On 07/02/2018 12:17, Volker Braun wrote:

On Wednesday, February 7, 2018 at 10:31:43 AM UTC+1, Jeroen Demeyer wrote:

##
Advertising

But what is the "desired implementation"?

Again, surely its "Sage integers" for users of Sage. Anything else is just
dumping a giant turd on your users just because its the lazy thing to do.

Certainly not. The lazy way is to use Python integers and rationals. It
was this way at the begining and I changed it partly because of Sage as
I wanted to avoid useless traductions (at C level) and better C level
integration. This was a non-trivial amount of work in upstream
gmpy2.

There are surely ways this can work. E.g. importing from pplpy.gmp gives
you stuff that uses gmp ints, and importing from pplpy.sage gives you stuff
that uses Sage integers. Or pplpy vs sage_pplpy, etc. Its not impossible to
make it work, it just requires a bit more than the minimal developer effort
to set it up.

I never claimed that it was technically impossible. I said that it was
not desirable to have two non-compatible API's. Having two libraries as
you proposed above is one solution. Though I don't like it to have two
copies of the same code. Another option is that I provide "universal
wrappers" that makes the necessary translations to make you happy.
Another option is to go with this small annoyance of dealing with
explicit conversions.
Vincent
--
You received this message because you are subscribed to the Google Groups
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at https://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.