OpenAtlas

We had the great opportunity to show colleagues the new system, especially the configuration part, at an ACDH Python User Group meeting.
Feedback was positive, we received a lot of valuable suggestions and hints.
For the configuration we will most likely go into the direction of Python classes. This will replace the current dictionary approach which was kind of a placeholder anyway.

A minor update:

I finished with all classes, what's missing are type hierarchy functions, some more complex relations between classes and some special functions like displaying root places at entity tabs of type detail views.

So I thought it's time to reactivate the tests. I got the base functionality working, went through them and added a return # Todo: continue tests where more adaptions are needed.
With that I got a coverage of 64.47%, so it's still a way to go but at least we now have passing tests again to help catch major issues in current development. And they are much faster now ;)

Type hierarchies are implemented and test coverage is at 85% (also thanks to @Bernhard Koschiček-Krombholz).

• Copy and dynamic entity add (in forms) are working again
• Removed obsolete code like former managers and unused functions
• Continued with tests
  • We are now at 91% coverage
  • I added comments about reasons where they are still turned off

Minor update: most of the type and admin checks are working again and we are at 92.81% coverage.

Files, image and import tests are working: 96.81%.

• 100% code coverage again. 315 of 12,573 statements are excluded with nocover comment, so about 97.50% real coverage.
• Most Mypy issues resolved (26 remaining)
• I think I managed to solve Actor relations, feel free to test it at the development demo

• Display at tabs, e.g. showing actors connected via events in place detail views
• Populating forms connected via location to object_location (or the other way round) when linking entities

I managed to solve the last remaining Mypy issues.
Beside being one step closer to the major release it also makes it easier to spot new Mypy issues in current development.

As already visible in the demo-dev version (https://demo-dev.openatlas.eu/entity/50625), tabs that do not yet contain any entries are grouped under the “+ Relation” button and can be populated from there. However, it is unclear whether “+ Relation” is the most appropriate label, as it may not be immediately intuitive for users. A possible alternative would be to rename this button “+ Add”.

Next OpenAtlas major version is now developed in feature/trixie which is already based on Debian 13.

I looked into more detail into gettext functions which are used for our user interface translations.

gettext vs lazy_gettext¶

lazy_gettext is only needed when used in class definitions (e.g. form classes) that are not inside a function.
For everything else we should use gettext because of various minor issues with lazy_gettext otherwise, e.g. it seems PyCharm gets confused and gives a lot of type warnings (Mypy works just fine).
I already re-written everything accordingly and put it online on demo + demo_dev. Please report if you notice anything (keeping in mind that newest translations of 9.0.0 are still missing in general).

Performance¶

I did some testing with gettext about performance while at it: translating 200,000 one word translations takes about 1 second at my local machine.
So we don't have to worry about translation performance in general. Only exception being column labels in tables, which I will take a look at.

• I finished translating new terms, so if you notice anything at demo or demo_dev in English or German please report
• With the newest version of Poedit it is now possible to use translations in f-string, e.g. like:
  

  label = f'What a wonderful {_('world')}'