2020-07-18T02:00:14 #kisslinux <E5ten> illiliti: now that only a few functions (the unexported ones in udev_list) are not supposed to be exported, and aren't marked static, couldn't you switch from having -fvisibility=hidden and marking every API function with the default visibility attribute, to not using -fvisbility=hidden, not marking API functions at all, and using the same attribute but with "hidden" visibility on those few udev_list functions?
2020-07-18T02:20:28 #kisslinux <illiliti> E5ten: i decided to not using symbol visibility at all
2020-07-18T02:21:18 #kisslinux <illiliti> leaking some functions in udev_list is not big problem i guess
2020-07-18T02:23:54 #kisslinux <E5ten> also in the Makefile, you do -D_POSIX_VERSION=200809L, but the correct define for that would be _POSIX_C_SOURCE=200809L, but either way since you use _XOPEN_SOURCE=700, which implies _POSIX_C_SOURCE=200809L, you can drop that define altogether
2020-07-18T02:29:20 #kisslinux <illiliti> "_XOPEN_SOURCE=700, which implies _POSIX_C_SOURCE=200809L" << that behaviour defined by posix ?
2020-07-18T02:46:00 #kisslinux <E5ten> yeah
2020-07-18T02:48:06 #kisslinux <E5ten> basically from what I can tell each _POSIX_C_SOURCE version will be enabled by a corresponding _XOPEN_SOURCE version that gets incremented by 100, so _POSIX_C_SOURCE=200112L (the one before 200809L) is enabled by _XOPEN_SOURCE=600, _POSIX_C_SOURCE=200809L is enabled by _XOPEN_SOURCE=700, and I'm sure whatever the next _POSIX_C_SOURCE is will be enabled by _XOPEN_SOURCE=800
2020-07-18T02:56:55 #kisslinux <illiliti> since all functionality enabled by _POSIX_C_SOURCE set equal to 200809L is enabled by _XOPEN_SOURCE set equal to 700, there should be no need to define _POSIX_C_SOURCE if _XOPEN_SOURCE is so defined
2020-07-18T02:56:59 #kisslinux <illiliti> https://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html
2020-07-18T02:57:16 #kisslinux <illiliti> found :)
2020-07-18T03:07:01 #kisslinux <illiliti> performance is really slow compared to eudev libudev
2020-07-18T03:07:52 #kisslinux <illiliti> eudev caches data to /run/udev/data
2020-07-18T03:10:43 #kisslinux <illiliti> i should speed up performance, but idk where to start. i don't want to implement caching
2020-07-18T03:13:05 #kisslinux <illiliti> eudev: LD_LIBRARY_PATH= libinput list-devices  0.07s user 0.00s system 7% cpu 0.950 total
2020-07-18T03:13:09 #kisslinux <illiliti> libudev-zero: LD_LIBRARY_PATH=~/projects/libudev-zero libinput list-devices  0.13s user 0.34s system 12% cpu 3.850 total
2020-07-18T04:24:57 #kisslinux <E5ten> Illiliti: you should add XCPPFLAGS, have it include CPPFLAGS as well as the _XOPEN_SOURCE definition instead of XCFLAGS, and then add it to the .c.o
2020-07-18T07:52:08 #kisslinux <tuxcanfly> hey folks, love the great work... keep it up 👍
2020-07-18T08:10:11 #kisslinux <dylanaraps> o/
2020-07-18T08:10:25 #kisslinux <dylanaraps> Great talk about LibreSSL's first 30 days: https://www.youtube.com/watch?v=GnBbhXBDmwU
2020-07-18T08:16:38 #kisslinux <tuxcanfly> *ssl's are doomed I'm afraid... have you seen: https://github.com/mit-plv/fiat-crypto
2020-07-18T08:17:16 #kisslinux <tuxcanfly> https://github.com/bcoin-org/bcrypto I use this project which uses fiat-crypto primitives
2020-07-18T08:41:54 #kisslinux <dylanaraps> Interesting
2020-07-18T09:25:18 #kisslinux <konimex> ah man rust has updated
2020-07-18T09:34:19 #kisslinux <konimex> anyway dylan: rare addition, I thought you were against gcompat a while back iirc
2020-07-18T09:49:34 #kisslinux <xzcvczx> i am confusde hte alternative dev manager post talks about unlinking udevd and losing hotplug, i don't have udevd in /var/service and hotplug ssems to work
2020-07-18T09:50:17 #kisslinux <konimex> which post
2020-07-18T09:50:45 #kisslinux <xzcvczx> https://k1ss.org/wiki/dev/replacing-udev , wiki sorry
2020-07-18T09:51:24 #kisslinux <xzcvczx> under "Purging udev" (all in caps)
2020-07-18T09:51:38 #kisslinux <xzcvczx> purging eudev rather
2020-07-18T09:59:17 #kisslinux <dylanaraps> konimex: I'm against the ripped out portions from other libcs. This is only useful for binaries as it provides no headers.
2020-07-18T10:02:03 #kisslinux <j`ey> is there a script that creates the kiss-chroot tarball?
2020-07-18T10:04:42 #kisslinux <konimex> ah I see
2020-07-18T10:04:52 #kisslinux <konimex> also rust 1.45.0 requires rustc 1.44.0 rather than 1.44.1, odd
2020-07-18T10:06:03 #kisslinux <j`ey> the .1 changes look minor, and dont seem to affect linux
2020-07-18T15:39:43 #kisslinux <E5ten> illiliti: what makes libusb not work with libudev-zero (Missing functions, stubs for functions it needs, etc.)?
2020-07-18T15:46:41 #kisslinux <konimex> finally! cargo-native-static can be done so if LibreSSL has a soname bump I don't have to build again, requires GNU make though...
2020-07-18T15:50:25 #kisslinux <illiliti> E5ten: it hangs at timerfd_create()
2020-07-18T15:50:54 #kisslinux <illiliti> try to compile libusb with udev support and you will see
2020-07-18T15:59:10 #kisslinux <illiliti> you can install usbutils to reproduce this bug
2020-07-18T16:21:36 #kisslinux <illiliti> kwin bug is much stranger than this
2020-07-18T16:22:25 #kisslinux <illiliti> reported by dilynm ix.io/2rmw
2020-07-18T16:57:53 #kisslinux <E5ten> I wonder what could cause it to hang on a non-udev function with libudev-zero but not libudev
2020-07-18T16:58:33 #kisslinux <E5ten> what did you use to find where it was hanging?
2020-07-18T17:23:21 #kisslinux <illiliti> strace
2020-07-18T17:24:46 #kisslinux <E5ten> I see, I just tried that and it doesn't help too much outside of seeing it SIGSEGV at timerfd_create cuz it doesn't actually show the functions calling those syscalls
2020-07-18T17:25:16 #kisslinux <E5ten> for the kwin thing, input.cpp doesn't seem to include libudev.h (even indirectly) so I don't understand how it works with normal libudev?
2020-07-18T17:26:02 #kisslinux <E5ten> udev_device_get_devnode is only used in input.cpp and udev.cpp, udev.cpp is the only file that includes libudev.h (no internal headers do) so there doesn't seem to be anything that leads to input.cpp including it?
2020-07-18T17:26:16 #kisslinux <E5ten> like I think just adding a libudev.h include to that would fix it
2020-07-18T17:26:49 #kisslinux <illiliti> E5ten: patch for segfault - termbin.com/homk
2020-07-18T17:28:03 #kisslinux <illiliti> i will get hang with this patch
2020-07-18T17:28:19 #kisslinux <E5ten> can you explain why (void *)1 is an invalid value for the udev_device struct?
2020-07-18T17:28:43 #kisslinux <E5ten> dilynm: can you try just adding a libudev.h include to input.cpp and see if that fixes it?
2020-07-18T17:29:12 #kisslinux <illiliti> (void *)1 is no-op
2020-07-18T17:29:14 #kisslinux <illiliti> https://github.com/illiliti/libudev-zero/blob/master/udev_monitor.c#L15
2020-07-18T17:29:30 #kisslinux <E5ten> oh
2020-07-18T17:29:45 #kisslinux <E5ten> and what's the reason for not returning NULL there?
2020-07-18T17:30:26 #kisslinux <illiliti> exit(1)
2020-07-18T17:31:18 #kisslinux <E5ten> what do you mean?
2020-07-18T17:31:23 #kisslinux <illiliti> xorg expected non null value to continue
2020-07-18T17:31:27 #kisslinux <E5ten> ah
2020-07-18T17:31:38 #kisslinux <illiliti> same for wlroots
2020-07-18T17:32:06 #kisslinux <E5ten> so maybe when that function is implemented that'll fix the libusb issues?
2020-07-18T17:32:38 #kisslinux <illiliti> yeah
2020-07-18T17:36:14 #kisslinux <illiliti> kwin uses c++, maybe c++ features provide ability to somehow indirectly include that header ? idk
2020-07-18T17:36:55 #kisslinux <illiliti> or maybe this is cmake magic
2020-07-18T17:37:45 #kisslinux <E5ten> cmake would only be adding the include directory for libudev, it wouldn't be including the file for it
2020-07-18T17:58:30 #kisslinux <himmalerin> I can't believe missed dylan again
2020-07-18T17:59:06 #kisslinux <himmalerin> Well, maybe he'll read the logs. Is something like http://ix.io/2riA ok for my personal repo? Don't want to misuse the logo or anything
2020-07-18T21:27:28 #kisslinux <dilynm> claudia02: my community PR for ECM 5.72 is gross, hence why it wasn't merged. Oops
2020-07-18T21:27:33 #kisslinux <dilynm> I'll have to look into the attics issue
2020-07-18T21:27:36 #kisslinux <dilynm> Attica*
2020-07-18T21:27:48 #kisslinux <dilynm> E5ten: I thought I pushed a fix for the kwin issue
2020-07-18T21:28:42 #kisslinux <dilynm> I did; it builds now. Unsure if it works, I don't see why it wouldn't. Also can't speak to why it complains about that header :/
2020-07-18T21:33:03 #kisslinux <dilynm> Attica builds fine for me
2020-07-18T21:43:02 #kisslinux <claudia02> dilynm: It was complaining that ecm is not a suitable version but since ecm has no other dep than cmake I didnt know what to rebuild. But honestly I spent not much time digging. I will try again soon
2020-07-18T21:56:06 #kisslinux <dilynm> It should just be as easy as upgrading ECM... Are you certain you had the right version?
2020-07-18T21:56:24 #kisslinux <dilynm> I've had times where ECM got rolled back because I forgot I had ran kiss u and it got downgraded
2020-07-18T22:00:14 #kisslinux <claudia02> I hope I didnt bother you because of inaccurate reading :D
2020-07-18T22:20:50 #kisslinux <dilynm> XD you're fine. You've motivated me to actually work on something
2020-07-18T22:23:58 #kisslinux <dilynm> shared-mime-info is cancer
2020-07-18T22:36:38 #kisslinux <claudia02> yes.
2020-07-18T22:37:17 #kisslinux <claudia02> For what is this acutally needed? To doubleklick on a file and then open the right application? I thought xdg-open does this kind of things
2020-07-18T22:37:56 #kisslinux <claudia02> Our webkit2gtk is not able to open a local html file without it..-.-
2020-07-18T22:39:32 #kisslinux <dilynm> kio-extras needs it
2020-07-18T22:39:47 #kisslinux <dilynm> And kio-extras is needed for... Thumbnails to appear in certain applications
2020-07-18T22:40:23 #kisslinux <dilynm> I think I spoofed shared mime info well enough to get away with it, but now kio-extras complains about taglib being broken (:
2020-07-18T22:40:47 #kisslinux <dilynm> This project sucks LMFAO why didn't I just learn C instead
2020-07-18T22:42:36 #kisslinux <claudia02> Its never to late to turn around ;)
2020-07-18T22:44:22 #kisslinux <dilynm> I've come too far to turn back now