[00:11] OK, so [00:11] First, congratulations, that's a huge deal. [00:19] Second, I think it's best to just make a spreadsheet of what the patch does [00:19] And then just consider why each choice was made [00:19] Maybe some make no sense [00:19] I note how he has two patches there, one for 2.4 and 2.4.4 and he actually made 2.4.4 significantly bigger [00:20] I pinged the guy a few times trying to be able to weigh in on thinking [00:46] I think something's wrong with the .patch file [00:47] https://i.imgur.com/H5RTodU.png [00:47] That shouldn't be the only change in that file, according to a manual reading of the .patch file [00:54] Well, I wouldn't expect the patch file to work unmodified! [00:54] I mean, it's been 5 years between versions here [00:54] 2.4 to 3.2, a huge leap [01:00] AFK [01:09] Sgeo_: if the patch adds lines that turn out to be present in the new version, or removes lines that turn out to already be removed, then those hunks will dissapear when you go to apply the patch [01:09] disappear [01:16] I mean... it's working, right? Just blowing up on devices and interaction [01:16] That's huge [02:21] Err, did I misstate something? [02:21] It's not working yet [02:22] That error message, in vice [02:22] That means it's working [02:23] It just errors out because it can't interact [02:23] Vice boots up and then does checks and fails those checks and gives the error [02:23] Take the W Sgeo_ [02:23] If you insist [02:24] This patch modifies a file that does not exist. It also modifies a file that does exist in the same exact way and I believe is used to make the currently non-existent file, so I'll change that [02:35] The patch switches the license text used, from a 40-column one to a larger column onbe [02:37] There's a file I think I attempted to patch, wasnt successful. It looks like the lines in question have already been commented out, in a different way from the VICE.js patching [02:38] "remove remains of old hack, supposedly makes free scaling work again on some setups" in the real VICE code, in 2013, commented out those lines [02:51] Rectangular select is so nice [02:53] Not sure what this syntax in the maincpu.c part of the patch is, I might try my prior approach there first and see what happens [03:14] /bin/bash: ./gentranslate: Permission denied [03:14] Buh? [03:15] hmm, in this version, it's LLVM IR bytecode like you reported. In the directory I was using before, it's an executable [03:19] configure; make x64 made a new gentranslate. The make died afterwards, but an executable gentranslate was made [03:24] wave.cc:34:10: fatal error: 'wave6581__ST.h' file not found [03:24] #include "wave6581__ST.h" [03:25] Ok so that's dynamically created somehow or other I believe. Not sure how [03:26] Going straight for "make x64" was a mistake [03:38] error: Linking globals named 'file_system_get_vdrive': symbol multiply defined! [03:39] I was sort of expecting multiple console_closes, since I didn't add in my change to fix that yet [03:39] Wasn't expecting file_system_get_vdrive thing [03:44] emmake make clean; emmake make x64 [03:47] So utterly bizarre. make leaving behind essential things that make clean does not clean up [03:52] I feel further away then before [03:52] https://i.imgur.com/l0DAw6Z.png [03:53] Although at least I learned about some nonsense hoops I didn't know about and inadvertently skipped before [03:53] Going to call it a night [04:03] Great work, thank you [04:17] You're welcome [04:38] Tomorrow going to try switching to SDL1. Already seeing why one change in the patch was made for that [14:49] *** Sgeo has joined #jsmess [14:50] *** Sgeo_ has quit IRC (Read error: Operation timed out) [20:46] *** Sgeo has quit IRC (Read error: Operation timed out) [20:55] *** Sgeo has joined #jsmess [21:42] *** Sgeo has quit IRC (Read error: Operation timed out) [22:16] *** Sgeo has joined #jsmess [22:20] *** Sgeo_ has joined #jsmess [22:23] *** Sgeo has quit IRC (Read error: Operation timed out) [22:48] Yay I have Internet back [22:49] sdlsound.c is giving me errors about undeclared functions. There are changes in the vice.js patch that affect sdlsound. I applied those changes on the SDL2 side of the config but not the SDL1 side. Obvious thing to do is try applying them [22:56] Yep [23:03] That didn't work. I could just go ahead and include this header the compiler tells me to include. [23:10] Well, a whole bunch of files wanted that same header, so I took the easy way out and threw it in vice.h. Would love to know why it wasn't being included properly. (Maybe I misapplied the patch to the config?). Compiling now though [23:31] _SDL_UnlockSurface@http://localhost:8000/x64.js:9106:7 is failing an assertion [23:34] assert(!SDL.GL); // in GL mode we do not keep around 2D canvases and contexts [23:34] bai, any ideas why Emscripten's SDL1 would assert !SDL.GL and why SDL.GL would actually be in use?