RFE:Delete comment should use Delete2

[atarkowska]

When we remove a Comment from P/D/I, we also have to check if it's linked to anything else, and if not then delete it.

Try using Delete2 instead.

Mark:
"I /think/ if you give a child option with excludeType as the list of all the link types (ImageAnnotationLink?, etc.) then I think it will fail if there is such a link.

http://www.openmicroscopy.org/site/support/omero5.1/developers/Server/GraphsMigration.html#examples-in-python
where we have,
keepAnn = [ChildOption?(excludeType=Annotation?)]
make that a list of ImageAnnotationLink? and all the other link types
"

[atarkowska]

Deleting command could also handle removing links while comment is linked to more then one objects

[jmoore]

Ola, do we need to add an RFE for the delete service, or is this a bug in your opinion?

[atarkowska]

Certainly, it would be easier to use server side method specific for that usecase. Currently what was fixed this by using 'remove' method implemented on the client side, see #5325. As Josh suggested, I will link it to API evolution #2656

[jmoore]

Is there anything someone needs from me for this?

[atarkowska]

The logic is already implemented in the python gateway. Would be good idea to push it to the server in order to serve the same functionality across both clients.

[jamoore]

Again, IDelete is now gone. See e.g. ​https://github.com/openmicroscopy/openmicroscopy/pull/3729

Is there any further cleanup to be done in the gateway?

[mtbcarroll]

With luck, Delete2 on the comment now addresses the motivating performance issue.

[jamoore]

This needs some discussion with Mark to know if all the hacks are now redundant.

[mtbcarroll]

(Also perhaps with Colin.)

[wmoore]

Since other Delete work is getting pushed to 5.1.2 ​https://github.com/openmicroscopy/openmicroscopy/pull/3725 makes sense to do same for this.

[wmoore]

Alternative is to use regular Delete2 command to delete the Comment with dryRun=True and see if the response lists any other comment annotation links.
Mark:
"easiest just to do a normal Delete2 where you set dryRun=True and target a CommentAnnotation? that is still linked to something and have a look at the response -- that should make the rest obvious"

[mtbcarroll]

Josh gives me the idea that it might work to simply have

excludeType=['IAnnotationLink']

it's worth a try. (I could test it myself next week if you like.)

[mtbcarroll]

It doesn't work, for tricky reasons that aren't easy to adjust. I suggest going the dryRun=True route for now and checking the response for any links.

[mtbcarroll]

(excludeType doesn't work because, as the ​http://downloads.openmicroscopy.org/omero/5.1.0/api/slice2html/omero/cmd/graphs/ChildOption.html documentation reminds me, these options are about how to treat children; while it makes sense to talk about orphan images or orphan annotations because ILink.child takes them as values, the links themselves aren't usually themselves considered as being children.)

[wmoore]

Tried using
keepAnn = [ChildOption(excludeType=['ImageAnnotationLink'])]
but this doesn't seem to prevent deletion of the Comment itself.
I guess that's because Image isn't a child of Comment?

This test fails, since Comment is deleted.

    def testDeleteComment(self):
        comment = omero.model.CommentAnnotationI()
        comment = self.update.saveAndReturnObject(comment)
        images = list()
        for i in range(0, 3):
            img = self.new_image(name="testDeleteComment")
            img.linkAnnotation(comment)
            images.append(self.update.saveAndReturnObject(img))

        assert self.query.find("CommentAnnotation", comment.id.val)

        keepAnn = [ChildOption(excludeType=['ImageAnnotationLink'])]
        command = Delete2(targetObjects={"CommentAnnotation": [comment.id.val]}, childOptions=keepAnn)

        handle = self.client.sf.submit(command)
        self.waitOnCmd(self.client, handle)

        assert self.query.find("CommentAnnotation", comment.id.val)

[wmoore]

Oops - sorry, missed your comments above!
Trying dryRun=True now.

[wmoore]

Fixed ​https://github.com/openmicroscopy/openmicroscopy/pull/3791