This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug localedata/10580] hr_HR: updated locale


https://sourceware.org/bugzilla/show_bug.cgi?id=10580

--- Comment #42 from Dragan Stanojevic - Nevidljivi <invisible@hidden-city.net> ---
> Before this change all abmon items (abbreviated month names) were 3 letters
> long. Now all are 3 letters long except the second item (February, Feb)
> which is "velj", 4 letters long. Previously it was "velj" therefore 3
> letters. Dragan, wouldn't you prefer it to be "vel", consequently 3 letters
> long?

True, before this change all were 3 letters, but through discussion with Mike
several arguments were made against using digraphs in LC_TIME:
- Unicode has since moved away from promoting them
- They have a lot of problems with digraphs and even tried to solve it with:
"U+034F COMBINING GRAPHEME JOINER" fix, so that digraphs would be glued with
it, but still written as two separate letters.
- Digraphs often look ugly in fonts, or are not contained in them so they're
substituted from another font, terminals in general don't have Unicode fonts,
and in TUI apps, it is better not to force digraphs, example would be `cal` or
TUI mail clients, shell prompt, tmux, ...
- abbreviations in many glibc locales isn't 3 letters. There is no rule that
they need to be, they just need to be shorter.
- I have wrongly assumed all abbr. needed to be of same length, they don't. If
that was the case I'd be more stubborn on digraphs, this way I'm more in favor
of "Velj".
- Many applications and many programmers decided to avoid glibc locale since it
was ugly. They either decided to make their own (LibreOffice for example), or
they do something like taking first 3 letters of a month or day name, giving
them wrong "Vel" values. "lj" is a digraph and a distinct phoneme, sounding
different from simple "l". IMO "Vel" is more wrong then "Velj".
- Glibc and CLDR were once very stern in what they'd accept. Now they've become
more pragmatic. One result is this issue with digraphs, but I hope that it is
clear that it was done with end users in mind. There are not many Unicode
digraphs used. And people will continue to type two letters for them since
entering digraphs is still awkward.

In the end, this patch was done more than 8 years ago. It was a complete
rewrite of the old locale and intention was to make it correct and easy to
read/maintain. During those 8+ years several bugs were issued towards hr_HR and
all of them were dups of this one, since I've solved all the issues back then.
Yet so many maintainers avoided this patch for one reason or the other. During
discussion with Mike, I really wasn't into forcing digraphs except in
LC_COLLATE, since that would be awkward for end users, and most other locales
avoid digraphs anyway. Even Unicode FAQ notes that they're troublesome in so
many practical ways.

In the end, I'm open to thoughts and arguments of others, especially end users,
but this patch, in any conceivable way compared to the previous state, is a
huge push towards maintainable and clear hr_HR locale.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]