00:00:08 oerjan: hmm 00:00:19 oh hi 00:01:15 I wonder if this is related to opitmal sorting networks or optimal interactive sorting protocols that minimize the number of comparisons 00:01:50 or that weird screws and nuts comparison problem 00:01:53 perhaps, although the difference is that the underlying tests here don't really come from a _total_ order 00:02:19 you still can structure them in a binary tree, though, which is why they must obey the OEIS function limit 00:02:53 -!- impomatic has quit (Quit: impomatic). 00:03:13 -!- impomatic has joined. 00:03:29 (the function can be described as: minimum sum of depths of leaves in a binary tree with a given number of leaves) 00:03:43 ftr I don't need an optimal method; close to optimal for large numbers and optimal for small numbers is already fine for me 00:04:03 true, as i said i just got obsessed about the question 00:05:22 and also finding the optimal method might get hairy in more complicated cases: for 16 files i'm having to look at the corner case with _5_ files out of 16 00:05:54 which looks like it _might_ be done better than by bisection but i haven't wrapped it up 00:06:45 and I logread this channel anyway 00:07:25 -!- impomatic has quit (Client Quit). 00:07:28 a general principle seems to be that the problem gets harder when the number of possible pairs is close to a power of 2 (because that minimizes the slack you have in balancing) 00:07:44 -!- impomatic has joined. 00:09:01 if you care about the minimum that is, not eg. the average 00:09:32 i'm actually caring about the average over all possible pairs 00:10:09 which is OEIS(n)/n for a sufficiently balanced tree 00:10:23 er 00:10:35 that's when n is the number of possible pairs. 00:11:04 also for a _perfectly_ balanced tree that also forces the minimum 00:11:28 s/perfectly/sufficiently/ 00:11:55 well that _is_ perfectly for that given number 00:12:39 because the only way to have that is for all the leaves to be at exactly 2 levels - otherwise you can improve by rearranging. 00:12:58 (well or 1 level for an exact power of 2) 00:16:52 anyway, i'm monologuing again (MWAHAHAHA) 00:36:26 -!- Lord_of_Life_ has joined. 00:36:27 -!- Lord_of_Life has quit (Ping timeout: 240 seconds). 00:37:40 -!- Lord_of_Life_ has changed nick to Lord_of_Life. 00:40:39 -!- impomatic has quit (Quit: impomatic). 00:50:17 -!- chiselfuse has quit (Quit: leaving). 00:50:35 -!- chiselfuse has joined. 00:52:56 -!- PinealGlandOptic has quit (Quit: leaving). 00:54:00 -!- chiselfuse has quit (Client Quit). 00:54:17 -!- chiselfuse has joined. 01:01:21 -!- impomatic has joined. 01:11:47 -!- chiselfuse has quit (Quit: leaving). 01:12:00 -!- chiselfuse has joined. 01:21:57 -!- chiselfuse has quit (Quit: leaving). 01:22:15 -!- chiselfuse has joined. 01:26:01 -!- Hooloovoo has joined. 01:26:15 j #dragonbox-pyra 01:26:29 woops 01:29:41 -!- impomatic has quit (Ping timeout: 250 seconds). 01:41:07 Which web browsers can you write dynamic extensions in C? 01:45:52 none anymore 01:46:26 they want to remove plugins, and remove even the feature of supporting plugins 01:47:38 whether it's activex, ppapi or npapi 01:48:14 does the south korean banking system still run on activex or did they fix that finally 01:48:55 LiveConnect and XPConnect might still works 01:49:05 None of those are what I had meant anyways though; I mean extensions and not plugins of web pages 01:49:58 well ppapi/npapi are sort of both 01:50:10 extension can only be in webextension nowadays 01:51:15 I think WebExtensions doesn't work well enough 01:51:47 i do too, they have LESS features than the previous solution 01:51:52 but we don't have anything else 01:52:39 I still use the old version of Firefox with XUL, since I dislike many of the features of the new one 01:53:14 then you can probably still use the old firefox extensions 01:53:18 and npapi 01:53:23 Although I have ideas how a better web browser should be written, one of the ideas is that extensions should be C 01:53:46 all plugins already are C 01:54:44 Also, HTTP, HTTPS, HTML, and many other things should also be implemented as extensions, although the "data" URI scheme will be a part of the core system 01:56:58 yeah, and we should restore support of ftp, ftps, sftp, gopher, gemini, ipfs, tor, etc... in browsers 01:57:07 alongside with rtf, sgml, tex, markdown and others 01:57:49 IDE (and toolchains) already have something similar with https://langserver.org/ 01:58:47 it WAS made by microsoft, but it's still a very nice idea, completely separating the tool, from the language, from the compiler 01:59:31 most IDE and tools already supports it 02:00:45 Yes, FTP, Gopher, Gemini, IPFS, etc can also be implemented as extensions and should be included with the browser. (Implementing nearly all of the URI schemes (except data: and some uses of about:) in extensions makes it more uniform and allow better customization, and ensures that the extension interface works good enough) 02:01:53