wchar in ansilib

Hello,
I have DDE27 and I am testing UTF-8 (I am French)
In Clib there is wchar.h but the librarie does’nt contain wchar.c, wctype !
I look up in the last BCM2835Dev, idem.
Please is it possible to get this library, or just wchar.c and wctype.c which are missing.

Perhaps it is better to use Unicode library?

Thanks
Jean-Michel

Perhaps it is better to use Unicode library?

Most likely, yes.

Wide character encodings were designed to deal with things like Asian characters before the adoption of Unicode – such as Shift-JIS and the like. Ironically wide encodings don’t specify any actual format, only that characters are represented by multiple byte encodings… The size of which is compiler specific.
If you’re thinking a pre-Unicode world was a mess, you’d be right.

Wide character support could, I suppose, be pressed into use for UTF-16, however if I was going to use Unicode on RISC OS, I would use the one most compatible with existing code – namely UTF-8 where a string is still a string in C¹, all you need to understand and support is a break from the idea that 1 byte = 1 character. Luckily the UTF-8 encoding is both simple to understand and gives it’s byte length (normally 1 to 4 bytes, but can be 1-6) as a part of itself.

The challenge, of course, is getting the Wimp to do something sensible with the text. While the machine can be switched to UTF-8 and it all magically works, the side effect is that everything else (assuming Latin1) breaks. As a Frenchie you’d see all your accents disappear.
How much I wish the Wimp had a flag that you could tell it “this application is UTF-8” and it would just go ahead and do something sensible regardless of the configured alphabet. RISC OS can so it – the NetSurf browser and my hacks to Ovation prove this, but the caveat is that you must paint the text yourself, which is somewhere between useless and impossible when we’re talking about window icons and error boxes.
It shouldn’t be hard for the Wimp to often a duplicate don’t handle, the second in UTF-8, and switch as the app requires, but since most users speak English and English looks the same in ASCII and UTF-8…..

And, no, I don’t expect fruity things like switching out fonts to support inline Cyrillic, Greek, kana, etc – that probably requires changes in FontManager for what might be a very narrow use case mais d’avoir des mots dans différents encodages qui semblent les mêmes sur l’écran and not with missing accents (other languages use twiddly bits over characters and they aren’t there as a fashion accessory like the umlaut in the names of certain metal bands) just because the alphabet is different… That is something the Wimp should do.²

¹ UTF-16, on the other hand, represents most English text as a character followed by a null, which would need continual special handling due to C treating a null as a string terminator.

² But, then, I’m well known as the sleeping monster that gets poked whenever something mentions Unicode and RISC OS in the same sentence. ;-) I probably would have ranted more had I a keyboard instead of a tiny screen to write this on. ;-)

Thanks Rick for these clarifications.
bq.(I would use the one most compatible with existing code – namely UTF-8)
I agree with you.
I am trying to use Unicodelib:
// --Extrait:
XML_Char *s;
printf(“car: %s carlen = %d\n”,s,len );
seqnb = UTF8_to_UCS4(s, ucode);
printf(“\tUcar: %s, seqnb= %d\n”, ucode, seqnb);
// — end
results:
car: é|B carlen = 2
Ucar: é, seqnb= 2

The € is difficult to get, is it the reason for Brexit :)

Hello,
Using UTF8 from Unicode library I can have French accent in my software:
<?xml version=“1.0” encoding=“ISO-8859-1”?>

<TaàTRé Atté = “purée” Attû = “fenêtre” > Les aventures légendaires de Huckleberry Finn
<SousTitre Attribut1 = "test de ô et ö " > sous titré par ST501

</TaàTRé>
Before the parser reports errors, now the caracters are correctly displayed.
If you wish the little XmlEd viewer/editor

[XmlEd]https://jeanmichelb.riscos.fr/Programmes.html