py3: some changes in misc

[chapoton]

extracted from #26321

[chapoton]

New commits:

​a7fcc75

various py3 changes in misc

[vklein]

-        sage: g = mypartial(f, 2, 3)
+        sage: g = mypartial(f, 2, 3)  # py2; on Python 3 this is an error
         Traceback (most recent call last):
         ...
         DeprecationWarning: object.__init__() takes no parameters

Why not testing the error case with py3 ? And if it is not worth it why keeping the test in py2 ?

[chapoton]

one should ask @embray, author of this

[embray]

Because it's checking behavior that is no longer applicable on Python 3.

[vklein]

Ok.

I am also surprised by a doctests behaviour.

When i see this

sage: sage_input(long(-5), verify=True)  # py2
# Verified
-long(5)
sage: sage_input(long(-7), preparse=False, verify=True)
# Verified
-7L
sage: sage_input(long(11), preparse=None, verify=True)  # py2
# Verified
long(11)

i expect the second test to fail in python3 but it doesn't whith a sage -t --long src/sage/misc/sage_input.py, what is the trick ?

[vklein]

I am ok with this ticket, i will set it to positive review after green bot.

[embray]

Replying to vklein:

i expect the second test to fail in python3 but it doesn't whith a sage -t --long src/sage/misc/sage_input.py, what is the trick ?

There are some tricks in the doctest parser for massaging differences between Python 2 and 3 w.r.t. to the Python 2 long type. For example, the L-suffix is just ignored on Python 3 as long as the preceding numerals are the same.

But I don't think the Sage pre-parser will take long(1) and output long(1) on Python 3. So there are a few exceptions. But the majority of long-related tests work on Python 3.

[vklein]

Ok, thank you for clarifying that.