Re: [privoxy-devel] windows 3.0.11 regression tests

[Lee <le...@gm...>]

On 2/23/09, Fabian Keil <fk...@fa...> wrote:
> Lee <le...@gm...> wrote:
>
>> On 2/22/09, Fabian Keil <fk...@fa...> wrote:
>> > Lee <le...@gm...> wrote:
>> >
>> >> On 2/22/09, Fabian Keil <fk...@fa...> wrote:
>> >> > Lee <le...@gm...> wrote:
>> >> >
>> >> >> Is it expected behavior to not run all of the regression tests?
>> >> >
>> >> > Yes. Tests that are known not to apply are skipped.
>> >>
>> >> How hard would it be to put out a "skipping <whatever>: known not to
>> >> apply" message?
>> >
>> > Trivial, after all it's written in Perl.
>>
>> trivial if one is sufficiently knowledgeable ... sure :)
>
> Skipped tests and the reason for skipping them are now logged
> if the --verbose options is used.

very nice - thank you!

> I'll probably add a dedicated --show-skipped-tests option later on.

that would be even better.


>> > If you provide some more details about which tests
>> > fail and why, I'll look into possible workarounds.
>>
>> I wouldn't bother - I'm over-riding the default behavior, so I'd
>> expect anything that tests for the default behavior to fail in those
>> cases.  For example, I get a lot of
>>   Ooops. '-filter' is not among the final results.
>> and
>>   Ooops. '-fast-redirects' is not among the final results.
>> and I haven't noticed any problems with pages not displaying correctly
>> or links not working.
>
> Usually the "Sticky Action" tests fail because a custom
> action file after default.action contains a match-all section.
> Now that we have match-all.action that should no longer be
> necessary.
>
> There are of course other reasons for overriding specific
> parts of default.action and I intend to address this by
> adding Privoxy-Regression-Test syntax to somehow override
> earlier tests.
>
> However, as the default.action file is intended to be a
> reasonable default for all users, there are also situations
> where it makes sense to simply commit your changes to the
> default.action file to solve the problem for all users.

I think I'm a lot more aggressive about blocking ads than most.  For
example, default.action blocks ad*. sites and I took those filters and
changed them to block .ad*. sites.  So I don't have specific filters
for things like
  content.served.by.adshuffle.com/
  media.adrevolver.com/
  newsletter.adsonar.com/

> I think that's the case for your recently mentioned
> changes for Yahoo.

I'll commit the actions for things I got sick of seeing during
testing, but do you also want the stuff for blocking .ad*. sites?


>> >> The problem was having "user-manual ./doc/user-manual/" in the config
>> >> file and not including doc/developer-manual/ in the distribution
>> >> package.  It works fine now :)
>> >
>> > The link should still lead to the 404 page.
>>
>> *sigh*  it does.  I thought I tried copying doc/developer-manual/ from
>> the source directory to the c:\program files install directory &
>> testing but obviously not :(
>>
>> I'm thinking I should just leave the Windows install package as it
>> was; defaulting to getting the docs via the web.  Thoughts?
>
> The problem existed since the user-manual config option has been
> added years ago and you are the first person who reported it.
>
> I think it's a minor issue and I will not fix it in the
> FreeBSD port until I have to change it for other reasons.
>
> Releases are never perfect and I'm sure there are several
> other issues we just don't know about yet. I changed the
> link in buildsource.sgml to point to the developer manual
> on the website and that's good enough for me for now.
>
> If you haven't created the package yet, you could fix the
> link too, but you might as well release the package with
> the broken link.

Ok - I'll release it as is.

> Having the user manual delivered by Privoxy is a useful
> feature (especially for Tor users) and in my opinion a
> single broken link doesn't change that.
>
> Of course you are the packager, so it's your decision.

I've been using the local docs feature for a long time with no
problems, but I figured there must be some reason for not doing it by
default..  But if you're OK with it I'll leave it in :)

Thanks,
Lee