OMERO.tables ArrayColumns #5632 (rebased onto develop) #728
Conversation
This needs reviewing by someone who knows about NumPy-Ice array conversions
…g a single recarray from the values and dtypes of each column in one go, instead of each co lumn creating its own numpy.array
AbstractColumn.__init__ uses properties such as self.name which aren't always set at initialisation, especially when Ice is involved. This commit removes self.recarrtype and replaces it with just the properties that should exist (based on testing since its unclear under what circumstances Ice will set a property or not).
Mask creation and test functions also put into separate functions for reuse within test.
Test for a array column of size 1. Test all possible column types in the same table.
Now that array columns seem to be working I can start to clean things up.
This replaces the need for a separate variable to indicate initialisation
This makes things logically nicer (new stuff goes at the end of the table)
Also moved HDF5 upload from class initialiser to the actual test method. This should make things easier if addition test files for future versions are needed.
…Table The OriginalFile ID is useful for compatibility tests, as it allows the underlying file to be manually retrieved
|
Tested locally with Python 2.6.
The baseline Python version we are currently supporting is 2.6, so the tests have to keep to 2.6 syntax. |
|
@bpindelski: at the moment Python 2.5 is still the minimum requirement. even if the http://www.openmicroscopy.org/site/support/omero4/sysadmins/system-requirements.html |
|
Added ticket http://trac.openmicroscopy.org.uk/ome/ticket/10403 |
|
@manics: is your intent to fix 10403 in this PR? |
|
@manics 👍 for the ticket. As this is a rebase, I'd vote for fixing it in a separate PR. Once it's open, we can merge this in. |
|
@joshmoore It needs to go into dev_4_4 and develop, so separate PR. |
|
Agreed. |
|
Now incorporating PR #753 (since AFAIK I can't open a PR which changes another PR). |
|
@manics: you can open a PR against your own PRs, but this is easily cleaner. Leave it for @bpindelski to give a thumbs up tomorrow. |
|
@manics, @joshmoore This fixes the failing |
|
👍 |
…n_5362-dev_4_4 OMERO.tables ArrayColumns #5632 (rebased onto develop)
This is the same as gh-538 but rebased onto develop.
Add support for Long and Double arrays in the columns of OMERO.Tables, previously only atomic elements and masks were supported.
Integration tests have been updated.
Docs/examples will be dealt with in #9957
This should be 100% backwards compatible with the current API and existing tables.
Also incorporating rebased #753 (Trac #10403)
--rebased-to #753
--rebased-from #538