[GNC-dev] Early locale initialization

[Geert Janssens]

In a moment of distraction I started looking at the locale initialization code 
in gnucash-core-app.cpp.

If I understand the actual code properly it's not exactly working as I had 
intended. So with this thread, I'd like to clear this out.

For reference I'm talking about the code that starts here:
https://github.com/Gnucash/gnucash/blob/
1756f76079221f77904ccb9c9a504f9ebca2eb0e/gnucash/gnucash-core-app.cpp#L418

Locale initialization is strongly influenced by platform, so there are 
different code paths to set locale.
1. Macos will use set_mac_locale
2. Windows will use set_win32_thread_locale
3. Linux will fall through and use setlocale

So far so good.

Looking deeper, the linux method will search for an environment variables 
LC_ALL or LANG in that order and uses its value as locale. If that fails, it 
will fall back to the C locale. There is also code  in place to allow the user 
to override any of these variables by setting them in the gnucash environment 
file.

On Macos the code will first try to use the Macos proper locale setup code. If 
that fails it falls back to locale "en_US.utf-8". Then it sets the LANG 
variable to the same locale and proceeds with calling gnc_localeconv which 
will fill in the locale details that will be used throughout the code.

On Windows the code will first look for environment variables "LC_ALL", 
"LC_MESSAGES" or "LANG" (in that order). If any of those exists it will try to 
convert that to a locale. If that fails it will fall back to the system 
default locale and if that fails it falls through to our main code which then 
defaults to "C".


Putting these next to each other there are a few differences:
1. the final fallback locale. On Macos that will be "en_US.utf-8", on the 
other platforms it will be "C". Should we use the same default on all or 
rather not ?
2. Priority of locale source. On linux and Windows adding variables in the 
environment will allow the user to override the default system locale. On 
Macos it will only allow to set LANG or LANGUAGE variables, not locale. Do we 
want to allow the user to override system locale via environment variables or 
do we only want a fallback mechanism ? Do we want to allow the user specify 
the locale to use for that fallback ?
I'll note the Macos code has a neat fallback trick in that it tries to find a 
related locale in case the default doesn't work. Related in the sense that it 
may be another language in same country or if that fails the language without 
country. Downside is that in  case of multiple fallback options the one chosen 
may not be the one the user actually would have preferred.
3. While all three allow to tweak some environment parameters, only linux 
allows the user to do so via the environment file. For Windows and Macos that 
file is only read after locale is set. Whether that's a good thing depends on 
the answer to question 2.

Regards,

Geert