[KPhotoAlbum] Category view problem

[Miika Turkia]

On Sun, Feb 17, 2013 at 11:55 PM, Robert Krawitz <rlk at alum.mit.edu> wrote:
> On Sat, 16 Feb 2013 22:28:20 -0500, Robert Krawitz wrote:
>> On Sat, 16 Feb 2013 18:14:09 -0500, Robert Krawitz wrote:
>>> I'm having a strange problem with category view.
>>>
>>> I frequently keyword every shot I take at an event.  I then select the
>>> ones I'm interested in, and keyword them with the same name, followed by
>>> "selection".  So I might have keywords of
>>>
>>> MIT-Babson Jan. 12, 2013
>>>
>>> and
>>>
>>> MIT-Babson Jan. 12, 2013 selection
>>>
>>> The latter is of course a subcategory of the former.
>>>
>>> The problem is that in category view, the number of photos in the parent
>>> category is always shown as either exactly the number of photos in the
>>> selection subcategory, or twice that number.  If I then click on that
>>> row, the correct number is displayed.
>>>
>>> It appears, though, that it isn't the name that causes this (if I
>>> keyword the subselection XXXX the same thing happens).
>>>
>>> But there's another, much more serious problem.  If I select all the
>>> photos in the subselect category, and type ctrl-2 to tag them all, I
>>> cannot selectively remove them from just the subcategory or the
>>> supercategory.  Initially, both subcategory and supercategory are
>>> checked.  If I check or uncheck either one, both are shown as checked or
>>> unchecked.
>>>
>>> If I don't make the selection a subcategory of the main keyword, I don't
>>> have this problem.
>>>
>>> In general, it appears that if there is a category/subcategory
>>> relationship, I cannot use ctrl-2 to change just one of the tags on
>>> images.
>>
>> There are actually two problems, based on git bisect.
>>
>> The first problem (showing bad counts) is:
>>
>> a54dfe608536ae9af2cd2a8407fd02af63cec6f6 is the first bad commit
>> commit a54dfe608536ae9af2cd2a8407fd02af63cec6f6
>> Author: Jesper K. Pedersen <blackie at kde.org>
>> Date:   Fri Jan 4 21:35:35 2013 +0100
>>
>>     s/Q3Dict/QHash for countedGroupDict
>>
>> :040000 040000 b3433e714a55b8cade49e21a3d32b3d6e28ee81e 6e4965242e91f5bfbc705e4e38bf370400b36805 M      DB
>>
>> I haven't figured out what's going on here.
>
> The double negative problem strikes again.
>
> diff --git a/DB/GroupCounter.cpp b/DB/GroupCounter.cpp
> index 7cf04ee..cb5bc87 100644
> --- a/DB/GroupCounter.cpp
> +++ b/DB/GroupCounter.cpp
> @@ -95,7 +95,7 @@ void GroupCounter::count( const StringSet& categories )
>              }
>          }
>          // The item Nevada should itself go into the group Nevada.
> -        if ( !countedGroupDict.contains( *categoryIt ) == 0 && _groupCount.contains( *categoryIt ) ) {
> +        if ( !countedGroupDict.contains( *categoryIt ) && _groupCount.contains( *categoryIt ) ) {
>               countedGroupDict.insert( *categoryIt);
>               (_groupCount[*categoryIt])++;
>          }

So is this a fix to the whole issue? Ready to be pushed to master?

miika