Fwd: commit - speller dialog refactoring [platform mtrs pls read]

From: Hubert Figuiere (hub@nyorp.abisource.com)
Date: Wed Jan 01 2003 - 14:01:00 EST

  • Next message: Hubert Figuiere: "Fwd: commit - UT_isWordSeperator made a little faster"

    From: Jesper Skov <jskov@amber.zoftcorp.dk>
    To: abiword-dev@abisource.com
    Subject: commit - speller dialog refactoring [platform mtrs pls read]
    Message-Id: <20030101105424.D543419C109@zoftcorp.dk>
    Date: Wed, 1 Jan 2003 11:54:24 +0100 (CET)

    Speller Dialog refactoring:
     Previously, both the text/fl_BlockLayout and the wp/ap_Dialog_Spell
     classes contained code for finding words suitable for spell-checking.
     One was used by the dialog, the other by the background spell-checker.

     I have now added a new class fl_BlockSpellIterator which contains the
     necessary logic. Both the dialog and the background spell-checker use
     this code now.

    As part of the refactoring I've fixed a few leaks in the unix platform
    dialog code. The number of mallocs/frees have also been reduced quite
    a lot.

    Unfortunately, I think I've probably broken all other platforms than
    Unix, so could platform maintainers please have a look?

    At least Cocoa, BeOS, and Win code contains (I think) broken
    assumptions about how long a UCS4 string is when converted to c_str. I
    can't test changes there, so I'm not even going to try and fix
    it. Both the QNX and Unix code seem to handle this OK though (but by
    using UTF8 - so it's not a trivial copy'n'paste to fix in the others).

    I think the unix (and thus XP) code should work, but there may be
    broken corner cases. Please let me know if you find any problems.

    Thanks,
    Jesper

    ----- End forwarded message -----

    This archive was generated by hypermail 2.1.4 : Wed Jan 01 2003 - 14:01:01 EST