Review Request: make sure we don't find more textranges than we ask for

[Friedrich W. H. Kossebau]

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/107621/#review23117
-----------------------------------------------------------

Ship it!


Seems the proper fix to me, to match what the API dox promises.
For possible regressions from what I looked at and tried to imagine I could not yet see one, should be safe.

ToCGenerator::fetchBookmarkRef(...) using block.position() + block.length() to wrongly calculate the last position is a separate problem :)
KoDocumentRdf::findXmlId(...) is broken anyway (though I miss how your patch improves the situation here, follow-up on IRC for this).
KoTextWriter::Private::saveParagraph(...) looks safe as well to me.
DeleteCommand::doDelete(), hm, not sure where textEditor->selectionEnd() points to and if it is one-off? But separate problem.

BTW, as note for a general TODO for us: I have seen the last days that *To and *End have not always the same semantic, sometimes this refers to behind the last included position, sometimes to the last itself. Should cleaning this up be made a Junior Job bug item perhaps?


libs/kotext/KoTextRangeManager.cpp
<http://git.reviewboard.kde.org/r/107621/#comment17659>

    These three ifs could be merged as well.



libs/kotext/KoTextRangeManager.cpp
<http://git.reviewboard.kde.org/r/107621/#comment17660>

    While you are at it, could you also add a comment about the option to set matchLast to -1 to the API dox?



libs/kotext/KoTextRangeManager.cpp
<http://git.reviewboard.kde.org/r/107621/#comment17658>

    These three ifs could be merged into a single one.
    


- Friedrich W. H. Kossebau


On Dec. 7, 2012, 1:58 a.m., C. Boemann wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/107621/
> -----------------------------------------------------------
> 
> (Updated Dec. 7, 2012, 1:58 a.m.)
> 
> 
> Review request for Calligra and Friedrich W. H. Kossebau.
> 
> 
> Description
> -------
> 
> Make the textRangesChangingWithin method actually check all that it's supposed to check
>     
> Normally we don't use this method in a too complex way so there should be no regressions,
> but still if bugs come up with saving too few textranges, this will be a patch to investigate
>     
> Basically we were only checking the matching tag against one of the parameters while we should be
> checking against both. I most case the our use is to have global match limits, so that is why we
> have not noticed the bug.
> 
> 
> Diffs
> -----
> 
>   libs/kotext/KoTextRangeManager.cpp 0940fda 
> 
> Diff: http://git.reviewboard.kde.org/r/107621/diff/
> 
> 
> Testing
> -------
> 
> None, except the skf test case, and some debug utput to see the method actually does what I want it to.
> 
> However there may be some regressions (espcialy in saving), because of changed behavour
> 
> 
> Thanks,
> 
> C. Boemann
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/calligra-devel/attachments/20121207/3ee97ae6/attachment.htm>