move integer ring to the new coercion model

[robertwb]

A couple of bugfixes are included as well.

CC: @aghitza @malb

Component: coercion

Issue created by migration from https://trac.sagemath.org/ticket/4058

[robertwb]

Attachment: 4058-integer-coerce.patch.gz

[aghitza]

comment:1

I'm getting an error trying to apply this patch to a fresh 3.1.2:

patching file sage/interfaces/expect.py
Hunk #1 FAILED at 1385
1 out of 1 hunk FAILED -- saving rejects to file sage/interfaces/expect.py.rej
patch failed, unable to continue (try -v)
patch failed, rejects left in working dir

[robertwb]

comment:2

I'll rebase this as soon as I get 3.1.2.

[robertwb]

Attachment: 4058-integer-coerce.2.patch.gz

[robertwb]

comment:4

Refreshed the patch so it applies cleanly to 3.1.2.

[mwhansen]

comment:5

Looks good to me.

[sagetrac-mabshoff]

comment:6

One thing I notice with this patch is that sr.py now takes around 650 seconds instead of 450 or so:

sage -t -long devel/sage/sage/crypto/mq/sr.py
         [655.1 s]

I am still merging the patch, but can we get this issue fixed next?

Cheers,

Michael

[sagetrac-mabshoff]

comment:7

Merged 4058-integer-coerce.2.patch in Sage 3.1.3.alpha1

[robertwb]

comment:8

I will certainly look into that.

[robertwb]

comment:9

Interestingly enough, on my machine sage -t sage/crypto/mq/sr.py, so it is just the (single) long doctest that provides the slowdown.

[robertwb]

comment:10

See #4186 for a fix.

[sagetrac-mabshoff]

comment:11

Replying to @robertwb:

Interestingly enough, on my machine sage -t sage/crypto/mq/sr.py, so it is just the (single) long doctest that provides the slowdown.

Yeah, in that doctest we do some wacky coercion into some mv polynomial ring with a couple thousand variables, so this is really an interesting test case. This was first discussed at SD6 in Bristol, so it just shows how much the coercion re-re-write has been an interesting road :)

Cheers,

Michael