2020-09-23T00:34:59 #kisslinux <admicos> finally, pgo looks like it's sorta gonna maybe hopefully work
2020-09-23T00:35:12 #kisslinux <admicos> apparently export MOZ_PGO=1 wasn't all and my previous attempt just did lto
2020-09-23T00:35:35 #kisslinux <admicos> also switched over to clang for cross-language (rust) optimizations
2020-09-23T00:36:15 #kisslinux <admicos> but i had to go back to mach instead of just configure+make so this is probably going to be a separate package and not a regular patch
2020-09-23T00:52:02 #kisslinux <sh4rm4^bnc> did you actually benchmark pgo vs lto vs normal build
2020-09-23T00:57:12 #kisslinux <dilynm> Why benchmark it when you can *feel* the performance
2020-09-23T00:57:16 #kisslinux <dilynm> 🔥🔥🔥
2020-09-23T00:58:46 #kisslinux <admicos> why benchmark it when you don't really care about performance but instead just want to do it because you can
2020-09-23T00:58:48 #kisslinux <admicos> :p
2020-09-23T00:59:47 #kisslinux <admicos> iirc mozilla themselves measured 1.2 - 1.4x performance improvements on _some_ benchmarks
2020-09-23T01:00:31 #kisslinux <sh4rm4^bnc> ofc, but since you invested a major effort into this wouldn't it be great to get and share some measurements
2020-09-23T01:01:19 #kisslinux <sh4rm4^bnc> imo if they really got 1.4x speedups they should look into what areas got the boost and improve that code
2020-09-23T01:01:35 #kisslinux <sh4rm4^bnc> since it's apparently where it's important
2020-09-23T01:01:57 #kisslinux <admicos> just checked the specific page i got the numbers from and it's 8 years old
2020-09-23T01:01:58 #kisslinux <admicos> so yeah
2020-09-23T02:29:16 #kisslinux <admicos> https://ebc.li/6ndw
2020-09-23T02:29:18 #kisslinux <admicos> aaaaaaaaah
2020-09-23T02:29:57 #kisslinux <admicos> "Firefox exited with code -11 during profile initialization" WHAT DOES THAT EVEN MEAN
2020-09-23T02:33:52 #kisslinux <dilynm> 'no settings file exists, new profile?'
2020-09-23T02:33:55 #kisslinux <dilynm> This seems obvious, firefox
2020-09-23T02:34:25 #kisslinux <admicos> that's just a warning iirc
2020-09-23T02:38:13 #kisslinux <admicos> i give up.
2020-09-23T02:38:16 #kisslinux <admicos> rest of y'all have fun: https://git.ebc.li/kiss/optim/src/branch/firefox-proper-pgo/repo/firefox
2020-09-23T02:38:35 #kisslinux <admicos> i'll probably clean up the clang+lto changes and push them, and pgo will stay for anothey day
2020-09-23T03:13:54 #kisslinux <konimex> lshw
2020-09-23T03:13:56 #kisslinux <konimex> whoops
2020-09-23T03:14:11 #kisslinux <konimex> wrong window
2020-09-23T04:04:44 #kisslinux <dilynm> Just realized I've been building rust with... -j1
2020-09-23T04:04:49 #kisslinux <dilynm> Smh
2020-09-23T04:05:02 #kisslinux <dilynm> No wonder it's so painful
2020-09-23T05:32:18 #kisslinux <dylanaraps> Can anyone confirm that Firefox 81.0 built fine?
2020-09-23T05:36:19 #kisslinux <konimex> with gtk2 patch or without?
2020-09-23T05:36:29 #kisslinux <dylanaraps> With
2020-09-23T05:36:54 #kisslinux <dylanaraps> Works fine here, just wanted to confirm with others
2020-09-23T05:37:01 #kisslinux <dylanaraps> (Under X11 and Wayland)
2020-09-23T05:37:05 #kisslinux <konimex> I'll compile it later
2020-09-23T05:45:30 #kisslinux <dylanaraps> I wonder if we can remedy this: ERROR: Could not find clang to generate run bindings for C/C++. Please install the necessary packages, run `mach bootstrap`, or use --with-clang-path to give the location of clang.
2020-09-23T05:45:46 #kisslinux <dylanaraps> Clang is required to build Firefox even when everything is set to GCC...
2020-09-23T05:46:12 #kisslinux <dylanaraps> So it forces users to have both compilers installed (unless pure clang system)
2020-09-23T05:46:22 #kisslinux <dylanaraps> (or unless it also forces gcc somewhere...)
2020-09-23T05:47:07 #kisslinux <dylanaraps> From: https://dxr.mozilla.org/mozilla-release/source/build/moz.configure/bindgen.configure#203
2020-09-23T05:52:50 #kisslinux <dylanaraps> Doesn't look to be possible
2020-09-23T05:54:19 #kisslinux <dylanaraps> Firefox uses https://github.com/rust-lang/rust-bindgen which depends on libclang/clang
2020-09-23T06:48:40 #kisslinux <muevoid> kiss-outdated is very spotty for me sometimes it shows certain packages othertimes it doesn't
2020-09-23T06:49:59 #kisslinux <muevoid> Oh yeah dilynm I will fix seatd quickly
2020-09-23T06:59:14 #kisslinux <muevoid> Opened up a pr dilynm
2020-09-23T08:04:21 #kisslinux <muevoid> dylanaraps do you know why sometimes kiss ou doesn't show all the packages sometimes?
2020-09-23T08:05:18 #kisslinux <dylanaraps> muevoid: kiss-outdated lists all packages in your KISS_PATH
2020-09-23T08:06:11 #kisslinux <dylanaraps> How it works is more or less a hack
2020-09-23T08:06:11 #kisslinux <muevoid> Yes but for example socat is currently outdated in my repo, however it only shows up sometimes when running kiss-outdated othertimes it doesn't show up. Appearingly random
2020-09-23T08:06:24 #kisslinux <dylanaraps> I'll try removing the async stuff
2020-09-23T08:07:20 #kisslinux <muevoid> kk
2020-09-23T08:07:29 #kisslinux <dylanaraps> I'll rewrite the whole thing
2020-09-23T08:07:34 #kisslinux <dylanaraps> Give me 15 minutes or so
2020-09-23T08:08:28 #kisslinux <muevoid> Kk now worries, thank you for doing so.
2020-09-23T08:44:35 #kisslinux <dylanaraps> muevoid: https://github.com/kisslinux/kiss/commit/cd6c95419db670a9988f9d1017a497a4da598210
2020-09-23T08:44:39 #kisslinux <dylanaraps> Let me know how this goes
2020-09-23T08:44:53 #kisslinux <dylanaraps> Will print error messages which should tell us why it's failing
2020-09-23T08:44:59 #kisslinux <dylanaraps> Will then be able to look deeper into the problem
2020-09-23T08:45:27 #kisslinux <dylanaraps> Just pop this in your $PATH somewhere https://github.com/kisslinux/kiss/blob/cd6c95419db670a9988f9d1017a497a4da598210/contrib/kiss-outdated
2020-09-23T08:45:39 #kisslinux <dylanaraps> Usage is now: kiss outdated /path/to/repo
2020-09-23T08:47:01 #kisslinux <dylanaraps> Ah, there's a bug.
2020-09-23T08:47:04 #kisslinux <dylanaraps> Hang on.
2020-09-23T08:51:23 #kisslinux <dylanaraps> Pushed more fixes.
2020-09-23T08:57:42 #kisslinux <muevoid> Ok thank you a lot!
2020-09-23T08:57:49 #kisslinux <muevoid> Testing it now
2020-09-23T08:58:23 #kisslinux <dylanaraps> Just pushed another commit
2020-09-23T08:58:44 #kisslinux <dylanaraps> https://github.com/kisslinux/kiss/blob/master/contrib/kiss-outdated
2020-09-23T08:58:58 #kisslinux <dylanaraps> What it is doing should be clearer now too
2020-09-23T08:59:13 #kisslinux <dylanaraps> I'll add some comments too
2020-09-23T08:59:23 #kisslinux <muevoid> Thay That is much better Thank you!
2020-09-23T09:03:01 #kisslinux <dylanaraps> I need to add a few more things
2020-09-23T09:03:16 #kisslinux <dylanaraps> But it should be better than it was
2020-09-23T09:04:58 #kisslinux <muevoid> It is much better! Thank you so much for your work
2020-09-23T09:05:06 #kisslinux <dylanaraps> No problem
2020-09-23T09:05:36 #kisslinux <dylanaraps> I just need to lowercase every package name (as repology does) and fixup a few known package naming issues (gtk+3 -> gtk).
2020-09-23T09:05:55 #kisslinux <muevoid> Kk
2020-09-23T09:35:11 #kisslinux <dylanaraps> https://github.com/kisslinux/kiss/blob/master/contrib/kiss-outdated
2020-09-23T09:35:35 #kisslinux <dylanaraps> muevoid: Should be good now. Will push more fixes to (known) naming conflicts later today.
2020-09-23T09:36:03 #kisslinux <dylanaraps> Let me know if the code is unclear anywhere too
2020-09-23T10:04:55 #kisslinux <eudaldgr> hola
2020-09-23T10:05:18 #kisslinux <eudaldgr> someone know what the hell means the name on i915 inside linux-firmware?
2020-09-23T10:06:51 #kisslinux <eudaldgr> i'm trying to understand these names, i think skl_* are skylake, bxt_* broxton and kbl_* kaby lake
2020-09-23T10:07:03 #kisslinux <eudaldgr> but the others i have no idea
2020-09-23T10:11:17 #kisslinux <dylanaraps> You should lookup which firmware is needed for your hardware
2020-09-23T10:11:24 #kisslinux <dylanaraps> I think you're doing this backwards
2020-09-23T10:11:47 #kisslinux <dylanaraps> Intel's website should have a page about it
2020-09-23T10:12:03 #kisslinux <dylanaraps> Maybe exists in a README-esque file in linux-firmware?
2020-09-23T10:17:59 #kisslinux <eudaldgr> No README explaining it inside linux-firmware, and inside intel website i'm trying also to find that info
2020-09-23T10:18:47 #kisslinux <eudaldgr> i'm trying to find wich of them are for ivy bridge
2020-09-23T10:19:29 #kisslinux <eudaldgr> but i have no luck
2020-09-23T10:19:51 #kisslinux <eudaldgr> maybe there is no one
2020-09-23T10:20:06 #kisslinux <dylanaraps> Try use the iGPU model name
2020-09-23T10:25:00 #kisslinux <konimex> most Intel iGPU should be i915, iirc
2020-09-23T10:26:42 #kisslinux <konimex> ah, the firmware
2020-09-23T11:02:11 #kisslinux <dylanaraps> muevoid: Finished https://github.com/kisslinux/kiss/blob/master/contrib/kiss-outdated
2020-09-23T11:12:00 #kisslinux <muevoid> Nice! I'm checking it out now. Thanks again!
2020-09-23T11:14:24 #kisslinux <dylanaraps> You can now also do: kiss ou /var/db/kiss/installed/
2020-09-23T11:14:35 #kisslinux <dylanaraps> To check your installed list of packages against repology.org
2020-09-23T11:25:55 #kisslinux <dylanaraps> Also 2>/dev/null to show only outdated packages (and no other messages)
2020-09-23T11:49:33 #kisslinux <dilynm> eudaldgr: iirc the i915bfirmwares are for post-haswell CPUs so there isn't one for ivy bridge
2020-09-23T12:41:10 #kisslinux <eudaldgr> @dilynm: thanks
2020-09-23T16:58:28 #kisslinux <himmalerin_kirc> Anyone ever try getting Avahi working? I seem to need it for printing
2020-09-23T18:44:32 #kisslinux <E5ten> I just messed around with elfutils a bit, trying to make the codebase less fucked by removing nested functions, and holy fuck there are a pretty significant amount of nested functions
2020-09-23T18:45:15 #kisslinux <E5ten> I've come to the conclusion that the elfutils developers hate humanity, and the elfutils source is an expression of their hatred
2020-09-23T18:52:51 #kisslinux <E5ten> dylanaraps: for checking if KISS_ROOT ends with /, instead of just stripping 3 of them and hoping that's enough, can't you use that thing from pure-sh-bible (or bash I forget) for stripping trailing whitespace, but use it for / instead?
2020-09-23T18:53:25 #kisslinux <E5ten> ${KISS_ROOT%${KISS_ROOT##*[!/]}}
2020-09-23T18:54:58 #kisslinux <dylanaraps> E5ten: Yes
2020-09-23T18:55:00 #kisslinux <dylanaraps> Good idea
2020-09-23T18:55:49 #kisslinux <claudia02> dylanaraps: I have successfully build latest 'firefox' with X
2020-09-23T18:56:38 #kisslinux <dylanaraps> Thanks
2020-09-23T19:09:02 #kisslinux <dilynm> E5ten: dropping nested functions in elfutils seems like a never-ending battle
2020-09-23T19:09:21 #kisslinux <dilynm> Somebody must've finished the work for at least libelf tho because clang can build it
2020-09-23T19:09:36 #kisslinux <E5ten> I gave up for now, I replaced all the ones in the libs with macros, but there are so many in the commands that I just can't keep going at least for now
2020-09-23T19:12:41 #kisslinux <dilynm> Are they even actually useful?
2020-09-23T19:12:56 #kisslinux <dilynm> Like, it seems like the authors would just rather not be writing C if they want to use them
2020-09-23T19:13:39 #kisslinux <E5ten> they're there because they use variables that are local to the function but don't take them as params
2020-09-23T19:13:51 #kisslinux <E5ten> so if they weren't nested you'd have to make all those variables global
2020-09-23T19:14:56 #kisslinux <E5ten> I turned the nested functions into macros, and for the ones that had return types (weren't void) I made a variable within the function, set it to the macro's "return", and broke from the do while loop, and then whenever the nested function's return value was used, I replaced that with a standalone call to the macro, and checking the variable it sets
2020-09-23T19:15:04 #kisslinux <E5ten> it's disgusting, but not as disgusting as using fucking nested functions
2020-09-23T19:17:19 #kisslinux <dilynm> Lmao
2020-09-23T19:17:41 #kisslinux <dilynm> I bet it doesn't look pretty enough for upstream to take it
2020-09-23T19:18:42 #kisslinux <E5ten> it's not even worth trying because there are so many others that would need dealing with, also their codebase is fucked and has both tabs and spaces so I wouldn't even know how to indent all my shit correctly, where they use spaces it's 2-space indentation so I just made all the code I changed into that
2020-09-23T19:19:20 #kisslinux <E5ten> In at least one place they have 2 nested functions in the same function with the same name
2020-09-23T19:20:00 #kisslinux <E5ten> They're just both nested under switches so they don't conflict, but for the macro version I had to #undef it after the uses of the first one
2020-09-23T19:22:35 #kisslinux <dilynm> This sounds like cancer
2020-09-23T19:23:10 #kisslinux <dilynm> Would it just be easier to extend elftoolchain xD
2020-09-23T19:23:20 #kisslinux <dylanaraps> And elftoolchain upstream is really slow :(
2020-09-23T19:24:30 #kisslinux <E5ten> Well to be fair the only issue I know of with elftoolchain (for libelf in the kernel I mean) is that one thing I made a patch for, although I'm not 100% sure the patch is correct (seems to be though)
2020-09-23T19:25:01 #kisslinux <dylanaraps> Yeah
2020-09-23T19:25:05 #kisslinux <E5ten> I do wish that bug I made would get a response though... I sent a ping like yesterday of the day before
2020-09-23T19:25:11 #kisslinux <dylanaraps> I saw
2020-09-23T19:25:13 #kisslinux <dylanaraps> :P
2020-09-23T19:25:13 #kisslinux <E5ten> Or*
2020-09-23T19:25:48 #kisslinux <dylanaraps> Are there any other elf libraries other than the old libelf?
2020-09-23T19:26:26 #kisslinux <E5ten> Not that I know of
2020-09-23T19:26:54 #kisslinux <E5ten> But is elftoolchain with one patch really a problem?
2020-09-23T19:29:00 #kisslinux <dilynm> It was for me (:
2020-09-23T19:29:05 #kisslinux <dylanaraps> Not really. I just hope the problem doesn't grow over time
2020-09-23T19:29:26 #kisslinux <dylanaraps> It's likely the kernel will support whatever elfutils does
2020-09-23T19:29:35 #kisslinux <dilynm> If they happen to upstream some patches then get data will be problematic for elftoolchain
2020-09-23T19:29:59 #kisslinux <mcf> E5ten: i reviewed your patch and it looks good to me. i just made a small additional change to remove the check for lid == NULL in the subsequent if statement (since it is impossible)
2020-09-23T19:32:44 #kisslinux <dylanaraps> The kernel seems to already depend on elfutils for some stuff https://github.com/torvalds/linux/search?p=2&q=elfutils
2020-09-23T19:36:54 #kisslinux <E5ten> mcf: thanks, did you send your updated one to their list or add it to the bug?
2020-09-23T19:37:55 #kisslinux <mcf> i did not (it's just what i'm using locally)
2020-09-23T19:38:02 #kisslinux <E5ten> Ah I see
2020-09-23T19:38:21 #kisslinux <E5ten> dylanaraps: almost all of that is in tools/perf, which doesn't matter
2020-09-23T19:38:58 #kisslinux <E5ten> The rest are either comments, or mention of elfutils in regard to finding a libelf cuz that's the main implementation
2020-09-23T19:40:42 #kisslinux <E5ten> mcf: maybe you could mention in the bug that you're using that patch locally so they know that there's more than just me interested in getting that working, and so they know it's got some "testing" (like because you're using it and it seems to work)
2020-09-23T19:44:10 #kisslinux <mcf> sure
2020-09-23T19:45:17 #kisslinux <E5ten> dylanaraps: https://github.com/search?q=elfutils+repo%3Atorvalds%2Flinux+-path%3Atools%2Fperf+-path%3Atools%2Fbuild&type=Code this search shows elfutils with tools/perf and tools/build removed (both are for perf, so not relevant to just building the kernel)
2020-09-23T19:53:03 #kisslinux <E5ten> mcf: thanks
2020-09-23T22:10:56 #kisslinux <E5ten> mcf: lol, looks like a second person commenting worked, new response already saying he'll be doing it soon
2020-09-23T22:24:09 #kisslinux <himmalerin_kirc> Just noticed foot 1.5 has a little scrollbar indicator, thats neat
2020-09-23T23:03:05 #kisslinux <dilynm> https://dev.getsol.us/T4824 I would love this from a KISS-kde perspective because I want to drop mozjs
2020-09-23T23:03:29 #kisslinux <dilynm> But the mailing list goes ~nowhere~, and it looks like ikey's work never got upstreamed. Sad.