Re: [Gramps-devel] Question Person Tree View

[Nick H. <nic...@ho...>]

Peter,

Yes, I get the same. When the pedigree view is not active it is not 
handling the nameformat-changed signal correctly.

I'll look at it this afternoon.

Nick.


Peter Landgren wrote:
> Did two more tests.
> 1.
> -  Switch to Pedigree View
> -  Change name format
> -  Switch to any other view with names (person, family, relationship, events)
> -  Check name format
> -  The format is correct.
>
> 2.
> -  Switch to any other view with names (person, family, relationship, events)
> -  Change name format
> -  Switch to Pedigree View
> -  Check name format
> -  Name format is still the previous one
>
> When I switch to  Pedigree View after a "change-name-format signal" the Pedigree View is not re-
> built. The other views are.
>
> /Peter
>  
>
>   
>> Tested it, but the pedigree view is not updated.
>>
>> /Peter
>>
>>     
>>> I  have created a bug report (#3691) and committed a fix for this -
>>> r14736 (gramps32) and r14737 (trunk).
>>>
>>> A re-build is now called for all views displaying names.  When a view is
>>> not active it will just mark the view as 'dirty'.
>>>
>>> This works well with the split views too.  In the prototype, there is a
>>> concept of an active workspace rather than an active view.  The
>>> workspace is responsible for setting the appropriate views to be active
>>> when they are displayed.
>>>
>>> I don't really have a concept of a view with focus, but I do have a
>>> primary view within the workspace which provides the main ui definition.
>>>
>>> Regards,
>>>
>>> Nick.
>>>
>>> Benny Malengier wrote:
>>>       
>>>> 2010/3/10 Peter Landgren <pet...@te...>:
>>>>         
>>>>> Nick,
>>>>>
>>>>> Changing name formats without a complete re-build/re-draw of the
>>>>> different views could be very confusing for a user. I think that we
>>>>> should do this, as you normally  don't change name format that often.
>>>>> At least I don't change it very often. The format reflects a personal
>>>>> preference. Maybe somehow indicate it will take some time.
>>>>>           
>>>> It should suffice to set the non-active views with names dirty on this
>>>> name-format signal (or you get a big performance hit). If active, then
>>>> rebuild seems in order.
>>>>
>>>> This makes me think about the experimental split views in the bug
>>>> tracker. Then there are two active views, and one view with focus....
>>>> all becomes a bit more complicated :-)
>>>>
>>>> Benny
>>>>
>>>>         
>>>>> /Peter
>>>>>
>>>>>           
>>>>>> Peter,
>>>>>>
>>>>>> I have fixed the multiple re-build problem with commit r14724
>>>>>> (gramps32) and r14725 (trunk).
>>>>>>
>>>>>> The other problem arises because only the active view is re-built
>>>>>> when the name format is changed. The other views which display names
>>>>>> will still display the old name format.
>>>>>>
>>>>>> The list views don't hold all the data in the table in memory. When
>>>>>> you scroll the view will get rows not in memory from disk. These rows
>>>>>> will be in the new name format, hence the mix of name formats in a
>>>>>> single view.
>>>>>>
>>>>>> One solution to this problem would be to re-build all views
>>>>>> displaying names when the name format changes. This would take some
>>>>>> time though because there are quite a few views with names: person
>>>>>> flat, person tree, relationship, pedigree, fanchart and family. I
>>>>>> suppose we should mark these views as 'dirty' and re-build them when
>>>>>> they become active.
>>>>>>
>>>>>> In a related issue the pedigree view can display a mix of name
>>>>>> formats. The pedigree view does not re-build in response to a
>>>>>> nameformat-change signal. Names in existing boxes are cached and not
>>>>>> refreshed. When the user navigates to new boxes they will display the
>>>>>> new name format, but cached boxes will display the old name format.
>>>>>>
>>>>>> Did you open a bug for this problem?
>>>>>>
>>>>>> Regards,
>>>>>>
>>>>>>
>>>>>> Nick.
>>>>>>
>>>>>> Peter Landgren wrote:
>>>>>>             
>>>>>>> Nick,
>>>>>>> If you start Gramps in Person Tree View it seems to be OK.
>>>>>>> If you
>>>>>>> - start Gramps in Person List View
>>>>>>> - do a name format change
>>>>>>> - switch to Tree View
>>>>>>> - do name format changes
>>>>>>>  this effect pops up.
>>>>>>>
>>>>>>> Note this happens both in 3.2 and trunk.
>>>>>>>
>>>>>>> /Peter
>>>>>>>
>>>>>>>               
>>>>>>>> Peter,
>>>>>>>>
>>>>>>>> Peter Landgren wrote:
>>>>>>>>                 
>>>>>>>>> I only changed the name format.
>>>>>>>>> Given Name Surname -> Given Name
>>>>>>>>> results in three re-builds.
>>>>>>>>>                   
>>>>>>>> This is what I thought you were doing.
>>>>>>>>
>>>>>>>>                 
>>>>>>>>> Any name format changed result in two or three re-builds.
>>>>>>>>>                   
>>>>>>>> I can't reproduce this.  Does it happen every time?
>>>>>>>>
>>>>>>>>                 
>>>>>>>>> I also run into another effect:'
>>>>>>>>> Two different name formats in the same display.
>>>>>>>>> See attached image.
>>>>>>>>> However, I'm not sure I can repeat it. The effect disappeared
>>>>>>>>> after some sorting.
>>>>>>>>>                   
>>>>>>>> This looks like the LRU cache not being cleared.  I can understand
>>>>>>>> why you may not be able to reproduce this one.  I'll look into it.
>>>>>>>>
>>>>>>>> Regards,
>>>>>>>>
>>>>>>>>
>>>>>>>> Nick.
>>>>>>>>
>>>>>>>>                 
>>>>>>>>> /Peter
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>>>> No, it isn't intended for it to re-build more than once.
>>>>>>>>>>
>>>>>>>>>> What are you changing?
>>>>>>>>>>
>>>>>>>>>> Nick.
>>>>>>>>>>
>>>>>>>>>> Peter Landgren wrote:
>>>>>>>>>>                     
>>>>>>>>>>> Hi,
>>>>>>>>>>>
>>>>>>>>>>> When I change the way names are displayed, Gramps always
>>>>>>>>>>> rebuilds the "Person Tree View" twice. I even think I saw it
>>>>>>>>>>> happens three times once.
>>>>>>>>>>>
>>>>>>>>>>> Is this correct?
>>>>>>>>>>>
>>>>>>>>>>> /Peter
>>>>>>>>>>>                       
>
>
>
>