> Generalizing code for the sake of generalization and at the expense of
> efficiency is probably a bad choice, especially for something so
> pervasive as the category framework, but if you can show that the more
> general code is necessary to do things people actually want, it
> becomes worth evaluating what the slowdowns are and if one can avoid
> the slowdowns, or even accept them if they are unavoidable.
Agreed. Otherwise, it will never stop - there are always more
generalisations which could be done in mathematics. Otherwise, your
additions sound very compelling and nice mathematically :-)

> The 1.5% you are giving is a nice guide, but it doesn't really show
> that there is a slow-down, nor how badly this is going to affect
> existing code. Could you try to identify which things are slowed down?
> All operations? just certain coercions? Perhaps look at the timing of
> separate modules and see which contribute most to that 1.5%.

Even more importantly: try removing your new doctests to show the
slowdown on only the original doctests; otherwise, the number 1,5%
doesn't reveal much (your new tests might just be very slow compared
to the average doctest out there).

Cheers,
Johan

-- 
To post to this group, send an email to [email protected]
To unsubscribe from this group, send an email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Reply via email to