It's expected that they would be different, since they are different
character sets with different requirements.

If you think something is not correct, please describe the exact
symptoms, and the *minimal, isolated* test cases for showing them.
Linking to a couple pages with dozens of links and just saying "they're
different" is not helpful.

gangleri wrote:

At http://en.wikipedia.org/wiki/User:Gangleri/tests/Unicode_ISO_8859-1/links additional
tests are available:

Summary: Only some character are responsible, that no links are generated. Such characters
are: [[Hebrew]] "נ" (נ), [[Runes|rune]] "ᚠ" (ᚠ) and probably
some more depending on other charactersets.

At the mentioned page these two links (with one single Unicode character) behave different
compared to other links with Unicode characters:
[[en:User:Gangleri/tests/Unicode_ISO_8859-1/נ]] with the "נ" character
*[[en:User:Gangleri/tests/Unicode_ISO_8859-1/Runes/ᚠ]] with the "ᚠ" character

Regards Reinhardt

They're invalid because they contain the byte %A0, which is not allowed in titles on a latin-1 wiki.

That the character references are converted as UTF-8 bytes in the first place is due to bug 65. Marking as duplicate.

• This bug has been marked as a duplicate of 65 ***