[Insight-developers] Unicode filenames and borland issue

[Tom Vercauteren]

Hi Bill,

Just realized that the Borland test I used was slightly different,
sorry about that.

I do indeed get the same error. It looks like borland does not
understand notations like
   L"\u03B1.txt"
for wide strings.

Replacing this notation with a more intricate one
  http://public.kitware.com/cgi-bin/viewcvs.cgi/Testing/Code/IO/itkUnicodeIOTest.cxx?root=Insight&r1=1.4&r2=1.5
fixes the borland failing test:
  http://www.cdash.org/CDash/index.php?project=Insight&date=2009-11-10#Continuous

Now comes my dear friend vs6 :(
  http://www.cdash.org/CDash/viewBuildError.php?buildid=468753
Using streams seems difficult with that compiler. Is there a
recommended way for handling this kind of errors?
* use iostreams from kwsys
* try adding some typedefs as in
http://public.kitware.com/cgi-bin/viewcvs.cgi/Code/Common/itkSymmetricSecondRankTensor.h?root=Insight&r1=1.4&r2=1.5
* replace "std::XXstream" with "std::basic_XXstream<char,
std::char_traits<char> >" as proposed in
http://list.sipfoundry.org/archive/resiprocate-devel/msg01085.html
* something else

I could try these approaches but will need to it it blindly as I don't
have access to VS6.

Regards,
Tom

On Tue, Nov 10, 2009 at 05:34, Bill Lorensen <bill.lorensen at gmail.com> wrote:
> What BUILD_TYPE are you using. I use MinSizeRel.
>
> I do not build shared libraries on Borland.
>
>
> On Mon, Nov 9, 2009 at 2:12 PM, Tom Vercauteren <tom.vercauteren at m4x.org> wrote:
>> Hi Bill,
>>
>> As you may have noticed, I have committed a unit test to experiment
>> with unicode filenames on windows:
>>  http://public.kitware.com/cgi-bin/viewcvs.cgi/Testing/Code/IO/itkUnicodeIOTest.cxx?root=Insight&view=markup
>>
>> Thanks to this webpage:
>>  http://edn.embarcadero.com/article/20633
>> I have been able to test the code on borland c++ 5.5. I worked out
>> without any issue.
>>
>> However on your system, it appears that the unit test is failing:
>>  http://www.cdash.org/CDash/testDetails.php?test=36142693&build=468077
>>
>> Apparently, MultiByteToWideChar does not work as expected on you
>> machine. Would you by any chance have a clue on why this is the case?
>> What version of windows, and filesystem are you using? If you have a
>> little spare time, could you try and experiment with itkUnicodeIOTest
>> and see what it gives? You may want to try by commenting out the calls
>> to removeAlpha and manually inspeting the txt files that are produced
>> by the test.
>>
>> Cheers,
>> Tom
>>
>