Newest at the top
| 2025-10-31 21:42:47 +0100 | dtman34 | (~dtman34@c-73-242-68-179.hsd1.mn.comcast.net) dtman34 | 
| 2025-10-31 21:40:49 +0100 | Googulator73 | (~Googulato@2a01-036d-0106-03fa-d858-cbac-c845-1acf.pool6.digikabel.hu) | 
| 2025-10-31 21:40:42 +0100 | Googulator73 | (~Googulato@2a01-036d-0106-03fa-d858-cbac-c845-1acf.pool6.digikabel.hu) (Quit: Client closed) | 
| 2025-10-31 21:38:35 +0100 | bonken | (~bonken@77.240.72.67) | 
| 2025-10-31 21:36:52 +0100 | remexre | (~remexre@user/remexre) remexre | 
| 2025-10-31 21:36:13 +0100 | inline | (~inline@2a02:8071:57a1:dc0:da65:d2f0:d460:56a8) Inline | 
| 2025-10-31 21:35:43 +0100 | Googulator31 | (~Googulato@2a01-036d-0106-03fa-d858-cbac-c845-1acf.pool6.digikabel.hu) (Quit: Client closed) | 
| 2025-10-31 21:35:38 +0100 | Googulator73 | (~Googulato@2a01-036d-0106-03fa-d858-cbac-c845-1acf.pool6.digikabel.hu) | 
| 2025-10-31 21:34:27 +0100 | inline | (~inline@2a02:8071:57a1:dc0:da65:d2f0:d460:56a8) (Ping timeout: 250 seconds) | 
| 2025-10-31 21:33:01 +0100 | Eoco | (~ian@128.101.131.218) Eoco | 
| 2025-10-31 21:27:08 +0100 | haltingsolver | (~cmo@2604:3d09:207f:8000:d250:ea0c:366a:6e73) (Ping timeout: 256 seconds) | 
| 2025-10-31 21:17:50 +0100 | haltingsolver | (~cmo@2604:3d09:207f:8000:d250:ea0c:366a:6e73) | 
| 2025-10-31 21:07:18 +0100 | phma | (~phma@2605:59c8:6211:ff08:7181:51cb:35db:c6be) (Ping timeout: 256 seconds) | 
| 2025-10-31 21:03:46 +0100 | <tomsmeding> | and all the llvm-* tools, though I expect they share most of the code with clang | 
| 2025-10-31 21:03:27 +0100 | <tomsmeding> | oh no there's clang++ too | 
| 2025-10-31 21:03:09 +0100 | <tomsmeding> | this is just clang, though with the NVPTX backend, I think | 
| 2025-10-31 21:02:56 +0100 | <merijn> | tomsmeding: Just clang or clang++ too? | 
| 2025-10-31 21:02:56 +0100 | <tomsmeding> | 2. apparently you have many more bells 'n whistles enabled than I have | 
| 2025-10-31 21:02:41 +0100 | <tomsmeding> | 1. rip your disk | 
| 2025-10-31 21:02:30 +0100 | <merijn> | tomsmeding: I have a full install in my homedir (multiple actually) and the smallest is 5GB | 
| 2025-10-31 21:02:22 +0100 | synchromesh | (~john@2406:5a00:2412:2c00:10d9:ceb6:778c:2519) synchromesh | 
| 2025-10-31 21:02:11 +0100 | jmcantrell_ | (~weechat@user/jmcantrell) (Ping timeout: 256 seconds) | 
| 2025-10-31 21:00:59 +0100 | synchromesh | (~john@2406:5a00:2412:2c00:10d9:ceb6:778c:2519) (Read error: Connection reset by peer) | 
| 2025-10-31 21:00:59 +0100 | <tomsmeding> | though you're right that 584 is larger than 124 | 
| 2025-10-31 21:00:16 +0100 | <tomsmeding> | (yes, the build directory is 11 GB) | 
| 2025-10-31 21:00:03 +0100 | <tomsmeding> | s/MB/GB/ | 
| 2025-10-31 20:59:56 +0100 | <tomsmeding> | yes, clang 15 is old, but it hasn't become 6.5 MB in a few years | 
| 2025-10-31 20:59:44 +0100 | <tomsmeding> | merijn: I happen to have a full install of clang 15 here in a prefix directory, and it's 584 MB | 
| 2025-10-31 20:57:42 +0100 | <merijn> | Which is like a solid 6.5 GB and that's not even with all bells and whistles | 
| 2025-10-31 20:57:39 +0100 | emmanuelux | (~emmanuelu@user/emmanuelux) (Remote host closed the connection) | 
| 2025-10-31 20:57:23 +0100 | <merijn> | sm: I mean, that's still nothing compared to clang | 
| 2025-10-31 20:56:01 +0100 | wootehfoot | (~wootehfoo@user/wootehfoot) (Ping timeout: 255 seconds) | 
| 2025-10-31 20:54:59 +0100 | ss4 | (~wootehfoo@user/wootehfoot) wootehfoot | 
| 2025-10-31 20:52:52 +0100 | <haskellbridge> | <sm> ouch | 
| 2025-10-31 20:52:30 +0100 | jmcantrell_ | (~weechat@user/jmcantrell) jmcantrell | 
| 2025-10-31 20:51:51 +0100 | <tomsmeding> | my GHC 9.12 is actually 3.6 GB | 
| 2025-10-31 20:51:14 +0100 | <tomsmeding> | maybe GC + some of the RTS + native codegen starts being sizeable? | 
| 2025-10-31 20:50:47 +0100 | <tomsmeding> | ghc-in-browser only needs to do bytecode interpretation; not sure if the process is smart enough to cut the native code generator, but it could in principle | 
| 2025-10-31 20:50:07 +0100 | jmcantrell_ | (~weechat@user/jmcantrell) (Ping timeout: 240 seconds) | 
| 2025-10-31 20:49:59 +0100 | <boonochrom> | Some years ago someone in the local Haskell group eventually suggested "use a standard exe compressor" >:) | 
| 2025-10-31 20:49:00 +0100 | <tomsmeding> | still, cutting >50% still seems much | 
| 2025-10-31 20:48:54 +0100 | <boonochrom> | OK OK GHC-in-browser probably can choose just one flavour and stick to it. | 
| 2025-10-31 20:48:32 +0100 | <tomsmeding> | merijn: okay very good point, stripping that .a yields 124 MB | 
| 2025-10-31 20:47:56 +0100 | <merijn> | tomsmeding: Also, don't forget debug symbols | 
| 2025-10-31 20:47:37 +0100 | <boonochrom> | I fear that RTS also has to have those 4 flavours. | 
| 2025-10-31 20:47:23 +0100 | <merijn> | Well, presumably literally all of the GC and multi-threading? :p | 
| 2025-10-31 20:46:26 +0100 | <tomsmeding> | which 82% of GHC does the wasm playground get to cut? | 
| 2025-10-31 20:46:16 +0100 | <lambdabot> | 0.18382352941176472 | 
| 2025-10-31 20:46:14 +0100 | <tomsmeding> | > 50 / 272 | 
| 2025-10-31 20:45:57 +0100 | <merijn> | tomsmeding: Oh, I forgot mentioning you ship "GHC-as-library" in all flavours too :p |