[01:09] Hi, back from dropping off friends. [01:09] OK, so. [01:09] I used the Pleasuredome CHD Torrent. It ABSOLUTELY splits things this way [01:09] All I split was the directories with "What are the ones with one image" vs. "what are the ones with multiple imnages" [01:11] So, we're now down to the J's with the PSX Uploads. [01:12] I'm re-running the Metadata injector, which is the thing that puts in the data from the XML document of the Software List. [01:14] It's at 820 items out of 2100 of the first easy batch. [01:15] From 2010 Blog: [01:15] Here are the default disk cache sizes for the major browsers: [01:15] Internet Explorer: 8-50 MB. [01:15] Firefox: 50 MB. [01:15] Safari: everything I found said there isn't a max size setting (???) [01:15] Chrome: < 80 MB (varies depending on available disk space) [01:15] Opera: 20 MB. [01:16] Ha, there's a tester for it. [01:17] Yeah, looks like it's http://stevesouders.com/tests/cachesize/ [01:24] I have no useful response to the Cache situation other than we're forcing an example of the usage of it. [01:44] you can use indexeddb to store more, but it looks like only up to about 2gb per origin? [01:44] and the browser will ask if it's more than like 50mb [01:45] it doesn't ask, as far as I know [01:46] yeah I had to go in there and clean out some stuff because my ssd was getting full [01:46] this could be an interesting application for service workers [01:48] Small update - I killed the "oops" item just because it's going to backfill into the regular thing. [01:48] I wrote a simple script to cross-match our items into Mobygames links and I REALLY WISH I could hand it to someone but I don't think that's realistic. [01:52] *** godane has quit IRC (Quit: Leaving.) [01:52] *** godane has joined #jsmess [02:02] Not surprisingly, Mobygames is hitting issues with various Japanese games. [02:18] https://archive.org/details/psx_alicecyb - This is what a "finished" entry will look like. [04:35] *** jvilk has joined #jsmess [04:35] What's going on guys? [04:36] SketchCow pinged me a few hours ago [04:44] jvilk: did any work ever get done towards lazy access of large files in browserfs using http range requests [04:46] we're trying to load playstation CD-ROMs in jsmess and it would be nice to not have to download 600mb all up front [04:55] DFJustin: oh nice! nothing has been done yet. there's something cooler we could do if sharedarraybuffers come back and we are able to put emulators into a worker. [04:56] It's something I'd like to return to once I finish my PhD (september!) [05:12] ok [05:15] there's also the issue of memory usage, since Chrome limits tabs to 1.4GB. loading up 600MB fills up quite a bit of that. [05:15] yeah [05:15] we are pushing it [05:16] any chance of workerizing the emulators? or is that something that's still lots of work w/ emscripten? [05:16] (workerizing = running in a Worker) [05:52] jvilk: had some initial success with that when w rried a couple years ago, main problems were proxying audio and canvas data to main thread to actually draw. browsers now almost support OffscreenCanvas, but audio still needs to be proxied [05:53] also input needs to be proxied from the main thread [05:53] it worked but there was a little bit of extra latency [06:57] Down to the R's. [06:57] 1,466 done [06:57] They're all just working. [06:58] It's of course weird and there's some flakiness with emulation [06:58] But a lot are just working. [07:09] *** bwn has quit IRC (Read error: Operation timed out) [07:18] *** bwn has joined #jsmess [07:29] https://archive.org/details/psxgames?&sort=publicdate [07:29] That's starting to have a nice look [07:29] nice [07:44] i always thought psx games was going be hard to emulate with java script cause of file size [07:46] i was thinking it was most likely to limit to less then 100mb games just say 'yes we can emulate psx' and nothing more [08:23] *** datajerk has quit IRC (ZNC 1.6.6 - http://znc.in) [08:48] *** datajerk has joined #jsmess [10:11] *** datajerk has quit IRC (Read error: Operation timed out) [10:26] *** datajerk has joined #jsmess [11:28] *** datajerk has quit IRC (Read error: Operation timed out) [11:39] *** datajerk has joined #jsmess [11:44] *** datajerk has quit IRC (Read error: Operation timed out) [12:16] *** datajerk has joined #jsmess [12:52] My main thing was thinking if we couldn't emulate Jaguar, how could we emulate anything beyond. [12:52] Apparently we can [12:55] So, the first set is in, that's the ones that were obviously not multi-disc. [12:55] 2,120 items! [13:21] DFJustin: I found what you mean. Even though this thing is semi-official, i.e. someone is doing this using the MAME data, they lumped together similar titles in the case of "japanese version" or "demo" or "beta" version. [14:18] "merged" sets based on the mame parent-clone relationships [14:37] *** bwn has quit IRC (Read error: Operation timed out) [14:45] *** bwn has joined #jsmess [14:49] *** godane has quit IRC (Ping timeout: 492 seconds) [15:21] *** godane has joined #jsmess [18:20] *** logchfoo0 starts logging #jsmess at Thu Mar 29 18:20:14 2018 [18:20] *** logchfoo0 has joined #jsmess [19:29] Ah [19:29] Whoooooooooooooooooooooooooooooooooooooooooooops [19:29] uh oh [19:30] *** bwn has quit IRC (Read error: Operation timed out) [19:40] Well, it's not terminal. [19:40] I'm getting most fixed pretty quickly. [19:46] *** bwn has joined #jsmess [21:03] *** datajerk has quit IRC (Read error: Operation timed out) [21:05] *** datajerk has joined #jsmess [21:42] *** bwn has quit IRC (Read error: Operation timed out) [22:03] I wrote a demerger script [22:03] #!/bin/sh [22:03] GOG="$1" [22:03] if [ -d $GOG ] [22:03] then [22:03] echo "Visting $GOG" [22:03] cd "$GOG" [22:03] for each in *.chd [22:03] do [22:03] FELT=`echo $each | sed 's/.chd//g'` [22:03] grep -i "$FELT" /2/MAMELAND/TEXTHELL/* [22:03] echo "-----" [22:03] echo "Enter a new directory name if you want to move this." [22:03] read BRUCE [22:03] if [ "$BRUCE" ] [22:03] then [22:03] mkdir "../$BRUCE" [22:03] mv "$each" "../$BRUCE" [22:03] fi [22:03] done [22:04] cd .. [22:04] fi [22:04] *** bwn has joined #jsmess [22:21] I know I have been fairly inactive lately; I really want to come back in a big way once my schedule is less crazy. I would like to poke at things this summer, but that's also when I'm finishing my dissertation and moving. But this project has not been far from my mind; I would love to see big CD-ROM-based games come to life in the browser without super long wait time and the ability to easily download and up [22:22] load save files into the file system. Anyway, that's where I'm coming from, in case anyone was wondering. [22:22] All talk, no action this month. Of course, if there are any bugs/issues with BrowserFS, I'm responsive to those. :3 [22:31] That's OK [22:31] The whole system is testing the limits of what browsers can do and I'm glad we're still on the forefront of that [22:31] Also fuck Larry Ellison [22:31] Also, I'm in London next week, I assume nobody here is from there. [22:39] So, my splitter thing is working nicely. I'll be through it in the next few minutes. [23:11] Ah, OK, I'm running into a strange thing... the Lua script stopped working and wouldn't start working, and I've found it's because when a new version is downloaded, it doesn't automatically truncate the file when overwriting it? [23:11] So I have the new beginning of the file with a lot of trailing garbage, which is causing the file not to parse. [23:12] I can have it manually delete/truncate the file before loading the new one, but I feel like I'm missing some flag value that's essential in cfgr.mountFile. [23:16] Ah, I see. We have flag_w defined with isTruncating: false. Is that intentional? [23:18] I feel like truncating the existing local data would be the expected action for fetching a file, but I'm not the one who wrote it, so it's good to check. [23:24] When db48x is around he might know [23:24] It must just be "default" and never hit an issue before. [23:25] Right, that's what I want to check. [23:25] But, as it turns out, changing it doesn't seem to fix the problem, hm. [23:25] So, moot point. [23:55] Hey jvilk: as long as you're here (if you're still here), is there an easy way to get to the ActionType global (or any other globals of BrowserFS) from the outside without calling BrowserFS.install()?