[10:16:34] Thats cool, thank you 😊 (re @vrandecic: I am randomizing them manually about once a day. If there is a way to randomize it automatically per view, that would be awesome...) [10:19:18] I vaguely recall a clever system that use the current time as a seed (re @harej: I guess you could put in a seed as the input, but it would result in the same exact randomness string every time. Then you'd hav...) [10:41:11] Guess how to get a true random seed? (re @NicolasVIGNERON: I vaguely recall a clever system that use the current time as a seed) [10:42:42] The cleverest system is a system from the nature! (re @NicolasVIGNERON: I vaguely recall a clever system that use the current time as a seed) [10:45:02] [[c:File:Lava_lamp_wall_at_Cloudflare_office_-2.jpg]] [10:45:54] This is the answer by Cloudflare on the question of "a genuinely random seed". (re @wikilinksbot: [[c:File:Lava_lamp_wall_at_Cloudflare_office_-2.jpg]]) [11:25:31] Thanks. I just nudged the caches on the two tests I suggested, and they pass the tests. I'm not aware of any problems yet, but will try to further investigate going forward. Can you now also enable Z28866 and Z28845? They will also improve our representation of the special values, bringing them into alignment with the binary machine representations as far as possible. [11:25:31] If they mes [11:25:31] s anything up, feel free to revert. (re @vrandecic: @u99of9 I tried switching to your float implementation for the subnormals. Looks good so far, some tests are stuck in cache for ...) [12:05:15] Yes, I confirm that every test I had labelled with the word "subnormal" now passes after a cache nudge. I think we can call that a success. (re @u99of9: Thanks. I just nudged the caches on the two tests I suggested, and they pass the tests. I'm not aware of any problems yet, but w...) [12:07:12] This one is satisfying: Z21923 (re @u99of9: Yes, I confirm that every test I had labelled with the word "subnormal" now passes after a cache nudge. I think we can call that...) [12:24:15] I agree with points 1-3, and have tried to implement them all where possible. Point 4 sounds like either never applicable (quantities are never for display, they represent calculations), or a graphic designer's answer, or post-hoc justification of leaving off the 0. I teach 2 as "Don't add *OR SUBTRACT* precision you didn't start with, so I think point 4 is wrong. (re@ [12:24:15] Al: That [12:24:15] ’s not the exact conversion from kg to oz. 1 kg = 35.2739619495804 oz, which gives 7.75921341. I asked it how to round this ...) [12:29:46] Yes, I got the same impression, but it is good to challenge these things so you can prompt it to change its “beliefs”. (re @u99of9: I agree with points 1-3, and have tried to implement them all where possible. Point 4 sounds like either never applicable (quant...) [12:31:40] I guess it depends what you're doing it for. I'd hate to have to have an argument about significant figures every time I wanted a calculation done. [12:34:04] Fair point, but I think you can just prompt it occasionally now. Still not as trustworthy as a deterministic function, of course. [15:22:23] It's randomized now! Thanks to a comment by NguoiDungKhongDinhDanh I found the shuffle template (re @harej: I thought maybe the entries would be randomized but if we both got "Wikipedia Babel" (one of many choices) then that's a coincid...) [18:07:46] Done! I misunderstood one direction was sufficient (re @u99of9: Thanks. I just nudged the caches on the two tests I suggested, and they pass the tests. I'm not aware of any problems yet, but w...)