[Issue 10902] some phobos unittests take an excessive amount of time

[d-bugmail at puremagic.com]

https://d.puremagic.com/issues/show_bug.cgi?id=10902



--- Comment #2 from Martin Nowak <code at dawg.eu> 2013-12-22 11:07:32 PST ---
> AMD Phenom II X6 1055T

Quite nice, I'm running on a i5-2400S. I'm still considering a Haswell or
Broadwell because of the 'gather' instruction.

> std.datetime totally takes the cake when it comes to tests, I'm guessing just
because of the sheer number of tests?

I think this is really low-hanging fruit. A lot of test times don't relate to
the complexity or I/O requirements of the module, so most likely the unittests
cover an overly huge amount of data.
Often random coverage is good enough, see
https://github.com/D-Programming-Language/phobos/pull/653.

If you look at a test run, e.g.
http://d.puremagic.com/test-results/test_data.ghtml?projectid=1&runid=57240&logid=6,
this is the tail.

2.747s PASS std.process
2.332s PASS std.uuid
3.278s PASS std.utf
4.290s PASS std.regex
4.447s PASS std.random
7.251s PASS std.parallelism
8.268s PASS std.conv
16.090s PASS std.file
45.696s PASS std.datetime

The suspect modules are std.uuid, std.random, std.conv and std.datetime.

-- 
Configure issuemail: https://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------