Re: multiple destinations in charset mapping files

[Kenichi Handa <handa@m17n.org>]

In article <wlab3y9hkj.wl%mituharu@math.s.chiba-u.ac.jp>, YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp> writes:

> I noticed that some charset mapping files such as
> etc/charsets/symbol.map contain entries where the same source is
> mapped to multiple destinations, and the latter one is preferred in
> decoding in such cases.

>   0x20 0x0020
>   0x20 0x00A0

>   (decode-char 'symbol #x20) -> 160

> But at least for symbol.map, it seems to be more natural to prefer the
> former entry (e.g., SPACE vs. NO-BRAKE SPACE, GREEK CAPITAL LETTER
> DELTA vs. INCREMENT).  WDYT?

I agree.  By this script in etc/charsets:

% for f in *.map; do awk '{print $1}' < $f | sort | uniq -c | grep '^ *[2-9] 0' && echo $f; done

I confirmed only symbol.map and stdenc.map contain such
duplications, so I regenrated those maps (simply by doing
sort -r) and committed to EMACS_23_1_RC and trunk.

Do you find any other maps that have duplications?

---
Kenichi Handa
handa@m17n.org