[Kmymoney-devel] csvimport plugin test is translation dependent

[allan]

On 16/09/10 18:45, Cristian Oneţ wrote:
> 2010/9/16 Alvaro Soliverez <asoliverez at kde.org>:
>> Comment out the short month test for now.
>>
>> Anyway, the test is inherently translation dependent, but it is
>> supposed to work ok. That is, until someone screwed up when
>> translating to Romanian. :/
> 
> I've done that but if it can't be rewritten in a translation
> independent manner I would prefer to remove it.
> 

As I see it, a test on the validity of an alpha month has to reflect the
current locale, so has to involve translation.  It would seem that the
test was valid, but the translation was not, if I understand correctly.
 Presumably in real use, a valid short date month would be rejected, or
perhaps just the one I chose at random?

I contemplated leaving the test just commented out, for possible future
inclusion once the translation is fixed, but the same could happen
again.  It was meant to be a test of convdate() rather than translation,
so I'll remove it totally.

Out of interest, what happens with the later test
QLatin1String("13-rubbishmonth-2000")?

Allan

>> On Thu, Sep 16, 2010 at 2:25 PM, Cristian Oneţ <onet.cristian at gmail.com> wrote:
>>> Hi Allan,
>>>
>>> Can you please check the test related fix in this commit. If the test
>>> can't be written in a translation independent manner then it should be
>>> removed.
>>>
>>> Regards,
>>> Cristian
>>>
>>> SVN commit 1176100 by conet:
>>>
>>> Fix compilation errors on MSVC due to a rather strange way of using
>>> QStringList. I think that the code using QStringList should have been
>>> more like the one in this patch. Also I commented out a test case
>>> which was failing in Romanian since it was locale dependent (the short
>>> month name returned 'nov.' which produced '5-nov.-1999' which can't pe
>>> parsed)
>>>