[sldev] [i18n] - Question about test strings

Soft soft at lindenlab.com
Sat Nov 15 17:30:48 PST 2008

On Sat, Nov 15, 2008 at 6:43 PM, Alissa Sabre <alissa_sabre at yahoo.co.jp> wrote:
>> to compile a
>> series of tests to check if the viewer can handle characters used in
>> different languages.
> This phrase attracted me, but,
>> Noelle Linden asked volunteers to have a look at CT-81
>> ( https://jira.secondlife.com/browse/CT-81 ) in order to compile a
> I looked at it, but I don't speak French.
>> We're now trying to decide on TEST STRINGS for the viewer i18n. In that
>> context, the question arrised that if one
>> greek/chinese/japanese/russian/... character is working in the viewer,
>> it would be an indication if ALL characters of that language would be
>> working. It was speculative and we decided to ask the
>> specialists/coders... So that is you guys (^_^)
>> Any ideas? Should we test whole alphabets in different languages or are
>> one to three characters enough to indicate that the language is working
>> as it should?
> I have some ideas.  I have my own test vector for this area.  However,
> it primarily targets at my own goal of supporting _more_ languages
> than the current viewer does.  So, most of my data doesn't work at all
> under the LL viewer.  I don't think such set is appropriate for Q/A
> people...

I just published maint-l10n-1 with Ukrainian, Polish and Russian
community contributions added. Hopefully more follow in the near

> A part of my test data is available in-world at
> http://slurl.com/secondlife/Hippotropolis/42/22/23 as a set of
> notecards.  Those who are interested in the Zai's original message
> will find these notecards interesting.

Awesome! Actually, a notecard with the full expected character set for
each language sounds perfect for working out font issues, at least.
That's also going to let people A/B issues in the currently active
viewer and development viewers.

> Please note when you view it that:
> - The result of viewing those notecards largely depends on the OS
>  environment.  For example, assuming you run US English version of
>  Windows XP, the test results differ greately by a checkbox
>  hidden deep inside of the Windows Control Panel.  (Control Panel >
>  Date, Time, Language, and Regional Options > Regional and Language
>  Options > Languages > Supplemental Language Support > Install files
>  for East Asian Languges).
>  # I'm a Japanese speaker living in Japan, and am using Japanese
>    version of Windows.  In Japanese version of Windows this checkbox
>    is turned on by default, and is unable to be turned off through
>    usual operations.
> - Only the Korean text appears right in the current viewer (under
>  Windows XP with the above checkbox on.  If the above checkbox is
>  off, Korean notecard doesn't appear right.)
> - Turkish, Hungarian, and Czech text appears readable, but some
>  characters are shown strangely; it is primarily because the current
>  viewer's character-by-character font selection mechanism from
>  multiple fallback fonts works badly.
> - Viewing Japanese and Chinese notecards shows all characters there,
>  but the paragraphs are folded into lines at wrong positions, causing
>  unconfortable texts.  (In Chinese notecard, some Hanzi are shown
>  using Japanese font, causing some strange appearance.)
> - Russian text appears fine if the above Windows checkbox (East Asian
>  language) is turned *OFF* and you have Microsoft Office installed
>  with full install (i.e., you have "Arial Unicode MS" in your Font
>  folder.)  If you turned on the checkbox, each Russian letter appears
>  too wide.  If you turned off the checkbox, but you don't have
>  Microsoft Office, Russian letters doesn't appear on the screen (even
>  if you have an ordinary Russian font.)
> - Hindi, Hebrew, and Arabic sample text doesn't work with the current
>  viewer, regardless of the Windows settings.

Does anyone know if there's a standard way for detecting these
dependencies and suggesting remedies under Windows? Surely this has to
be a pretty common problem.

With Mac so far it seems that the correct characters are shown so long
as the proper system language pack is installed, and newer versions of
OS X may even include all languages in a default installation.

More information about the SLDev mailing list