Bug #102796 for Map-Tube-Plugin-FuzzyFind: Test failure if newest Map::Tube::London is installed

Mon Mar 16 02:19:40 2015 SREZIC [...] cpan.org - Ticket created

Subject:

Test failure if newest Map::Tube::London is installed

The t/07-levenshtein.t may fail if the newest Map::Tube::London is installed: # Failed test 'Finding many lines Packalu fuzzy with distance 6' # at t/07-levenshtein.t line 36. # Structures begin differing at: # $got->[5] = Tunnel # $expected->[5] = Does not exist # Looks like you failed 1 test of 26. t/07-levenshtein.t ... Dubious, test returned 1 (wstat 256, 0x100) Failed 1/26 subtests It looks like the changed London map due to the new "other_link" feature is unexpected for your test script.

Mon Mar 16 09:10:20 2015 GWS [...] cpan.org - Taken

Mon Mar 16 09:13:26 2015 GWS [...] cpan.org - Correspondence added

Am 2015-03-16 02:19:40, SREZIC schrieb: Show quoted text

> The t/07-levenshtein.t may fail if the newest Map::Tube::London is > installed:

[...] Show quoted text

> It looks like the changed London map due to the new "other_link" > feature is unexpected for your test script.

Yeah, it's fragile to use structures not under ones own control for offline testing. Maybe I should use synthetic test data that are under my control. The potential disadvantage of that is that they would be synthetic :-(

Mon Mar 16 09:13:26 2015 The RT System itself - Status changed from 'new' to 'open'

Mon Mar 16 16:46:55 2015 SREZIC [...] cpan.org - Correspondence added

On 2015-03-16 09:13:26, GWS wrote: Show quoted text

> Am 2015-03-16 02:19:40, SREZIC schrieb:

> > The t/07-levenshtein.t may fail if the newest Map::Tube::London is > > installed:

> [...]

> > It looks like the changed London map due to the new "other_link" > > feature is unexpected for your test script.

> Yeah, it's fragile to use structures not under ones own control for > offline testing. Maybe I should use synthetic test data that are under > my control. The potential disadvantage of that is that they would be > synthetic :-(

In my opinion using real data is fine, even if it means that things may break now and then. The danger with synthetic data is always that the real thing is changing, and one does not notice the changes.

Tue Mar 24 07:36:58 2015 GWS [...] cpan.org - Correspondence added

Am 2015-03-16 16:46:55, SREZIC schrieb: Show quoted text

> In my opinion using real data is fine, even if it means that things > may break now and then. The danger with synthetic data is always that > the real thing is changing, and one does not notice the changes.

I fully agree. Issue should be fixed in 0.07.

Tue Mar 24 15:48:21 2015 SREZIC [...] cpan.org - Correspondence added

On 2015-03-24 07:36:58, GWS wrote: Show quoted text

> Am 2015-03-16 16:46:55, SREZIC schrieb: >

> > In my opinion using real data is fine, even if it means that things > > may break now and then. The danger with synthetic data is always that > > the real thing is changing, and one does not notice the changes.

> I fully agree. Issue should be fixed in 0.07. >

Unfortunately not, I see the same error with 0.07. [BTW, are you located in Berlin? Will you attend tomorrow's berlin.pm.org meeting?]

Tue Mar 24 17:29:50 2015 GWS [...] cpan.org - Correspondence added

Am 2015-03-24 15:48:21, SREZIC schrieb: Show quoted text

> Unfortunately not, I see the same error with 0.07.

Mhm, I can't reproduce that here. Failing any smoke testers' report up to now, can you tell me whether it's still the same line (36) that generates that error, and still exactly the same message? The updated test script does contain 'Tunnel' as element #5 of the expected return value. Show quoted text

> [BTW, are you located in Berlin? Will you attend tomorrow's > berlin.pm.org meeting?]

Yeah, I am, but I can't make it tomorrow :-( Will try on some other occasion, I do read berlin.pm's mails.

Tue Mar 24 17:45:14 2015 SREZIC [...] cpan.org - Correspondence added

On 2015-03-24 17:29:50, GWS wrote: Show quoted text

> Am 2015-03-24 15:48:21, SREZIC schrieb:

> > Unfortunately not, I see the same error with 0.07.

> Mhm, I can't reproduce that here. Failing any smoke testers' report up > to now, can you tell me whether it's still the same line (36) that > generates that error, and still exactly the same message? The updated > test script does contain 'Tunnel' as element #5 of the expected return > value.

Ah, I see it now --- the failed tests have it the other way: # Failed test 'Finding many lines Packalu fuzzy with distance 6' # at t/07-levenshtein.t line 36. # Structures begin differing at: # $got->[5] = Does not exist # $expected->[5] = 'Tunnel' # Looks like you failed 1 test of 26. This seems to happen only for my perls with an older Map-Tube-London. So it's best to skip the test if Map-Tube-London is not new enough, or handle both cases. Show quoted text

>

> > [BTW, are you located in Berlin? Will you attend tomorrow's > > berlin.pm.org meeting?]

> Yeah, I am, but I can't make it tomorrow :-( Will try on some other > occasion, I do read berlin.pm's mails.

OK, then maybe next time!

Tue Mar 24 18:42:29 2015 GWS [...] cpan.org - Correspondence added

Am 2015-03-24 17:45:14, SREZIC schrieb: Show quoted text

> Ah, I see it now --- the failed tests have it the other way: > > # Failed test 'Finding many lines Packalu fuzzy with distance 6' > # at t/07-levenshtein.t line 36. > # Structures begin differing at: > # $got->[5] = Does not exist > # $expected->[5] = 'Tunnel' > # Looks like you failed 1 test of 26. > > This seems to happen only for my perls with an older Map-Tube-London. > So it's best to skip the test if Map-Tube-London is not new enough, or > handle both cases.

Nah, I'm checking for M::T::London 0.71, the test won't be run if it's older. However, I don't check for older versions of Map::Tube itself, and older versions of that module don't deliver the other_line values. M::T::London also doesn't require a newer M::T version (rightly so). I'll try to catch that case too.

Tue Mar 24 18:51:41 2015 GWS [...] cpan.org - Correspondence added

Am 2015-03-24 18:42:29, GWS schrieb: Show quoted text

> Nah, I'm checking for M::T::London 0.71, the test won't be run if it's > older. However, I don't check for older versions of Map::Tube itself, > and older versions of that module don't deliver the other_line values. > M::T::London also doesn't require a newer M::T version (rightly so).

EDIT: Also(?), as of M::T 2.94, get_lines() doesn't return other_lines anymore. I'd rather not track compatibility with individual versions of M::T in combination with different versions of M::T::London (-> maintenance nightmare), but rather stick to the behaviour by their latest versions.