harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ivanov, Alexey A" <alexey.a.iva...@intel.com>
Subject RE: [testing] locale dependent tests
Date Wed, 26 Jul 2006 08:34:04 GMT
Tim,

Tests for DecimalFormat are not locale-specific. They fail because of
the difference in the default data. I mean those five mentioned in the
other thread (HARMONY-925). There's a problem with using
DecimalFormatSymbols in this class which causes
test_setDecimalFormatSymbolsLjava_text_DecimalFormatSymbols to fail.
(This test is now excluded from running with 'failing' prefix; moreover
the whole test class is excluded.) I logged this issue in HARMONY-965,
and prepared a patch to fix the problem. Could you look at it?

As for failing test ChoiceFormatterTest.test_toPattern(), I've also
attached a patch to HARMONY-925 which fixes the problem.
Later today I'm going to create a patch for DecimalFormatTest. I think
to set the locale to en_US, so that the result of formatting is the same
as expected. What do you think about it?
Still it has several other tests failing, and I can't figure out why.

>-----Original Message-----
>From: Tim Ellison [mailto:t.p.ellison@gmail.com]
>Sent: Tuesday, July 25, 2006 7:47 PM
>To: harmony-dev@incubator.apache.org
>Subject: Re: [testing] locale dependent tests
>
>On 25/07/06, Richard Liang <richard.liangyx@gmail.com> wrote:
><snip>
>> Not sure what the "fake" hard-coded locale is. I used to set the
default
>> locale to some particular one, e.g., en_US, but it seems that someone
>> (Tim) does not like the idea. :-)
>
>What I objected to was setting the locale to a specific value that
>made the tests pass even where the logic is locale-specific.  So, for
>example, if an application using the ru_RU locale would experience the
>problem that Alexey is seeing with the DecimalFormat, then the tests
>should use the default locale, and we (Harmony) get the benefit of a
>wide community of hetrogeneous testers.
>
>i.e. don't make all tests run on a uniform machine set-up -- let them
>be realistic.

Do you think we should always use the default locale, take the data from
it, generate the result of formatting using these data, and then compare
the strings?
In my opinion, this will lead to almost unreadable tests? On the other
hand, tests may still use the default locale until they proved to fail
on some of the locales. And then we'll take a decision how to update
test and whether we need an additional test for another locale. I mean,
if necessary, we can create several tests for one method but with
different locales, like en_US, ru_RU. Does it make sense?


Thanks,
Alexey.


--
Alexey A. Ivanov
Intel Middleware Product Division

>
>Regards,
>Tim
>
>--
>
>Tim Ellison (t.p.ellison@gmail.com)
>IBM Java technology centre, UK.
>
>---------------------------------------------------------------------
>Terms of use : http://incubator.apache.org/harmony/mailing.html
>To unsubscribe, e-mail: harmony-dev-unsubscribe@incubator.apache.org
>For additional commands, e-mail: harmony-dev-help@incubator.apache.org

---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: harmony-dev-unsubscribe@incubator.apache.org
For additional commands, e-mail: harmony-dev-help@incubator.apache.org


Mime
View raw message