``validate`` method and *specify* in your configspec what
conversion/check to apply to each value.

I can write enough general case functions that will cover all the above
examples - and anything else that you will need.

According to YAGNI I shouldn't write anything you don't actually want -
so give me your use cases and I'll work to those..... The above ones are
obvious though I guess.....

Best Regards,

Fuzzball
http://www.voidspace.org.uk/python

Re: [Rest2web-develop] ConfigObj Commit

[Nicola L. <ni...@te...>]

> But seeing as you *specify* what you want, in these cases you probably
> want to leave it as a string. See below about the fact that I'm not
> *really* suggesting *automatic* type conversion.
> 
> As you say in the other email - it's a bit like duck typing, the
> conversion is only done if you actually need (specify) it.

And that's as it should be. :-)


> I can write enough general case functions that will cover all the above
> examples - and anything else that you will need.

I can write something too sometimes, you know, even if it doesn't look like
that. ;-)


> According to YAGNI I shouldn't write anything you don't actually want -
> so give me your use cases and I'll work to those..... The above ones are
> obvious though I guess.....

You shouldn't write anything that *you* don't actually want. I'll take care
of my use cases in due course, your hands are full enough already. :-)

-- 
Nicola Larosa - ni...@te...

Adding things just because you can leads to monstrosities like Common LISP,
PL/I, Algol 68 and Perl 6. Adding features only when they add functionality
(or better yet, by removing restrictions) leads to jewels like Python,
Scheme and Eiffel. -- Mike Meyer, comp.lang.python, April 2005