MIF queries

[jamoore]

In order to implement the proper delete/chgrp actions described in #10780 et al., it may be easier to detect server-side which images have siblings in a separate container and provide this information to the clients.

Logically, this means:

• pass in a container (e.g. Dataset id)
• find all the images in the dataset (CONTAINED)
• remove those which are either:
  • not in a MIF, or
  • the entire MIF is in the container
• call this set SEPARATED
• find other containers of the same type that contain images from SEPARATED.

The proper name for this type of query is still unclear. "find separated siblings" might work. This is related to findContainerHierarchy and loadContainerHierarchies.

[mtbcarroll]

Perhaps this new method should be offered alongside those listed in IContainer.ice and implemented by PojosImpl.java.

[jburel]

actually adjusting some calls from the existing methods in Icontainer will be required so the graph was be nicely displayed in client

[mtbcarroll]

What if some of the images in SEPARATED are orphaned?

[jamoore]

Then I would expect images to be returned as with ​findContainerHierarchies

[mtbcarroll]

Ah, return a set with both containers and images in it? (Basically, a container if a separated image is in it, and an image if the image is orphaned?)

[jamoore]

Right. findContainerHierarchies returns a collection of either Projects Datasets or Images. The highest one in the hierachy possible.

[wmoore]

What I'm doing in the web just now when chgrp fails is to show the user how they attempted to 'split' each fileset: see ​https://github.com/openmicroscopy/openmicroscopy/pull/1138

If I were to provide this same UI before they hit the chgrp "OK" button, then I'd like to know which filesets would be split, and which images would be in/out.

E.g.

filesets = listSplitFilesets("Dataset", [123, 456])          # Trying to move 2 datasets

#  { 1: { true: [3, 4, 5],  false: [6, 7, 8] },   11: { true: [ 20, 21], false: [ 22 ] } }

In the example, 2 filesets (1 & 11) would have been split if I tried to chgrp 2 datasets. Image IDs are true: (those inside the 2 Datasets) and false: Not included in Datasets.

NB: If a fileset is split between the 2 Datasets, it should not be returned by this query.

[jamoore]

Wasn't needed during FS demo 4.1

[mtbcarroll]

Would it be enough to know which images are missing (need to be added to avoid split filesets) or do we actually need an excitingly structured return value?

[wmoore]

I'm just saying that this is what we're currently displaying to users in the web UI. If you don't give me the missing AND included image IDs then I've got to load the filesets and their images to get that data, which is possible but not as nice.

[mtbcarroll]

I'll indeed try to provide something that is pretty much what you suggest then.

[jburel]

Moved from sprint FS demo 4.2

[mtbcarroll]

fixed by ​https://github.com/openmicroscopy/openmicroscopy/pull/1214 but leaving open so that the accompanying integration tests aren't forgotten

[wmoore]

Python tests (all passing) are in ​https://github.com/openmicroscopy/openmicroscopy/pull/1217

[Will Moore <will@…>]

(In [9f269cd9ac0f1051c71709869ed39bcba9f6712d/ome.git] on branch develop) Various fileset query tests. See #10782

[Josh Moore <josh@…>]

(In [424a167af0fc9957cbc0e7c02d2f3cb6797a0c06/ome.git] on branch develop) Merge pull request #1217 from will-moore/split_fileset_query_tests_10782

Various fileset query tests. See #10782

[mtbcarroll]

The accompanying integration tests I had in mind in comment 17 are now in ​https://github.com/mtbc/openmicroscopy/commit/67629c8fe6d62cfc8b393e85a928ce8ef6034e39 and more will follow from #11013.

[Mark Carroll <m.t.b.carroll@…>]

(In [d1c4ec7580f40171f92b3aee1f440f78742a1368/ome.git] on branch develop) fix #10782: add IContainer.getImagesBySplitFilesets

[Josh Moore <josh@…>]

(In [ba6e48595b876c67d99a247eb8d115184a86c12e/ome.git] on branch develop) Merge pull request #1214 from mtbc/trac-10782

fix #10782: add IContainer.getImagesBySplitFilesets