The cypari2 related files in sage/libs/pari/ are rather minimal
$ wc -l sage/libs/pari/*
4 pari/all.py
13 pari/convert_flint.pxd
158 pari/convert_flint.pyx
13 pari/convert_gmp.pxd
209 pari/convert_gmp.pyx
3 pari/convert_sage.pxd
296 pari/convert_sage.pyx
217 pari/__init__.py
3 pari/misc.pxd
25 pari/misc.pyx
1798 pari/tests.py
2739 total
As you can see, 65% of it are leftover tests from the old time
when we had the Cython interface to PARI inside sage. The remaining
941 lines of code are interfaces (with gmp, flint and Sage).
Le 07/02/2020 à 11:04, Han Frederic a écrit :
First, thank you for this discussion.
About the sage pip install project. The problem is that I don't know at all
what will be available.
But if you are done with sage's pyx files in src/sage/libs then it should
be similar with giacpy_sage.
It is indeed rare for an spkg to be linked to the sage library and
giacpy_sage is
more similar to the cython files in src/sage/libs than to an external
package.
The original ticket https://trac.sagemath.org/ticket/15226 six years ago
about giacpy_sage
was a sage patch. It just ended with no answer but with the current
giacpy_sage
it is still easy to put 4 files of giacpy_sage in sage's source tree
src/sage/libs, adjust
src/module_list.py and let sage's setup.py build giacpy_sage. Then you
don't need an external package at all.
This is related to Vincent's question. There are many cython files in
sage/libs related
to cypari2. Currently giacpy_sage have none. We need one of the following
(to not lose any features) either giacpy_sage is an external package linked
to the sage library
or a part of giacpy_sage (or all of it, because it is just few files) is
inside sage. Any of these
solutions is fine for me. But being in between is more work and my skills
and time are limited.
The python package have a different name: giacpy. This was the only request
I had during the previous vote where all the focus turned on giac and where
giacpy_sage was forgotten.
(cf https://groups.google.com/d/topic/sage-devel/HI0uZra5_iI/discussion)
I don't see any conflict in having both. Sage could load the python package
but it is just less interesting than loading giacpy_sage.
A minimal python is able to use the giacpy package obtain a full computer
algebra system. I don't want to break this with the in between solution (1
python package and files in sage).
It is important for me that a window user with no development tools looks
to be able
to do with a minimal python installation:
python -m pip install giacpy
and have it. It was very difficult for me to provides these binaries, and I
don't want to break this.
giac and sage are both computer algebra systems, so there have have many
interactions that are meaningless in a standalone python.
Another remark: giac is a standard package and giacpy_sage test it more.
For instance this ticket
https://trac.sagemath.org/ticket/27306
was a giac bug in a new version of giac. It was not related to giacpy_sage
and I guess that it would have been detected during the upgrade of the giac
package if giacpy_sage was standard, while it was detected after.
Best
Frederic
--
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 view this discussion on the web visit
https://groups.google.com/d/msgid/sage-devel/4c16560a-9536-14b4-95de-fb4ac20aedbe%40gmail.com.
