Thanks for reviewing the proposal. 

>
> > 1. Remove incompatibilities between the old and new assumptions. 
> > [...] This project aims to correct that by adding a compatibility 
> > layer between the old and the new assumptions.  
>
Is such a compatibility layer needed? 
>  From previous discussion, I have been under the impression that it is 
> not, but then I haven't followed it very closely. 
>
> I thought the shift from old-assumptions to new assumptions was too big a 
change
for previous users to completely deprecate them in one go. So we keep both 
for some time and gradually remove the old ones.
 

> > 2. Since we ultimately want to shift to the new assumptions system. 
> > Make a few core components ( symbol, Add, Mul, etc ) use the new 
> > assumptions system and expand gradually. 
>
> Sounds like a good plan to me. 
> BTW I think this alone could well be enough work for a full GSoC. Making 
> changes of this kind involves not just review, but also writing 
> documentation and unit tests. These can easily double or triple the 
> total amount of work (but are *very* important). 
>
> In retrospect, I now think that I bit off more than I could chew. They 
will take a much longer time than I expected.
Thanks for pointing it out.
 

>  > 3. instead of, 
>  >>> facts  = Q.positive(x) & Q.real(x) & Q.commutative(x) 
>  >>> facts &= Q.positive(y) & Q.real(y) & Q.commutative(y) 
>  >>> facts &= Q.positive(z) & Q.real(z) & Q.commutative(z) 
>  > we could do something as 
>  >>> factset =  Q.positive & Q.real & Q.commutative 
>  >>> facts=factset.apply(x, y, z) 
>
> That's a very cool idea. 
> If requires some changes to the API of Q, and probably some background 
> in higher-order functions. 
> Can you say a bit about how Q's API would look for that? I.e. a list of 
> functions that need to be made (or modified), and what each would do. 
>
> I'm not sure that the new assumptions system is ready for this addition 
> right now. 
>
> If you think so, we could do it as a separate project. I think a few 
changes in Predicate class will do the trick.
 

> Timeline Week-2 
> I suspect that making the old system interpret new assumptions isn't 
> going to work very well. 
> Logic inference algorithms tend to break down if you work with just a 
> subset of a body of known facts. 
>
> Overall, I suspect you're trying to do too much in too little time. 
> I may be wrong :-)  
>
Just my thoughts. I bet others will have other things to say. 
>
> Regards, 
> Jo 
>

I have updated the proposal. (read: reduced some load) 
Thanks
Aaditya M Nair

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sympy.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/sympy/6be25e1f-9af1-4a03-94a4-c4d9b44c98ee%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to