#5072 BLOC Update.: Activities dont load in 640

[Zarro Boogs per Child]

#5072: Activities dont load in 640
------------------------+---------------------------------------------------
  Reporter:  yani       |       Owner:  tomeu            
      Type:  defect     |      Status:  new              
  Priority:  blocker    |   Milestone:  Update.1         
 Component:  datastore  |     Version:                   
Resolution:             |    Keywords:  Update.1? review?
  Verified:  0          |  
------------------------+---------------------------------------------------

Comment(by tomeu):

 Replying to [comment:11 marco]:
 > Spent some time tracking this down, here are my findings and thoughts.
 >
 >  * Tomeu, we configure the model according to the database but we don't
 apply the changes immediately (like it's done a few lines above for the
 default model properties by datamodel.apply). Because of this the source
 property error can be fired at any time and the problem becomes hard to
 reproduce. I think it explains why we are not able to reproduce yani
 crash. If I apply the changes immediately, the mount fails consistently as
 expected.

 Model.apply() looks to me like it syncs from Model to Xapian, isn't it? If
 we are sync'ing from Xapian to Model, then we don't need to call apply().

 >  * I'm not convinced Kim/Scott problem is the same of yani. It don't see
 how it could happen on a clean datastore. I think it might be #5127 or
 something else entirely (possibly related to the custom property changes).
 We need logs to be able to do anything about it.

 Yeah.

 >  * For Ship.2 I'm likely going to recommend to *not* apply the custom
 property patch. Source should not be a string in first place, we can keep
 defining it as "text" in model.py until it's worth to break model
 compatibility. Anyway we should think a bit more about it tomorrow morning
 and make a call in the Journal meeting.

 Perhaps that's the best thing we can do, but I'm a bit worried of setting
 source as "text", as this will be indexed, reducing performance in some
 unknown measure. 'string' is wrong but 'text' looks definitely worst to
 me.

-- 
Ticket URL: <http://dev.laptop.org/ticket/5072#comment:12>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system