Internationalization

FDR wrote: No it works!
Strange!
Everything seems equal: the line feeds, the encoding...

Mine is larger, but it did the same when it was only a few rows too.

Can you try pulling the latest code.
I did find one bug:
I forgot to switch back to English when writing the ini, so it would be written in your language, but read in English which caused issues. This might explain why it failed only on startup, since it only reads tx settings on reboot.

No, it does the same...

FDR wrote: No, it does the same...

Can you enter USB mode when it is in a broken state (after switching languages and powering down..hold 'Ent')
And check the tx.ini and model.ini to see if they are in English and/or look corrupt. You could also just send them to me and I could try them in the emulator

They look fine!
BTW after clearing the language row from the tx.ini it works again in english...

FDR wrote: They look fine!
BTW after clearing the language row from the tx.ini it works again in english...

I'll check it tonight and see if I can figure out what is up.

The problem was a buffer overflow. I don't know why valgrind didn't catch it.

Anyhow, it is fixed now, and seems to work.

Yep, it works now...
Thanks!

wuselfuzz wrote: /edit: I really wouldn't run symbols for the mixer etc through the translate function, so model files are language independent.

Model files are language independent. The strings in the model file (and tx.ini) are always in English regardless of your language.

Corrected a typo in the hungarian language file.
Pushed...

FDR wrote: Corrected a typo in the hungarian language file.
Pushed...

Wait with pulling, I've found some more. I will correct them at home...

OK, it's done, I pushed...

I'm automating the extraction of the strings so it can be easily regenerated. Do you have any issue with me alphabetizing them in the output?
It will make it much easier to detect when new strings are added. (I'll alphabetize the existing language files too). Then we can use 'diff' to find new strings that need translating.

Edit: I've committed updated language files that are sorted.
Strings that were previously translated but are no longer used will be prefixed with '<:' and '<'. Strings that are new and have not yet been translated are prefixed with '>:'

You can run 'make language' at any time to extract the latest strings and updatethe existing language files as necessary. I think it will work on Windows too.

A (may be too late) question:
Is it necessary, that the language files are numbered instead of "lang.hu" or "lang.hu.txt" or "hu.txt"?
It makes harder to identify the language files.
I would prefer the named method for the model files too, but I understand, that they must exist, because the FS cannot create files, so they cannot be created or renamed on the fly...
But the language diretory (and for the matter the templates too) could be read as a whole directory, since they are read only.

there is no way to search for files in the code today (because of how I have petit_fat compiled) So yes, today it is necessary to use numbered files.
Someday I'll enable the full filesystem code support, but not before the 1st release.

Ah, that's a pity...

I saw that you've done it!
(I think the german version should be named lang.de, but Wuselfuzz can tell for sure...)

Now there is nothing prohibits you to make the templates the same!

BTW, I pushed some changes to the lang.hu

Sorry, I still found typos in lang.hu
Pushed...