Re: 3.0.20 regression testing

On 1/4/13, Fabian Keil <[email protected]> wrote:
> Lee <[email protected]> wrote:
>
>> On 1/3/13, Fabian Keil <[email protected]> wrote:
>> >
>> > Using only the official files:
>> >
>> > actionsfile match-all.action
>> > actionsfile default.action
>> > actionsfile regression-tests.action
>> >
>> > I get:
>> >
>> > fk@r500 ~/git/privoxy $privoxy-regression-test.pl --max-level 109
>> > --show-skipped-tests
>> > 2013-01-03 20:26:44: Asking Privoxy for the number of action files
>> > available
> [...]
>> > 2013-01-03 20:26:57: Executed 658 regression tests. Skipped 7. 658
>> > successes, 0 failures.
>>
>> 13 seconds!!!  I wish.  Cygwin is *slow*
>
> On FreeBSD Privoxy-Regression-Test spends most of the time
> (IIRC more than 99%) executing curl, so you could try using
> a native curl binary which might make a difference.

I suspect a native curl binary would be more trouble than it's worth
since I'd also need a native perl binary + native whatever else the
regression test uses
.  The cygwin maintainers seem to place a much higher priority on
posix compatibility than playing nice with windows.  In other words, I
usually have problems when trying to mix cygwin & windows programs.

> You could also try profiling Privoxy-Regression-Test yourself first,
> but getting Devel::NYTProf running on Windows might not be your idea
> of a having a good time.

I think the problem is process creation in cygwin.  Try this on your FreeBSD box

$ cat speed.sh
#!/bin/sh
T=`date +"%S"`
N=0
while [ 1 ]; do
  if [ "$T" != "`date +"%S"`" ]; then
    printf "Loops: %s, time=%s\n" $N $T
    T=`date +"%S"`
    N=0
  fi
  N=`expr $N + 1`
done

$ ./speed.sh
Loops: 3, time=51
Loops: 23, time=52
Loops: 23, time=53
Loops: 22, time=54
Loops: 23, time=55
Loops: 23, time=56
Loops: 22, time=57
Loops: 23, time=58
Loops: 23, time=59
Loops: 22, time=00
Loops: 23, time=01
Loops: 22, time=02


>> > I usually silence them by setting the ignore flag,
>>
>> hrmm... maybe more documentation would help :)
>
> Now that I know that there is demand I should probably implement
> a proper solution. Of course this doesn't rule out explicitly
> documenting the work around as well ...
>
>> I took a quick look, but didn't see anything for an ignore flag.  Is
>> all your documentation inside of privoxy-regression-test.pl or is
>> there somewhere else I should be looking?
>
> Currently "the code is the documentation":

*sigh* I kinda thought so, my perl abilities leave much to be desired :(


> fk@r500 ~/git/privoxy $cat -n tools/privoxy-regression-test.pl
> [...]
>    578              } elsif ($token eq 'ignore' && $value =~ /Yes/i) {
>    579
>    580                  l(LL_FILE_LOADING, "Ignoring section: " .
> test_content_as_string($regression_tests[$si][$ri]));
>    581                  $regression_tests[$si][$ri]{'ignore'} = 1;
>    582                  $ignored++;
>    583
>    584              } elsif ($token eq 'expect status code') {
>
> Thus you can put "# Ignore = Yes" below any test you want to skip.

neat!  I'll give it a try

Thanks,
Lee

------------------------------------------------------------------------------
Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and
much more. Get web development skills now with LearnDevNow -
350+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122812