[11:49:05] Anyone in the mood for writing an easy function to configure lists of notable people into your language? Here's the config: Z47849. Most of you won't have to change much from the English composition. It's already getting some use, and is pretty relevant for lists of notable people associated with populated places. : [11:49:05] https://tools-static.wmflabs.org/bridgebot/e17f0d64/file_85244.jpg [13:55:23] The FAQ says "Currently, it is _not_ possible to call other functions from implementations written in any programming language. This is currently only possible through composition." Is it still true? [13:56:23] There are some things in the FAQ that are no longer true. E.g. "We hope to add at least one further programming language in 2025 (but have not yet decided which one)." [14:10:22] What's the problem with https://www.wikifunctions.org/view/he/Z53443 ? [14:11:20] It says "Argument type mismatch". The function expect two lexemes, and I'm sending it two lexemes. Am I missing anything? [14:16:12] Oh, now it seems to be fine. Failing for a good reason. Probably caught up to caching. [14:48:51] Is there something like "virtual lexemes"? Or "virtual forms"? I made https://www.wikifunctions.org/view/he/Z53432 , which takes two lexemes and returns a string that combines according to Hebrew spelling rules. The first lexeme is a one-letter Hebrew clitic word that is always written with the next word. These are words like "the", "that", "from", "and", etc. But it [14:48:51] should be ab [14:48:51] le to be recursive: to take a clitic and another word that already has a clitic and to combine them, to make something like "and the cat". The inner call combines "the" and "cat", and the outer call combines "and" and "the cat". [14:49:47] It's possible to do it all with strings, but then the metadata about the lexeme gets lost, and it's not great, because it can be useful for grammar, semantics, etc. [14:51:57] ... Not exactly. All the tests pass now when I'm editing the implementation, but fail when I'm viewing the function. The function page probably still sees an old version of the implementation. Is it the usual thing that everyone here is already used to?.. (re @amire80: Oh, now it seems to be fine. Failing for a good reason. Probably caught up to caching.) [19:48:09] @amire80 yes it is a particular irritation that may have arisen recently. (re @jsamwrites: This is a major issue.) [19:51:18] Yes strings for now. There have been some theoretical discussions about virtual lexemes, but they are a long way from finished. (re @amire80: It's possible to do it all with strings, but then the metadata about the lexeme gets lost, and it's not great, because it can be...)