Originally posted by kichik
There are none in NSIS. And even if language files are added for those, it still doesn't justify having every example in Unicode as most examples don't demonstrate multiple languages and those that do, usually do this with a small collection of languages.
I find the idea of a unicode capable NSIS compiler great.
I've built an (standard NSIS)installer for network drivers which currently supports 25 languages, including 2 RTL languages.
If we had to add some extra dialogs or messages all translation services would deliver unicode files.
I haven't yet understood or found a detailed explanation of how to safely convert unicode texts into NSIS codepage-based strings.
Being able to directly use unicode strings would significantly ease localization.
In my work I normally use PERL and the automated generation of Windows driver .inf files (UTF-16) is done by PERL scripts.
But NONE of the sources/templates, except for the NLS dependent parts (strings and readme files), are in UTF-16. As PERL uses UTF-8 internally and ASCII is a real subset of UTF-8 there is no need to have program sources in UTF-16.
When including a NLS file, which is in UTF-16, this is detected at read time.
I don't know if the Microsoft C++ API only supports UTF-16, but I think it should not be a great problem to have a smart "read file" function which does detection and conversion, even if internally all is handled in UTF-16.