Agenda

1. Currently major changes which Carlos might consider to back port goes through email. Lets use GitHub and Pull Request.
2. #6826 remove listGroups.
3. listChildren/listParents does not link loaded sequence to the root object. We should verify other methods how extra information is linked to the root object. Not linking might be too expensive, especially when people call the same methods many time in the lifetime of the object. The question is if we want to do it or not? #6911 ExperimenterWrapper and ExperimetnerGroupWrapper need own listChildren and listParents
4. #6752 webgateway caching.
5. How to rationalize JS libraries #6466 and #6465.

Attending

Carlos, Chris, Ola, Will

Notes

1. Currently major changes which Carlos might consider to back port to 4.1.custom goes through email. We could use GitHub to share those information. We could push 4.1.custom but it will be harder for Carlos if he has a lots of pull requests. There is no way to filter pull requests per branch, might be tricky for Carlos to filter it. If anyone sends pull requests to O.org repository then Carlos gets email.

   Action: Carlos is going to push 4.1.custom to his own repository, but not invite others (we do not want to push there). Then we will pull requests against o.org what gives Carlos nice overview.

2. #6826

   Action: remove listGroups. post 4.3.3

3. #6911

   • Naming specific type of Children or Parent is fairly accepted. List Parents and Children should be only available on top hierarchy data objects (Project/Dataset/Image or Screen/Plate/Well). For other object wrappers we should use alternative solution, existing methods directly from omero model or proper naming based on class names. Otherwise is risky and makes confusion. Carlos is not ready to remove listChildren/listParents from the superclass wrapper. If we need it in the specific classes we can add them directly. We can just use omero model linkedExperimenter(Group)List and add it to the Wrapper directly. This will stick to the model.
     
   • Loading map links. There is no standard procedure for loading map and storing it. Mostly object is loaded without map and then map is loaded on demand every time method is called (Dataset.listChildren will make a query to the server every time). We have few deviations. They should be documented then. Originally Carlos's idea was to cache what is too expensive for several calls. Generators are used frequently to prevent holding HTTP requests (used in Templates).

   Action: We should declare, there is no caching by default. In places where this is already done, just document it. post 4.3.3

4. #6752 webgateway caching.

   Action: post 4.3.3

5. How to rationalise JS libraries #6466 and #6465.

   Action: JS, Djanog upgrade MUST be done as first thing post 4.3.3.

6. Other tickets:

   • multiple selection on well does not correspond with metadata panel while using Shift and Ctrl.
   • not every time entire image graph is being put in the share, noticed on the course.