nauty_geng is naughty

[sagetrac-azi]

I think nauty_geng should not ignore geng's return value and return an exception if the returned value is not 0. Consider the following:

def find_counterexample(n):
   for G in graphs.nauty_geng("-c" + str(n)):
     if is_counterexample(G):
        print 'Yeees, there is a counterexample'
        return 
   print 'No counterexamples! Conjecture holds'

the code appears to be just fine but it has a bug! Namely it tries to run the program geng with the parameter "-cn" instead of "-c n". Hence geng returns an error and instead of getting an error in Sage, it simply does not generate any graphs.

The branch implements the catching of geng's error.
We also clean up the code of graph_generators a bit.

---

---

CC: @dimpase

Component: graph theory

Author: David Coudert, Dima Pasechnik

Branch/Commit: 93edfb0

Reviewer: Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/14068

[dcoudert]

Author: David Coudert

[dcoudert]

comment:5

A possible fix is to raise an error only when parameter debug is False and that something goes wrong.

When debug is True, the behavior remains the same as before and so we continue to display the message output by geng in stderr.

---

New commits:

759fab2

trac #14068: check input when debug is False

[dcoudert]

Branch: public/graphs/14068_nauty

[dcoudert]

Commit: 759fab2

[dcoudert]

comment:6

I don't understand the reported pyflakes error with sage.graphs.strongly_regular_db. Other modules are imported the same way without causing pyflakes errors.

[dimpase]

comment:8

Replying to @dcoudert:

I don't understand the reported pyflakes error with sage.graphs.strongly_regular_db. Other modules are imported the same way without causing pyflakes errors.

perhaps pyflakes is confused by staticmethod() thing.
The only place where a method from strongly_regular_db is used is

strongly_regular_graph = staticmethod(sage.graphs.strongly_regular_db.strongly_regular_graph)

by the way, I don't understand why it's not written as

from . import strongly_regular_db
...
strongly_regular_graph = staticmethod(strongly_regular_db.strongly_regular_graph)

Is it cause this code pre-dates from . syntax?

[sagetrac-git]

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

93edfb0

refactoring using 'from .' syntax

[sagetrac-git]

Changed commit from 759fab2 to 93edfb0

[dimpase]

comment:10

OK, otherwise looks good. let's see if this commit makes pyflakes happy

[dcoudert]

comment:11

pyflakes is now happy !

[dimpase]

Changed author from David Coudert to David Coudert, Dima Pasechnik

[dimpase]

Reviewer: Dima Pasechnik

[fchapoton]

comment:13

moving milestone to 9.0 (after release of 8.9)

[vbraun]

Changed branch from public/graphs/14068_nauty to 93edfb0