sage -tp N should store times when some files fail

[rlmill]

line 364 of $SAGE_LOCAL/bin/sage-ptest is where to start looking:

    if len(failed) == 0:
        if interrupt == False:
            print "All tests passed!"
        else:
            print "Keyboard Interrupt: All tests that ran passed."
        #Only update timings if we are doing something standard
        if opts=="-long" or len(opts)==0:
            with open(time_file_name,"w") as time_file:
                pickle.dump(time_dict, time_file)
                print "Timings have been updated."
    else:
        if interrupt:
            print "Keyboard Interrupt, not all tests ran"
        print "\nThe following tests failed:\n"
        for i in range(len(failed)):
               print "\t", failed[i]
        print "-"*int(70)

The reason I want this is that if you're making lots of changes and testing frequently, and you never get a completely clean run, all the good files still run in a random order, which is inefficient.

Component: doctest coverage

Author: Robert Miller

Reviewer: Rob Beezer

Merged: sage-4.3.1.rc1

Issue created by migration from https://trac.sagemath.org/ticket/7941

[rlmill]

Author: Robert Miller

[rlmill]

Attachment: trac_7941.patch.gz

apply to scripts repo

[rbeezer]

comment:3

Works as expected. Removed timings, introduced a doctest that would fail.

Ran sage -tp 2 -long devel/sage/sage/graphs/

twice, and repeated experiment without -long argument.

In both cases, second run obviously employed timings.

Positive review.

[rlmill]

Merged: sage-4.3.1.rc1