Re: [Tuxpaint-i18n] Confused by font scoring strings

[Pere P. i C. <per...@gm...>]

Hi Karl,

El dg. 26 de 02 de 2023 a les 15:36 +0100, en/na Karl Ove Hufthammer va escriure:
> Hi!
> I’m confused by the four new font scoring strings. The extracted comment says:
>  
> > If neccessary, translate any of following strings using at least
> > two locale specific characters required to render your language.
> > Then, the scores for those fonts having such characters will increase.
> > You can use two different weight for scoring, 1 or 9, according
> > to the importance.
> And the strings are:
>  
> > <1>spare-1a <1>spare-1b <9>spare-9a <9>spare-9b
> Which parts of the strings should I translate? 

You should replace the whole string  by chars needed in your language.
Or just 2 identical chars if you are happy with the scores of the other font quality checks. 


> Should I just append two (or more) characters:
>  
> > <1>spare-1aæø <1>spare-1bæø <9>spare-9aæø <9>spare-9bæø

No, that lets 2 identical chars in each string('1' or '9') and the check will fail.


> Or should I replace the entire strings? 

Yes, if needed by the language.


> Does the numbers (the ones on start of each string and/or the ones on the end) signify anything?

Yes, they match the weight the string has in the font scoring process,
the '1' strings wheight 1 point, the '9' strings wheight 9 points.


>  Can I change them?

Yes, you can put there any string that reflects the needs of your language, 
but the wheight will not change, they will continue to  be of 1 and 9 points


>  What is the difference between ‘a’ strings and ‘b’ strings? 

'a' and 'b' strings are there to leave translators the option to use 2 strings for 1 point and 2 for 9 points.
They are independent, if one of the '1' strings fails, the other can sum if it passes.
Same for the '9' ones


> Why are there four different strings? 

Depending on the language requirements you may need 0 to 4 of those strings.
If you are fine with the font scoring provided by the other translations, then you don't need any of the spare strings.

You could use the '9' strings to give the fonts that passes their tests  a very high score, 9 + 9
ensuring that those fonts will end up in the first position despite other tests.

This may allow to not need to use the blacklisting "QX" "qx" strings but still get a high punctuation 
to the fonts suitable in a language, allowing fonts for other languages be available more down the list.


> What should I (not) put into them?

If you want some of those tests count in the font score wheigthing, you must not put two (2) identical chars in them.

If you don't need some of those tests, you must put two (2) identical chars there.
Note that by default they contain already two identical chars each of them, so leaving them untranslated is safe if you don't need them.


The tests work as this, IIRC:
render each of the chars of the string and compare to each other, 
if a char fails to render, then the test fail.
if one char renders as a square/rectangle but renders anyway, the test doesn't fail (still). 
if a char renders identical as another char(example 2 chars rendering as squares/rectangles), then the test fail

So many fonts replace missing chars by rectangles instead of simply not rendering, so to catch them you need to make the font render 2 rectangles


See also the thread at
https://sourceforge.net/p/tuxpaint/mailman/tuxpaint-devel/thread/20221217110312.c9b6e27cb0349790aaac288c%40wmail.plala.or.jp/#msg37749918



HTH
Pere