Python 3 preparation: The semantic of filter() function is changed

[wluebbe]

In Py2 filter() returns a list, while in Py3 filter() returns an iterator (as itertools.ifilter() does in Py2).

The tool 2to3 wraps filter() usages with a call to list().
An alternative approach is to add from future_builtins import filter and to check where a wrapping with list() is required.

There are 53 effected modules.

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

[wluebbe]

2to3 does even better as its documentation ​2to3 filter suggests:
"Wraps filter() usage in a list call."

Often it changes the code as indicated by ​filter():
"Note that filter(function, iterable) is equivalent to [item for item in iterable if function(item)] if function is not None and [item for item in iterable if item] if function is None."

This is more readable and more pythonic as a combination of filter and lamda. And it allows sometimes further improvements!

---

New commits:

​e574c68

trac #16067: changes generated by "2to3 -w -f filter src/sage"

[wluebbe]

These are the generated changes. Some manual fixups / improvements to come.

[wluebbe]

Apparently I did not run ./sage -b before the test - Grrr ... :-(

Now I will always use ./sage -b;./sage -tp 4 --all --long >logs/sage-tp4-all-long-16067.log in one line!

[git]

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

​65c8620

trac #16067: fix an error generated by "2to3"

[wluebbe]

In sage/graphs/generic_graph.py the name filter was overwritten (by a local function definition. Therefore 2to3 got confused and wrapped the call filter() with list(). This caused a lot of doctest failures and 4 segfaults!

[git]

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

​1d50592

trac #16067: tidying up the generated patch

[wluebbe]

I think this a complete fix for the filter() function.

This ticket may really belong to stage 1 ...

[git]

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

​c68a033

Merge branch 'develop' into u/wluebbe/ticket/16067

[wluebbe]

Hi André, would you like to review this Py3 ticket?

[aapitzsch]

Hi, please change ==None back to is None in src/sage/combinat/k_tableau.py.

Could you check that the indentions are correct, e.g. in src/sage/misc/sagedoc.py:

match_list = [s for s in match_list if re.search(extra, s[1],
                                  re.MULTILINE | flags)]

should be

match_list = [s for s in match_list if re.search(extra, s[1],
                                                 re.MULTILINE | flags)]

[git]

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

​0cebea1

Merge branch 'develop' into u/wluebbe/ticket/16067

[git]

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

​0c2486e

trac #16067: answered reviewer comments and further improvements

[wluebbe]

Resolved merge conflict, answered the reviewer comment and added a few additional improvements. All doctests pass.

[aapitzsch]

Check your changes in src/sage/doctest/control.py and src/sage/quadratic_forms/quadratic_form__ternary_Tornaria.py. Some functionality appears twice.

[git]

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

​49e52c3

trac #16067: fixed duplicate lines found by reviewer

[wluebbe]

Hi André, thank you for reviewing and finding these problems! Apparently I wanted to go too fast :-/

"Eile mit Weile ..."