| boemann |
Rebasing was done using parameters from QTextDocument::contentsChange() signal incorrectly.
This caused the markup to be displayed in wrong position, and sometimes in an assert
in KoTextBlockData::appendMarkup().
This patch depends on that a simple edit is always the same as adding one character.
I found this method the easiest way to trigger the assert:
Add a (long) line with spelling mistakes.
Save to file.
Open file and position at the end of the line.
Quickly add a few words.
| No Linters Available |
| No Unit Test Coverage |
Not quite
simple edit could also be simple deletions and removes iirc
it seems to me that your assumption should at least be tested. ie added==1 && removed ==0, before just rebasing with +1
Iirc simple editing is to just move the other words after so they are still placed correctly
then later when the cursor leaves the word that word is spellchecked (possibly the entire block at that pooint, though i can't remember)
Camilla Boemann skrev den 2017-08-10 14:54:
boemann added a comment.
View Revision [1]Not quite
simple edit could also be simple deletions and removes iirc
No, I don't think so. Looking at the code (and testing) shows
startingSimpleEdit() is called only on insert character.
it seems to me that your assumption should at least be tested. ie
added==1 && removed ==0, before just rebasing with +1
Yeah, thought so too, but it seems that
QTextDocument::contentsChange(from, removed, added) is emitted with:
from: always 0 (or it might be
removed: length of complete text before edit
added: length of complete text after edit
So, imho useless, but I might have missed a trick.
To make it a little bit more robust in case of upstream changes, I could
check:
added == removed + 1
Hmmm, I think a number of asserts would be the thing, to detect upstream
changes.
Iirc simple editing is to just move the other words after so they are
still placed correctly
then later when the cursor leaves the word that word is spellchecked
(possibly the entire block at that pooint, though i can't remember)
Yes, the intention of this code is to keep the markup in the correct
place and to avoid spellchecking until a complete word has been entered.
But afaics the whole doc will be spellchecked, as the code iterates over
all blocks.
REPOSITORY
R8 Calligra
REVISION DETAIL
https://phabricator.kde.org/D7228
TO: danders, boemann
CC: vandenoeverLinks:
But afaics the whole doc will be spellchecked, as the code iterates over
all blocks.
I wouldn't be surprised no - there are issues like that for sure
And also on initiating a spellcheck when just loading the doc iirc