[00:09:05] Me neither. Maybe it helps prevent us from making broken objects? (re @David: Oh, right! Thanks for reminding me and clarifying. I sort of knew about that limitation, but I've almost never stumbled across...) [00:09:57] No concerns here, nor even forseeable risks. (re @David: Thanks, Al! Good points. Regarding T411947 - that's a simple task and i think it's a no-brainer – just ensuring there's alway...) [00:20:38] No… apart from providing meaningful tests for the fetch functions, it’s not really been an issue. No one wants to construct a whole item or lexeme! (re @David: Oh, right! Thanks for reminding me and clarifying. I sort of knew about that limitation, but I've almost never stumbled across...) [00:29:05] I searched for set type and functions and found nothing today. Are they missing? (re @u99of9: Conflating ordered lists and unordered sets will eventually cause headaches.) [01:01:03] Truly. In the coming quarter the team will do some work allowing for copying a portion of a composition (ZObjects, of course) and pasting it in another location in a composition. I'm hoping that will lead, sooner or later, to an ability to copy the output of a function call and paste it as the expected value in a test. But I can't say how soon. (And one can imagine [01:01:03] even quicker [01:01:03] ways to automatically populate a test with a complex expected value.) (re @Al: No… apart from providing meaningful tests for the fetch functions, it’s not really been an issue. No one wants to construct a wh...) [01:05:20] Right. The team has considered making a built-in typed-set type and related functions, but so far it hasn't become a high priority. (re @Npriskorn: I searched for a typed set and related functions (add/remove/union/difference/complement/symmetric difference) and found nothing...) [01:32:11] no haste, we have so much else to play with already 🤩 (re @David: Right. The team has considered making a built-in typed-set type and related functions, but so far it hasn't become a high prior...) [01:47:04] I'm curious if anyone else have had the need for OR-ing an input? Z30573 [01:47:04] We need either side length or radius for this. [09:16:16] It sounds like multilingual texts are thought of like a typed set. Changing that definition will cause a lot of work if we are not careful. The longer we wait the more it will take. (re @David: Right. The team has considered making a built-in typed-set type and related functions, but so far it hasn't become a high prior...) [09:33:20] To add more context to this: this feature will indeed allow to copy an output of a function and paste it somewhere else. However, Wikidata literal entities have a special behavior. We disallow creation of Wikidata literal objects, to always keep Wikidata as the source of truth for all Wikidata entity information. We can of course visualize and use the output of a [09:33:20] function call wh [09:33:21] ich returns a Z600* object, but not create it nor persist it, only through function calls or references that dynamically fetch and resolve them. [09:33:22] This will be the same in the context of the new Copy+Paste feature, while it will be possible to copy an output and paste it as a part of your composition or test call (or anywhere else), there will be a number of restrictions in place to keep enforcing all system-wide restrictions (e.g. type matching, restrictions on root-level objects, or avoid copying wikidata [09:33:22] literals) [09:33:24] It sounds like that the only reason for wikidata entity creation would be for testing fetch functions or other functions that return wikidata items, which makes sense. There are ways around it, but if it becomes a problem, we should approach the issue with that particular use case in mind and maybe come up with a different set of helpers or a differently [09:33:24] pre-configured test call/ [09:33:25] validator keys when the output is a Wikidata literal... (re @David: Truly. In the coming quarter the team will do some work allowing for copying a portion of a composition (ZObjects, of course) a...) [09:43:44] The longer we wait the more it will take. (re @u99of9: It sounds like multilingual texts are thought of like a typed set. Changing that definition will cause a lot of work if we are n...) [10:17:06] When we make a change to an existing type, the effort is roughly proportional to the number of functions that already use that type. (re @Npriskorn: Would you be willing to elaborate this? [10:17:06] The longer we wait the more it will take.) [10:30:24] Yes, I found out the hard way that it's not that nice an experience to change a functions input/output types after creating tests and impl.