On Jun 14, 2005, at 1:32 PM, Kevin Scaldeferri wrote:
A little more interesting information. I ran a coverage test for the
full code base. Then I did this:
[kevin]% time perl -MDevel::Cover -e 1
...
perl -MDevel::Cover -e 1 14.19s user 0.88s system 79% cpu 18.997 total
After spending a little time reading the code, I tried simply
commenting out line 582 of Devel/Cover.pm:
# $Structure->read_all;
This brought the above test down to about 6 seconds instead, which is
still 4.5 seconds longer than when no coverage database is present at
all, but a substantial improvement, none-the-less.
Unfortunately, this change seems to make eval_use.t barf badly and I am
out of time to work on this for today. Here's a subset of the
failures:
t/eval_use..............ok 29/80# Test 30 got: '14 4 100
if ($x) { eval 'use E2' }
'
(/home/kevin/.cpan/build/Devel-Cover-0.53/blib/lib/Devel/Cover/Test.pm
at line 277 fail #30)
t/eval_use..............NOK 30# Expected: '14 4 100
100 if ($x) { eval 'use E2' }
'
t/eval_use..............ok 31/80# Test 32 got: '15 4
print "3 - $x\n";
'
(/home/kevin/.cpan/build/Devel-Cover-0.53/blib/lib/Devel/Cover/Test.pm
at line 277 fail #32)
# Expected: ' 2
'
t/eval_use..............NOK 32# Test 33 got: '16 4 100
100 if ($x < 4) { eval 'use E3' }
'
(/home/kevin/.cpan/build/Devel-Cover-0.53/blib/lib/Devel/Cover/Test.pm
at line 277 fail #33)
# Expected: ' 2
'
t/eval_use..............ok 34/80# Test 35 got: ' 4
'
(/home/kevin/.cpan/build/Devel-Cover-0.53/blib/lib/Devel/Cover/Test.pm
at line 277 fail #35)
# Expected: '15 4 print "3 -
$x\n";
'
...
Failed Test Stat Wstat Total Fail Failed List of Failed
------------------------------------------------------------------------
-------
t/eval_use.t 80 45 56.25% 30 32-33 35-42 44-48 50-58
60-68
70-80
1 subtest skipped.
Failed 1/34 test scripts, 97.06% okay. 45/3180 subtests failed, 98.58%
okay.
-kevin
