Hi Alexey,
On Mon, April 7, 2014 20:06, Alexey Zakhlestin wrote:
>
> On 07 Apr 2014, at 19:45, Anatol Belski <[email protected]> wrote:
>
>
>> Hi Ferenc,
>>
>>
>> On Mon, April 7, 2014 14:48, Ferenc Kovacs wrote:
>>
>>> Hi,
>>>
>>>
>>>
>>> I have never managed to make this test pass for me, even thought that
>>> I
>>> was trying with different ICU versions. I'm currently testing with ICU
>>> 52.1
>>> from homebrew on Mavericks and with the same ICU version on debian
>>> wheezy and it fails with the same diff:
>>>
>>> 209+ 'timezone_type' => 2,^M
>>> 210+ 'timezone' => 'PDT',^M
>>> 209- 'timezone_type' => 3,^M
>>> 210- 'timezone' => 'America/Los_Angeles',^M
>>> 217+ 'timezone_type' => 2,^M
>>> 218+ 'timezone' => 'PDT',^M
>>> 217- 'timezone_type' => 3,^M
>>> 218- 'timezone' => 'America/Los_Angeles',^M
>>> 225+ 'timezone_type' => 2,^M
>>> 226+ 'timezone' => 'PDT',^M
>>> 225- 'timezone_type' => 3,^M
>>> 226- 'timezone' => 'America/Los_Angeles',^M
>>> 233+ 'timezone_type' => 2,^M
>>> 234+ 'timezone' => 'PDT',^M
>>> 233- 'timezone_type' => 3,^M
>>> 234- 'timezone' => 'America/Los_Angeles',^M
>>> 241+ 'timezone_type' => 2,^M
>>> 242+ 'timezone' => 'PDT',^M
>>> 241- 'timezone_type' => 3,^M
>>> 242- 'timezone' => 'America/Los_Angeles',
>>>
>>>
>>>
>>> Could anybody who has this test passing tell me what ICU version are
>>> you testing with?
>>>
>>> --
>>> Ferenc Kovács
>>> @Tyr43l - http://tyrael.hu
>>>
>>>
>>>
>> I've forked this one into
>> ext/intl/tests/dateformat_format_variant3.phpt and adjusted the output
>> for 52.1. The variant 2 is worky with at least ICI 51.2, i've added the
>> corresponding condition to the skipif. Looks like this change happens
>> somewhere between ICU 51.2 and 52.
>
> well… the question is: should something like this be tested anyway? tests
> should cover things which are guaranteed to work specific way by us, not
> by external libraries.
>
> --
> Alexey Zakhlestin
> CTO at Grids.by/you
> https://github.com/indeyets
> PGP key: http://indeyets.ru/alexey.zakhlestin.pgp.asc
>
>
we can't test every single version, but at least we ensure some of the
changes rapidly happening in ICU. There's pretty much data, and that's
somehow an issue because it's causing erroneous bug reports. On the other
hand, if we would reduce the tests only to the functionality within PHP
responsibility, then they were quite a few.
So I would say it's fine to test the way we do, at least at the time where
some new ICU version arises, failing tests can show what was changed and
whether PHP is fine with that. Almost that's data change, but sometimes it
could be something more serious, too.
Regards
Anatol