This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Charset request for iconv (patch included)
OK then, let's use the math-sigma. Can the
Unicode-->HP-48 conversion be set to accept either?
I still don't fully understand why Unicode has
accented letters when zero-width combining marks are
available. Best I can figure is that it allows C code
to treat each letter as an integer instead of a string
of indeterminate length.
Having an "x with macron" character would also
mean that when "x" appears in Unicode text, iconv can
write "x" to the HP-48 text without first checking to
see if the next character is a macron. But there's
obviously a reason why you implemented charsets with
DLLs instead of simple tables -- was it not to handle
special cases like this?
How much power do you guys have to add new
characters to the Unicode standard?
Dave Coffin 10/2/2016
On Sun, Oct 02, 2016 at 10:00:08AM +0200, Florian Weimer wrote:
> * Andreas Schwab:
>
> > On Okt 02 2016, dcoffin@shell.cybercom.net wrote:
> >
> >> I don't believe there is an official mapping. The author
> >> mentions two other mappings and how they differ from his. He
> >> also explains why he chose Greek letters over mathematical
> >> symbols. (Your Greek sigma looks correct in my editor while
> >> the math sigma looks like "???", so I can guess which is more
> >> widely supported.)
>
> Considering that the uppercase Greek sigma is next to an integral
> sign, I'm prety sure that its intended meaning is the mathematic
> operator. It's also how it is used in the HP-48 user interface. (I
> once had one myself.)
>
> > Just because some broken programs cannot display the characters should
> > not be a reason to use a wrong mapping.
>
> Right. And I would say the same applies to intentionally using a
> wrong mapping. “x̄” would have to be added to the Unicode to standard
> to enable compatibility with the HP-48 character set.
>
> Would we accept the mapping in case these issues were addressed? I
> still don't think so.