Ok, so I've fixed that bug where spellcheck & images don't play nicely with
one another, or at least in the cases that we discussed. Right now, as far
as the spell-checker is concerned, images don't exist. This was a 3 line
change. I'll locate and close the bug(s) in bugzilla.
Why did we get the bug in the 1st place? See fl_DocLayout.cpp around line
340 for some "suspicious" code. We insert UCS_ABI_OBJECT into the stream as
a placeholder for various reasons. Our problem was that we were also
spell-checking this character in our spell-checking routines.
So I added logic in fl_BlockLayout::_checkMultiWord to strip ABI_UCS_OBJECT
from the text being selected for spell-check.
Now we get a more interesting bug, as far as I'm concerned (this bug has
always happened to us): commence[IMAGE]ment isn't squiggled because
"commencement" is a legal word and images don't count. Should this be our
behavior or should this be treated (and spell-checked) as two words:
"commence" "ment" (thus marking commence as correct, but ment as
misspelled)? What about this counter-case: commence[IMAGE]mint
("commencement" was the intended word, but "commence" and "mint" are both
legal words by themselves)?
I honestly don't know what we should do, but either way it's easy to
implement once we reach some sort of consensus. Our behavior now is to mark
he[IMAGE]llo as a correctly spelled word. Screenshot of the bug at
http://198.139.111.101/abiword/img-spell.jpg
Dom
CVS: ----------------------------------------------------------------------
CVS: Enter Log. Lines beginning with `CVS:' are removed automatically
CVS:
CVS: Committing in .
CVS:
CVS: Modified Files:
CVS: fl_BlockLayout.cpp
CVS: ----------------------------------------------------------------------
_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com
