Re: 5291 RTF import slow

From: Hubert Figuiere (hfiguiere@teaser.fr)
Date: Wed Sep 24 2003 - 03:52:34 EDT

  • Next message: Mike Nordell: "Re: 5291 RTF import slow"

    On Wed, 2003-09-24 at 04:10, Dom Lachowicz wrote:
    > > Result: it added a minute to the import time (5:34
    > > vs.
    > > 4:39). So much for that idea ;-)
    >
    > New result - s/UT_XML_stricmp/strcmp in a bunch of
    > places. These are properties we're talking about -
    > they should all have the same case and we are
    > standardized on a case - lowercase. This will probably
    > break some compatibility with reading abiword 0.1.0
    > files or therabouts.
    >
    > 3:40m vs. 4:39m on my machine, or about a 21%
    > improvement. I know, it's another band-aid over the
    > problem. Getting it much better than this will
    > probably necessitate a rewrite of a nice chunk of the
    > PT.

    Something we should investigate is whether we often compare property
    names from inside the PieceTable or beetween PieceTable and stuff loaded
    from the file.
    The idea is to use "token" inside the piecetable (think numeral
    constants or enums) and only compare that. After that we just need to
    associate a string property name with a constant. BTW, that would
    probably save some RAM too as a 4byte int is smaller than a string.

    Just my .02 EUR

    Hub

    -- 
    "<erno> hm. I've lost a machine.. literally _lost_. it responds to ping,
    it works completely, I just can't figure out where in my apartment it
    is." --  http://www.bash.org/?5273
    


    This archive was generated by hypermail 2.1.4 : Wed Sep 24 2003 - 04:07:22 EDT