Check existence of members/methods

[Nicolas Fleury]

Alex Martelli wrote:
> try: meth = object.setXmlFilename
> except AttributeError: meth = lambda x: setattr(object,'xmlFilename',x)
> meth(currentFillename)
> 
> This doesn't assume that object.xmlFilename must already exist before
> you can set it, which IS implied by your code here quoted -- it just
> seems a slightly weird condition to me.

In my case, a xml parser/saving creating objects corresponding to 
elements and vice-versa, forcing the existence of members before setting 
them makes the code more readable (and avoid a lot of errors).  I agree 
it would be a weird restriction in other situations.

> I personally prefer the try/except/else variant:
> 
> try: meth = object.setXmlFilename
> except AttributeError: object.xmlFilename = x
> else: meth(currentFillename)
> 
> it seems way simpler to me.  However, if you think of objects lacking a
> setter method as weird and exceptional ones, I see why this might seem
> backwards.  Personally, I consider setter methods the anomaly (it's
> exactly to avoid them that we have property...:-) but I do understand
> they're frequently used.  If I often had to fight with objects full of
> getThis, setThat methods I'd wrap them into a generic wrapper with a
> __setattr__ and __getattr__ to be able to use attribute get and set as
> common sense and decency require, e.g, something like....:

Actually, I prefer the "hasattr" solution.  I'm thinking of removing the 
setter functionality from the parser.  If some class needs to do 
specific stuff when a member is updated, as you said, the built-in 
function property can still be used.

Regards,
Nicolas