Subject: re: comments on the BiDi patch
From: Tomas Frydrych (tomas@frydrych.uklinux.net)
Date: Sat Dec 30 2000 - 10:55:12 CST
Thanks Jesper,
I think I have managed to fix the typos and indentation you pointed
out so far, just a few comments/questions concerning more
substantial issues ...
>I wish you'd use an enum for the dominant direction type. Bool is
fine if
>you know the code intimately, but an enum makes for self
documenting code.
>Same goes for the return type of getVisDirection()
OK, where do you suggest to put the typedef?
>I'm not sure if this is the case, but won't
>fp_Line::_createMapOfRuns() end up having a monstrously big
>s_pMapOfRuns after a load/import?
>The reason I ask is that on load/import all Runs of a paragraph is
>stuffed into the same line, whereafter it is split up in mutiple lines
as
>necessary for the flow of the text.
>If that does indeed affect your code, you'll end up with a large
array (or
>large arrays in the non-static version) which is never fully used
after
>the initial load.
This would only happen if we have a very long paragraph with lot of
formating changes. I will though add a mechanism to shrink the
map if it is twice the default size and the current line in fact
contains less runs than the default would accomodate; this should
get rid of any huge maps without forcing unncessary delete/new all
the time. The default size of the map is 80; currently the map
element is UT_sint32 (I could change this to a signed character, if
desired ...)
>+ //if we follow a run that is consistent with the
>+ direciton of the para, then so will be this //(dont have
>+ to worry about the i-1 since this will never happen on
>+ position 0 if(s_pMapOfRuns[i-1] == 0)
>+ s_pMapOfRuns[i] = 0;
> Why? Can't a space be the first thing on a line?
Yes, it can, but if the first run is directionally neutral, the code
that precedes this has already set its direction to the dominant
direction of the paragraph; I will expand the comment to make this
clear.
Further comments are eagerly awaited.
Tomas
This archive was generated by hypermail 2b25 : Sat Dec 30 2000 - 10:59:59 CST