[00:51:04] Hello, all! We found a few bugs and were able to squash them, it seems. James did an emergency deployment. Some issues will persist until the on-wiki cache expires. [00:52:59] @Al, on recursion limits: the backend linearizes lists before processing them, so you don't usually see long lists hit the recursion limit. You could hit that limit if using a recursive composition to process the list, though. [00:54:32] As far as we could tell, the Z504 was due to a bad cached value for Z13518, which then affected every function using Z13518 :( :( :( [00:59:29] Here is an example of the issue using a brand new test on an existing function for the new long-list chemical element type: Z27960 (re @wmtelegram_bot: @Al, on recursion limits: the backend linearizes lists before processing them, so you don't usually see long lists hit t...) [01:22:29] Ahh, yeah. That would be the case of a recursive composition. Still, we *should* be able to address that case. The near-term solution will be to increase our recursion limit, which I think we'll be doing soon. Long-term, the obvious solution is tail recursion, which we've whispered about but haven't acted on yet. (re @Al: Z27960) [01:59:56] From a quick look I can't tell where the recursion is in this composition. The implementation seems to boil down to Z26879 and Z26908. If we can identify the recursive call, we may be able to work around it. (re @wmtelegram_bot: Ahh, yeah. That would be the case of a recursive composition. Still, we *should* be able to address that case. The near-...) [02:01:26] Z26879 (https://www.wikifunctions.org/view/en/Z26879) is implemented with Z26881 which has a hardcoded list of elements [02:27:59] I get a similar message at Z27961 where I'm pretty sure I haven't relied on any recursion (in the frontend? at least). (re @wmtelegram_bot: Ahh, yeah. That would be the case of a recursive composition. Still, we *should* be able to address that case. The near-...) [02:29:15] https://tools-static.wmflabs.org/bridgebot/c5b637cc/file_73591.jpg [03:22:27] also in Z27965 (re @u99of9: I get a similar message at Z27961 where I'm pretty sure I haven't relied on any recursion (in the frontend? at least).) [04:44:54] I've traced the issue all the way back to a built-in function. Z27969 fails. David might want to take a look at that one. (re @u99of9: From a quick look I can't tell where the recursion is in this composition. The implementation seems to boil down to Z26879 and Z...) [20:28:50] Thanks for clarifying… that makes sense. It’s not clear to me why the type generator consumes the enumeration recursively, though. (re @wmtelegram_bot: @Al, on recursion limits: the backend linearizes lists before processing them, so you don't usually see long lists hit t...) [20:37:56] Thanks for fixing this so quickly. Z13518 is pretty fundamental! I have to admit it has never occurred to me to nudge a type… (re @wmtelegram_bot: Hello, all! We found a few bugs and were able to squash them, it seems. James did an emergency deployment. Some issues w...)