hacked fonts in MS-Windows: rev. solidus vs Yen/Won(was..RE: Character identities)

From: Jungshik Shin (jshin@mailaps.org)
Date: Fri Oct 25 2002 - 13:09:57 EDT

  • Next message: Michael Everson: "Re: Sarati"

    On Fri, 25 Oct 2002, Marco Cimarosti wrote:

    > 1) Assigning arbitrary glyphs to some Unicode characters. E.g., assigning
    > the "$" character to long S; the ASCII letters to Greek letters; the whole
    > Latin-1 range to Devanagari characters, etc.

      There are several Japanese and Korean fonts with exactly this
    problem that are shipped with MS Windows. As is well known, at U+005C
    they have glyphs for Yen sign and Won sign instead of reverse solidus.
    IMO, MS is trying to solve a well-known problem of 'Yen/Won vs reverse
    solidus' in a totally inappropriate way(a font-hack !) . Nobody would
    argue that reverse solidus and Yen/Won signs are identical. However,
    what these fonts do is exactly that. Just switching fonts suddenly turn
    all reverse solidus to Yen/Won signs.

     MS-Windows has to provide distinct ways to enter 'reverse solidus' and
    'Yen/Won' sign (both full-width and half-width) in Japanese and Korean IMEs. It
    must be very easy to modify their IMEs that way. In Japanese and Korean
    input mode, pressing the key marked with 'vertical bar, Yen/Won' (this
    marking also has to be changed to have three 'vertical bar, reverse
    solidus and Yen/Won in a diff. color') should generate 'Yen/Won' sign
    (half-width/full-width version can be controlled the same way as for
    US-ASCII characters. Both Japanese and Korean IMEs in MS Windows offer
    this run-time configurable option) while in non-Japanese/Korean mode
    (usually English) they should generate 'reverse solidus' for the key.

      Somebody may argue that this would be problematic because there are
    a lot of old documents in traditional/legacy encodings (Shift_JIS/CP932
    and CP949) that use 0x5c for Yen/Won sign. There's no easy solution
    other than some heuristics combined with manual correction for this
    problem. This conversion has to be done for all old documents. It may
    be painful and time-consuming to life the degeneracy between 'reverse
    solidus' and 'Yen/Won sign' (in financial documents, most of them must be
    'Yen/Won sign'. In TeX/LaTeX, most of them must be 'reverse solidus'), but
    it has to be done at some point and it's better to do it early than later.
    After that, there should be no old documents/data with this problem IF
    MS Windows stops tacitly promoting its users to produce documents with
    U+005C meant for Yen/Won sign by shipping hacked-font mentioned above.

      Another argument against this change may be that it's quite
    inconvenient to toggle between EN mode and JA/KO mode when typing a long
    file path(in Japanese and Korean) with a lot of path separators (U+005C
    in MS-Windows) embedded. Well, most Windows users rarely type file paths
    directly. For a small number of users that do type them often but do
    not type currency signs as often (another group of people that need this
    option is TeX/LaTeX users, perl/shell/etc programmers), JA/KO IME can have
    another run-time option, whether or not hitting the key labelled with
    'reverse solidus/veritcal bar/Yen/Won' always produce 'reverse solidus'.

      Jungshik Shin

    This archive was generated by hypermail 2.1.5 : Fri Oct 25 2002 - 13:51:27 EDT