Python 3 preparation: Py2 vs. Py3 return value of write() in doctests

[Wilfried Luebbe]

The write() function does return nothing in Py2

>>> sys.version
'2.7.5+ (default, Feb 27 2014, 19:37:08) \n[GCC 4.8.1]'
>>> with open("tracked", "w") as f: f.write("boo")
... 
>>> 

while in Py3 it returns the number of bytes written

>>> sys.version
'3.3.2+ (default, Feb 28 2014, 00:52:16) \n[GCC 4.8.1]'
>>> with open("tracked", "w") as f: f.write("boo")
... 
3
>>> 

This can be unified by adding something like ignore =

>>> sys.version
'3.3.2+ (default, Feb 28 2014, 00:52:16) \n[GCC 4.8.1]'
>>> with open("tracked", "w") as f: ignore = f.write("boo")
... 
>>> 

About 79 py module are effected.

This ticket is tracked as a dependency of meta-ticket ticket:16052.

[John Palmieri]

New commits:

​d020f64

trac 16086: In Python 3, FILE.write('...') has a return value: the

[John Palmieri]

I used _ = file.open('...'), since _ is commonly used as an ignored variable in Python. Also, _ has a special meaning at the command line: the result of the previous command. But _ is treated differently in doctesting: in libs/pari/tests.py, if I use

    sage: _ = gpfile.file.write("polcyclo(5)\n")

on line 187, then on line 377,

    sage: _.Strchr()

the doctesting framework thinks that _ still refers to the result of the call to write. I would say this is a bug in the doctesting framework. To get around it, I used a double underscore on line 187 (and the previous few lines) in lib/pari/tests.py.

[Travis Scrimshaw]

Needs a rebase, but LGTM otherwise.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​0234816

Merge branch 'develop' into t/16086/ignore_write_output

[John Palmieri]

Okay, rebased.