Hi Fabian!

On Sat, Feb 21, 2009 at 2:50 AM, Fabian Seoane <[email protected]> wrote:
> Hi all. As part of my work in the assumptions branch[1], I'm faced to the
> problem of finding a new syntax for the expr.is_* property.
>
> It seems that last time we talked about it, we used the following syntax for
> querying an element (although we did not discuss it)
>
> IsPositive(x)
> IsElement(x)
> etc.
>
> This is ok, but I find it cumbersome to have so many methods (at least one
> for each assumption we support). Alternatively, my implementation
> proposes another solution based on the query method. You can see it's
> implementation here [2]. Examples would look like:
>
> query(x, 'positive')
> query(x, 'real')
>
> It works the following way:
>     Each query key ('positive', 'real') defines a QueryHandler that is
> responsible for handling the resolution of the current query. Then
> query_dict maps each assumption
> to it's handler. Of course we could define shortcuts:
>
> def IsPositive(x):
>     return query(x, 'positive')
>
> The nice thing here is extensibility: If any third-party module would like
> to add new query keys (query(x, 'my_key')), just append the appropriate
> handler to query_dict (all this is better understood looking at [2]).
>
> Of course, Is* syntax can also be made extensible by subclassing, so it all
> comes at the end to a matter of consistency.
>
> Any comments? Which syntax do you prefer ?

Thanks for working on it. I think your approach is fine. We can always
define such methods, or properties if needed (like is_positive).
Important is, that the assumptions will be detached from the
expression, because then we can merge the cython core in.

Ondrej

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to [email protected]
For more options, visit this group at http://groups.google.com/group/sympy?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to