Linux Receiving Generic Casefolding Implementation For File-Systems

[phoronix]

Phoronix: Linux Receiving Generic Casefolding Implementation For File-Systems

In making for easier code re-use among file-systems and allowing a unified implementation to focus on a single code-base for optimizations moving forward, a generic case-folding implementation for Linux file-systems is being prepared for mainline...

http://www.phoronix.com/scan.php?page=news_item&px=Generic-Casefolding-FS

[ermo]

On the surface, this seems a sensible move assuming that case-folding is here to stay (which appears to be the case *ahem*).

[boxie]

On the surface, this seems a sensible move assuming that case-folding is here to stay (which appears to be the case *ahem*).

iseewhatyoudidthere.gif

[JustRob]

Looks simple: https://git.kernel.org/pub/scm/linux...ore.c?h=v5.8.5

But far from it: http://perl11.org/blog/foldcase.html

[Nth_man]

Looks simple: https://git.kernel.org/pub/scm/linux...ore.c?h=v5.8.5

But far from it: http://perl11.org/blog/foldcase.html

It was written in other places:

You don't always name your files, sometimes someone else did. For example, take these three words:

• Maße
• MASSE
• Masse

In case-insensitive English, the first word is distinct from the others, but the last two are equal. However, in Swiss, the first and third are equivalent, but in German the first and second are (but not the third)...

[chithanh]

• Maße
• MASSE
• Masse

It gets even worse since Unicode 5.1 introduced ẞ (U+1E9E LATIN CAPITAL LETTER SHARP S) so you have to add MAẞE to the list.

ẞ has lower case of ß but the upper case of ß is SS.

[JustRob]

... and the mapping changes: so files that were converted from the filename given to how the name was stored must be compared with what you have on your list (as of the date / version that information was stored). So you might be normalizing and folding two things that weren't stored based on the current and (supposedly) correct Standard.