00:01:47 hm, the website says it was released a month ago, I wonder why portscout didn't pick it up and notified the maintainer 00:02:24 you should open a problem report to notify the maintainer of the new version 00:02:48 i dont really know how to do that 00:02:55 but i can learn 00:02:57 :) 00:11:52 you need a account on https://bugs.freebsd.org/bugzilla/ and then use the 'Report an update or defect to a port' link there, the summary should start with the port name so games/nethack36: in this case 00:11:53 Title: FreeBSD Bugzilla Main Page 00:14:23 as a fast but not so useful alternative (as the rest of the project can't really see it) you can email the maintainer directly (and if the maintainer has no time currently nobody else can pick it up as nobody knows) 00:15:02 it wont let me type in the component or version box on bugzilla 00:17:01 you don't need to change them, they are already correct 00:17:51 ok 00:18:28 (they are about the ports tree, not the individual port) 00:20:55 there we go, all sent 00:57:49 anyone with py3 able to make a quick venv and see if they can install the cryptography package? I've been chasing issues with it and just want to make sure I am not the only one 00:59:06 (https://pypi.org/project/cryptography/) 01:15:50 Hello. I am trying to upgrade a port I maintain (www/hiawatha). When I see the Makefile as it currently is, I see there is no exact filename for MASTER_SITES and the ports tree pulls down the old version. How do I tell the ports tree to pull down the more recent version of the source code? Am I thinking about this all wrong? I am also going off of: https://docs.freebsd.org/en/books/porters-handbook/upgrading/#preamble 01:15:51 Title: Chapter 11. Upgrading a Port | FreeBSD Documentation Portal 01:25:00 hey jauntyd, this question has better chances of being answered in #freebsd-ports channel 01:25:18 (you can also try #bsdports on efnet) 01:27:32 ty 04:14:19 /j #fog 04:14:22 oops 04:51:45 I suddenly miss 2 devices under /dev/sndstat and have no idea where to start :( 04:57:35 any idea where to start? 05:25:39 So, the port net/wireguard is a metaport which installs wireguard-kmod, whose descr says is "new, unvetted, possibly buggy," etc. Should I be worried and used the "go" version instead? 06:24:22 ghoti: no 06:24:56 the scare warnings should probably go away, we deemed it ok enough to go into base 07:48:45 ghoti: the current wireguard code has been thoroughly scrutinized and reviewed. 07:56:10 I have installed freebsd on one disk. Now i want to use raid0 to add another disk. So now my OS will be on 2 disks via raid0. Can I add the second disk in freebsd or I have to reinstall the OS again? note that the OS is disk encrypted via GELI I think. The other disk should be encrypted as well. 08:02:20 Letiute: what filesystem do you use? if it is zfs it is easy to add drives to a pool, with ufs no idea 08:02:50 zfs 08:03:23 nimaje will it be raid0 after adding drive and can I configure raid0 like that? 08:11:07 nimaje thre? 08:16:29 nimaje I am worried about freebsd and zfs community support. Not much users as compared to linux 08:20:26 you want to setup encryption for your disk and then add it see https://docs.freebsd.org/en/books/handbook/zfs/#zfs-zpool-attach (you probably want to read the whole page not just that section) 08:20:28 Title: Chapter 21. The Z File System (ZFS) | FreeBSD Documentation Portal 08:20:45 nimaje the disk is already encrypted . I chose it during installation 08:20:54 now I am adding a new disk. 08:20:59 thats the thing I want to confirm 08:23:08 yeah, the new disk you want to add, the other one is already in your pool 08:25:11 will the new disk be GELI encrypted? 08:26:16 don't think so, I have to do it myself. OR reinstall freebsd, choose raid0 at installation. IF I do so, will install process encrypt BOTH disks via GELI? 08:26:20 that was the only question 08:33:55 you have to setup the encryption yourself (installer would encrypt all disk if you do a encrypted setup with multiple disks) 08:34:17 I gues resinstalling and choosing "striping" with two disks would be quicker than i do it myself AND BOTH disks will be encrypted with GELI. correct? 08:34:28 https://docs.freebsd.org/en/books/handbook/bsdinstall/#bsdinstall-part-zfs 08:34:29 Title: Chapter 2. Installing FreeBSD | FreeBSD Documentation Portal 08:34:59 nimaje if I do while installation process, both disks will encrypt. Correct? 08:37:01 if you put both in the stripe there and choose to encrypt disks then both will be encrypted 08:37:55 I see. that would sum up. thanks! 08:38:11 nimaje better for me to just reinstall... 08:39:56 When I add disk to simple pools for storage, the data is written on random places ie some files will go on one drive, others on other. If I loose one drive, that would mean I might loose some random/unknown data? Not talking about raid this time. 08:43:47 not sure if zfs can rescue any data if a part of a stripe fails 08:46:41 it cant 08:46:46 I think 08:46:48 what level of zstd compression is unsane for OS and database? zstd-15 is ok? 08:47:14 nimaje but I was not talking about raid/strip this time as said in last line of parah 08:47:23 what will happen i nnormal pool? 08:52:08 is there any know-working solution on 13-RC2 for listening to music (either natively or in linuxolator)? 08:54:11 a pool is a stripe of vdevs, where vdevs can also be single disks, so in your setup you would have a stripe of two single disk vdevs, so if a disk fails the corresponding vdev fails, not sure if zfs can recover from that situation 08:55:08 elgrande: any music player? what problems do you have with listening to music? 08:59:51 spotify and apple music neither work anymore in linux brave nor in freebsd firefox nor in freebsd chromium. testing linux chrome now. 09:02:50 cannot call chrome 09:03:00 elgrande: I'm using Deezer and Qobuz several hours a day and it works just fine with native Firefox. 09:03:12 elgrande: Can you give more details on what doesn't work? 09:03:23 I am not allowed to get a third abo 09:03:47 My abo is not recognized, I am logged in, but can only listen to the 30 seconds version 09:03:59 elgrande: Like Spotify show an error saying it can't play? Or it starts to load the stream, but the seek bar stays at 00:00? Or it plays the stream but you hear nothing? 09:04:07 and chrome is totally broken on ubuntu 09:04:35 dumbbell nothing happens when I press play 09:05:42 elgrande: Can you play a video and hear the sound on YouTube for instance? 09:05:55 I can hear sound on native apple music, too 09:06:05 but only 30 seconds 09:07:22 can I rollback to 13-1 somehow? 09:07:36 maybe it helps 09:08:16 elgrande: Did you try YouTube? 10:53:55 Hi 10:54:01 the pool name set by freebsd is zroot I think. I had another drive named ssd1. How can I use it? `zpool import ssd1;` ? If I do, it says cannot import 'ssd1': pool was previously in use from another system Last accessed by xubuntu (hostid=53eb6008) at Thu Mar 16 10:57:50 2023. The pool can be imported, use 'zpool import -f' to import the pool. 10:54:01 Is this safe? 10:54:20 will this mess the OS or I create a new pool? 10:55:06 You need to export your pool from the system it was last used on. 10:55:57 debdrup am.. ok that system is 'gone'... 10:56:13 debdrup what do i do now? 10:56:56 Then you don't have much option but to use -f; it _should_ be alright, if the pool was created recently. 10:57:34 debdrup ok, so if it hadn't been created recently on linux, I would have been in trouble? 10:57:43 debdrup also, shouldn't I create a new pool? 10:57:45 Not necessarily. 10:57:57 Why do you think you should create a new one? 10:58:00 importing it in root bool is ok? 10:58:14 importing it in root p*ool is ok? 10:58:16 I don't think you understand how ZFS' pooled storage works. 10:58:44 I agree with you. this is my second day and a lot of reading and paranoid behaviour is exppected :) 10:59:18 Is zstd-19 the highest compression? 10:59:21 A pool can consist of an arbitrary number of disks (well, there is a limit, but it's pretty ridiculously high), in various configurations. 10:59:38 debdrup understood 11:00:06 zpool is one pool and zroot is some primary drive label. right? 11:00:06 No, zstd goes arbitrarily high up to zstd-1000 - but at that point you're wasting so much cputime, it's probably not worth it. 11:00:13 No. 11:00:22 zpool is the name of the pool of drives. 11:00:35 debdrup not even for backups where storage capacity is primary concern? 11:00:46 yes and zroot is the drive? 11:00:49 no. 11:01:33 debdrup the command says " 'compression' must be one of 'on | off | lzjb | gzip | gzip-[1-9] | zle | lz4 | zstd | zstd-[1-19] | zstd-fast | zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]' 11:01:34 " notice that zstd-[1-19] VS zstd-fast-[1-10,20,30,40,50,60,70,80,90,100,500,1000]' 11:02:13 Right, sure - the ridiculous levels are locked behind zstd-fast. 11:02:43 Oh right, those are negative levels. 11:02:50 so zstd-1-19 is higher compression? 11:03:24 in a nutshell zstd19 is the highest or zstd-fast-1000 is? 11:03:24 zstd-19 is the highest compression, but at that level you can end up in a situation where one backup can't complete in the time it takes before a second one starts. 11:04:00 ok 11:04:14 debdrup i cannot find a list of compression ratio for each 11:04:14 I think you need to buy a book on ZFS and read it. 11:04:24 * Letiute reads handbook 11:04:32 but cannot find the benchmar for1-19 11:04:46 You won't find a benchmark. There are too many variables. 11:05:00 ok what level might 19 give? 11:05:02 rough idea? 11:05:10 and what might 3give 11:05:16 Does your CPU have SSE2 or not? How well does your data compress? 11:05:28 Nobody can tell you, if they're being honest. 11:05:29 its textual mostly 11:05:36 I can make up a number, but it won't be right. 11:05:40 I know 11:05:44 differs with data type 11:05:58 just wanted textual data idea 11:06:05 You're not gonna get it from me. 11:06:11 ok :) 11:06:14 no issues 11:06:19 I am setting it to 19 for backup 11:06:25 and 7 for database + os 11:06:27 kevans do we have a fork() function in our flua? 11:07:01 one concern is that zfs won't recompress old files if I change algorithm. How can I achieve that. BTRFS had defrag doing this automatically 11:07:13 Letiute: it can't. 11:07:25 ZFS would need block pointer rewrite, and that's not happening. 11:07:58 debdrup not even if I use `mv` on same disk? 11:08:03 Block pointer rewrite also breaks the implicit promise of copy-on-write, which is to never rewrite already-written data. 11:08:31 I have no idea what sort of situation you're talking about where what you're saying makes sense. 11:08:59 Why move things around on a disk? It achieves nothing. 11:09:55 ok 11:10:08 Also, it doesn't matter what BTRFS has. ZFS isn't BTRFS. 11:10:33 You need to absolve yourself of this notion that just because X can do something, Y should be able to do it too. 11:10:55 so rsync /old/ /new/ ; rm -rf /path/to/old 11:11:02 you can use something like this: cp -a /current /new && rm -rf /current && mv /new /current 11:11:03 Okay, just stop. 11:11:05 debdrup understood 11:11:32 You've been asking the exact same questions in two different channels, and you're getting different answers. So that tells me you're going to not remember half of what you've been told, and then you'll blame someone else when things inevitably go wrong. 11:12:27 thats your wrong positively made assumption. Also, I am getting different perspectives in each channel 11:12:42 VVD I SEE 11:12:45 Have some respect for the people whose time are being taken up by helping you, and go study how ZFS works instead of just asking endless questions that won't help you understand anything about the underlying concepts. 11:12:49 VVD I see 11:13:14 debdrup I just respect your opinion to which I differ 11:13:32 but I agree with the notion. 11:14:27 How to make these settings , mountpoints and compression ration permanent 11:14:31 after reboots 11:14:41 You set them. 11:14:43 <_xor> kevans: Around? I noticed you're the maintainer fdor the Arduino ports. 11:16:16 debdrup and the OS will remember the values or the pool/drive itself willr emember it in metadata/info? 11:16:32 Letiute: go read a book on ZFS. 11:18:01 debdrup https://freebsdfoundation.org/wp-content/uploads/2021/05/Zstandard-Compression-in-OpenZFS.pdf 11:18:42 Yes, that's a known set of data; namely the FreeBSD source code. 11:19:03 Unless the data you're storing is the FreeBSD source code, it won't tell you anything. 11:19:42 I said "text" earlier but I just link fyi 11:19:47 code = texzt 11:20:00 No, source code compresses better than text. 11:20:16 I wonder how 11:20:23 Because of how compression works. 11:20:37 code has more reused chars? 11:20:39 thats why? 11:20:41 It looks for patterns in data, code has more patterns. 11:25:24 a lot of users say "text" but mean "docx/odt", but they are zip 11:26:10 The open document formats use DEFLATE? 11:26:16 As in, by default? 11:26:28 VVD Does the OS will remember the values of compression, mountpoing etc or the pool/drive itself will remember it in metadata/info? Unable to find it in a "book" 11:26:53 Welp, you clearly don't read what you're told. 11:27:15 what did I missed? 11:28:41 [1678965281] You set them. 11:32:49 debdrup, compression method: deflated 11:32:49 compression sub-type (deflation): normal 11:33:03 VVD: Huh, interesting. 11:33:42 zipinfo -v random.ods 11:34:15 Wait, so can you pass random.ods through tar? :P 11:34:28 libarchive will decompress any compressed file I've thrown at it. 11:34:42 tar - no, it's zip archive 11:34:43 debdrup my question is (elaborating) that when I "set" them, "where" is that setting saved? in OS setting, drive meta, /FS . where? 11:35:05 I read "1678965281] You set them." the first time 11:35:20 zfs set are dataset options, not OS options 11:35:33 VVD thanks for the simple asnwer 11:35:35 answer 11:35:36 Letiute: it's metadata. 11:35:46 debdrup thanksf or the simple answer 11:35:48 zpool set are pool options 11:35:51 ok 11:36:03 All of this is covered on the actual books that cover ZFS. 11:36:47 sometimes 1st steps are hardest 11:37:08 Doesn't mean you have to make it harder for yourself. 11:38:04 debdrup people come to irc, forrums etc to discuss insecurities, opinions etc. books have their place. Else irc would shutdown. and yes; what VVD said. Also, by no means this comment covers my incompetence to read more in less time. 11:38:27 and no means it is not respecting others time 11:38:33 * Letiute closes the topic from his side 11:39:04 Letiute: irc isn't a doctors visit, you don't need to solicit a second opinion. 11:39:36 If you want to test things, you set up a temporary pool, and try your hardest to break it. 11:39:47 Worse-case, you trigger some bug that you can report, and it gets fixed. 11:39:55 Letiute: may I suggest you to read : https://mwl.io/nonfiction/os#fmzfs 11:39:56 Title: Operating Systems – Michael W Lucas 11:40:13 * Letiute closes the topic from his side 11:41:56 When I started using ZFS in 2008, I'd spent a few prior weeks testing all manner of things, deliberately trying to break them. I didn't succeed, so I felt fairly confident in the data resiliency claims of ZFS and have used it ever since. 11:42:36 debdrup: i just beat the crap out of my new server, repeated os install, pop drives, rebuild, pop drives, try to reformat or trash, satisfied, reinstall 11:42:53 Demosthenex: that's a more risky way of accomplishing the same 11:48:26 had a lot of hard time fixing mic issue. I have two options it seems. one is builtin speaker/mic /mixer5. speaker working fine. another /mixer8 is external usb mic. How to change to that? the += command does not work... gtk-mixer all volums up 11:54:22 https://pastebin.mozilla.org/xhyo2TX8 similar issue https://forums.freebsd.org/threads/internal-microphone-setup-question.77091/ 11:54:23 Title: Mozilla Community Pastebin/xhyo2TX8 (Plain Text) 12:20:25 Letiute: look at virtual_oss 12:21:45 i use that as a virtual device and hot swap between mics (ie: usb and local) 12:29:41 Demosthenex doing; thanks! 12:34:45 Demosthenex don't see anythign https://imgur.com/mItcz5e.png 12:40:08 Letiute: https://dpaste.org/XbRLJ 12:40:10 Title: dpaste/XbRLJ (Plain Text) 12:42:51 Demosthenex sysrc -a | grep oss prints nothing 12:44:30 Demosthenex should I do virtual_oss_cmd /dev/dsp.ctl -R /dev/mixer8 -P /dev/mixer8 ? 12:44:46 mixer8 is my mic I guess 12:45:24 dumbbell yt music works with add blocker, thanks 12:48:51 elgrande: Then I would clear cookies for Spotify/Apple Music and reconnect. Look if ffmpeg is up-to-date too, native Firefox depends on it. 13:01:03 Letiute: try mixertui, and get a list of your devs there 13:12:39 what am I missing https://bsd.to/Mm9L 13:12:41 Title: dpaste/Mm9L (Plain Text) 13:12:46 vbox related 13:16:29 a solved. 13:16:30 unable to select pcm8 USB audio for F4 [Capture]. If I do, nothing is shown https://imgur.com/5AsPKd8.png 13:18:11 it is somehow not assigning recording device as pcm8 /mixer8 13:19:29 Demosthenex ^ 13:19:54 instead it is showing playback for it 13:20:03 not using playback but showing 13:20:06 that should not be the case 13:32:42 does kqueues EVFILT_VNODE work to be notified of changes to files/dirs in a nullfs mount within a jail? 13:35:44 Demosthenex can I have pavucontrol in freebsd? 13:35:51 that worked with same hardware on llinux 13:36:35 Letiute: yes. i think it has its own package. 13:36:43 dont see in search 13:37:00 ok got it 13:37:02 i see pavucontrol as well as pavucontrol-qt 13:40:18 rejoining 13:42:39 wb 13:45:19 phryk: re kqueues and EVFILT_VNODE: my naïve guess would be: probably. 13:47:05 i have no idea 13:51:39 pavucontrol did it. but I was not able to get it working on chrome. works on firefox. 13:51:43 any ideas why? 13:52:46 also, cannot make volume more than 100% 13:56:13 chrome can't even see any mic except default option (which is incorrect/not working) 13:58:49 maybe this helps but unable to understand https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=245251 13:58:52 Title: 245251 – www/chromium: Doesn't allow to choose a microphone 13:59:51 Letiute: so firefox and chrome i think pick the first audio device 14:00:14 which is why i use dsp (no number) with virtual_oss 14:00:21 it's first selected 14:00:24 Demosthenex no, firefox shows full list of devices 14:00:27 freebsd audio is a bit different from linux 14:00:28 not chrome 14:00:37 and once i got it setup, i've never had to change it again 14:00:51 Demosthenex what commadn should I use then? 14:01:33 virtual_oss_cmd /dev/dsp.ctl -R /dev/mixer8 -P /dev/mixer8 ? 14:01:47 this will make it default? 14:04:39 _xor: arduino, you say? *runs away* 14:05:20 Letiute: you have to setup the virtual_oss service first 14:05:29 virtual_oss_cmd is for changing the service while it's running 14:10:28 Demosthenex what commands do I need to set it up? 14:10:56 meena: heh, guess i'm just gonna have to try then^^ 14:11:53 phryk: usually quicker and easier than waiting for a qualified answer 14:12:47 but once we do know, and it seems to run against our intuition, we can open a bug, or fix up the docs 14:12:54 meena: not in my case, tho. i have very little C experience and want to do it from D, with which i have even less experience. Plus I don't even know yet when I'll get to work on this :'D 14:14:34 which D? dtrace D or walter bright dlang? 14:15:58 C is very easy, once you accept that it's impossible to get right. or even close to sensible 14:17:47 dlang. 14:18:35 I want a compiled language that's more sensible than C and C++, but i want to be able to use C APIs, because that's just the standard. D is the only language that ticks all those boxes. 14:19:21 wait for cppfront :-) 14:19:26 Letiute: the sysrc variables i put in the dpaste link are what i used 14:19:48 paulf: that sounds like one of those "let's just add an abstraction layer on top" kinda deals. i think i'll pass on that.^^ 14:19:52 Zig for C, Rust for C++, D if you have infinite time to contribute yourself into a black hole 14:20:10 Demosthenex sysrc -a | grep oss shows nothing 14:20:15 https://dpaste.org/XmQxv 14:20:17 Title: dpaste/XmQxv (Python) 14:20:29 phryk that's waht C++ was originally 14:20:32 meena: I *am* kinda planning on writing my own GUI toolkit :'D 14:20:41 paulf: yeah, and look how that turned out :D 14:20:48 Letiute: if you do sysrc -a | grep oss and have nothing, you haven't customized it. i gave you my settings for reference 14:20:59 it's not over yet, by a long way 14:21:12 speaking of gui toolkits, does the freebsd installer use dialog or something else? 14:23:27 Letiute: the second paste is my puppet config which sets things for sound 14:23:55 looks like dialog, but I am only sure for ports 14:24:57 https://man.freebsd.org/dialog(1) 14:24:58 Title: dialog(1) 14:27:10 fun. damn, i keep wanting a TUI library for common lisp that isn't an ncurses wrapper :P 14:28:05 hrm, now that i think about, iirc, bapt replaced dialog with something else, but i don't remember if that's done, actually, and what the name is 14:28:29 with bsddialog? :) 14:29:52 yeah, which has no man page 14:30:16 oh, it does 14:30:31 it's mit bsdialog, it's actually bsddialog 14:31:00 I guess somebody didn't like the implications that they're producing BS 14:31:23 Demosthenex unable to follow what exactly needs to be done here 14:31:48 Letiute: install virtual_oss. configure and start the service. then administer it with virtual_oss_cmd. 14:31:48 "bsddialog was written by Alfonso Sabato Siciliano .", so i guess it has nothing to do with BS (being bad) 14:32:52 yuripv: maybe monolog would have been a cool name? it's not too late. 14.0 has not been released yet 14:33:42 that would get the system stuck talking to itself then? :) 14:34:34 gosh that would be cool, if it had an easily scriptable backend 14:34:53 so you feed it a seed file and watch it install itself 14:35:25 but i feel like we have different options for those scenarios already 16:43:59 Greetings all. What NIC + driver combo would I need on 13.1 to get NBASE-T working on FreeBSD with my ISP's new 2.5 Gbit connection? Searching for FreeBSD + NBASE-T on the web and searching NBASE-T in the forums is not getting me any answers. I don't want to waste money buying something that won't work. TIA! 16:44:43 FTR, I am planning to build a new router box from scratch for this. 16:48:30 unixman_home: great nick. what's NBASE-T anyway 16:49:25 ah, not quite 10G and on copper, got it. 16:49:30 i'd suggest starting with card models ;] 16:50:37 Yeah, it is also known as multi-rate. Basically rates between 1 Gbit and 10 Gbit. 16:51:03 pick a card first, then confirm support 17:00:13 I looked at https://ark.intel.com/content/www/us/en/ark/products/189463/intel-ethernet-network-adapter-x710t2l.html but do not want to spend that much before I know it will work for multi-rate / NBASE-T with FreeBSD 13.1. Pretty sure those use the ixl driver, but I have not gotten firm information that the driver supports multi-rate successfully. I can't just toss money at the problem with trial and error. 17:00:14 Title: Intel Ethernet Network Adapter X710T2L Product Specifications 17:01:05 I was hoping someone here had experience getting NBASE-T working and would recommend a NIC. 17:06:08 seems like last 2 versions of firefox make firefox not create new net connections and even close some. making it unusable 17:20:02 unixman_home: might be better to post on the forum, asking about the ixl driver 17:20:07 given how specific that question is 17:31:17 Yeah, Demosthenex, I am planning to ask on the forum as well. But wanted to check here first. Thanks. 17:40:38 I'm sticking with Google Fiber's 1 Gbit service, until they ditch the requirement to use their router for 2 Gbit. 17:40:59 'Course. . . the 1 Gbit is stupid fast for a home connection. 17:49:06 I've been digging into a test failure for a piece of software, and I think the problem comes down to posix_fallocate(2) fails on FreeBSD/ZFS, but I cannot find much discussion about that on the internet at large 17:53:11 * rtyler is rather surprised 17:59:05 elgrande: ports uses dialog4ports, appropriately enough 18:01:48 unixman_home: are you sure whatever's at the other end implements NBaseT properly? A lot of the teething problems of NBaseT seems to relate to improper implementations. 18:02:11 nmz-: I can't say I've been experiencing that; have you tried with a brand new profile (created with firefox -P)? 18:09:00 Hi. I understand the difference between pkg and using ports, but my question does the same source / binary ways exist to update maintain the core system? Like, is there a source based way of using the system instad of freebsd-update if I am on -RELEASE? Don't have a solid reason except tuning the system with march=native 18:10:12 acommonman: Yes. 18:10:21 acommonman: read build(7) 18:10:26 Ok Thanks 18:10:37 debdrup, that's it 18:10:39 thank you 18:10:56 Alternatively, https://docs.freebsd.org/en/books/handbook/cutting-edge/#makeworld 18:10:57 Title: Chapter 25. Updating and Upgrading FreeBSD | FreeBSD Documentation Portal 18:11:37 thanks 18:11:40 acommonman: do note that march=native doesn't do anything, since nothing in the FreeBSD base system (with the exception of OpenZFS, which does it irrespective of what the build system says) takes advantage of SSE and MMX. 18:13:01 This is deliberate because the vast majority of data, both in terms of frequency and size, are sets of data that aren't very big - so because there's a latency associated with using all SIMD instructions like SSE and MMX, it can take _longer_ to execute short bursts when using SIMD instructions. 18:13:17 debdrup, I see. Well, apart from march=native Gentoo also use https://wiki.gentoo.org/wiki/CPU_FLAGS_X86 .. They generate it with a python script cpuid2cpuflags which can be put in make.conf. No such CPU instruction set declaring options for FreeBSD to take advantage of SSE / MMX and newer sets? 18:13:18 Title: CPU_FLAGS_X86 - Gentoo Wiki 18:13:33 FreeBSD isn't Gentoo. 18:13:48 debdrup, I know it isn't. I meant is there a similar option to force it? 18:14:13 acommonman: there's nothing in the base system that takes advantage of it. 18:14:26 Ok Got it 18:14:31 Also, watch https://www.youtube.com/watch?v=r-TLSBdHe1A 18:14:32 Title: "Performance Matters" by Emery Berger - YouTube 18:14:41 Thanks 18:15:08 acommonman: though, you can set CPUTYPE= yourcpu to maybe gain something 18:15:16 acommonman: no, it won't gain anything. 18:15:32 That's the entire point of what I just said. 18:16:10 I understand 18:16:39 he can try and judge for himself. There is such possibility to set cpu optimized variable 18:16:55 angry_vincent: that's why I linked to the presentation above. 18:21:03 It's not a matter of "judge for yourself"; statistics is an actual field of science where we have the tools to figure out if A´ is faster than A, with the correlary that if we don't control for the act of optimization, we don't know if A or A´ is faster despite the fact that A´ may appear faster. 18:21:11 acommonman: read this one https://cgit.freebsd.org/src/tree/share/mk/bsd.cpu.mk 18:21:12 Title: bsd.cpu.mk « mk « share - src - FreeBSD source tree 18:21:26 acommonman: if you still want to try 18:22:14 acommonman: i have set CPUTYPE= sandybridge for my cpu 18:22:24 acommonman: in /etc/make.conf 18:23:10 CPUTYPE doesn't build code that suddenly includes SIMD instructions. 18:25:06 debdrup: that talk looks class 18:25:24 meena: there's also a follow-up that's called "performance really matters" 18:27:34 In reality, the binaries produced by FreeBSD target the Pentium Pro micro-architecture, according to https://cgit.freebsd.org/src/blame/contrib/llvm-project/clang/lib/Driver/ToolChains/Arch/X86.cpp#n115 - but even if you change that, it doesn't suddenly produce machine-code that includes SIMD instructions. 18:27:35 Title: X86.cpp « Arch « ToolChains « Driver « lib « clang « llvm-project « contrib - src - FreeBSD source tree 18:31:26 debdrup, got it. But the ports are a different matter right? Can they benefit from CPU instructions? 18:32:19 acommonman: that depends on their build infrastructure, which ports have no control over. 18:32:30 The third party software's build infrastructure, I mean. 18:32:39 Oh ok 18:37:45 debdrup, I have no way to test the NBASE-T implementation yet as I don't have a system that can take advantage of it. Once I have hardware that I know is supposed to work, then I can bug my tech guys at the ISP if it does not. That said, one of my cow-orkers has a MicroTik router build that he showed me is working at 2.5 Gbit with the local ISP. 18:39:16 unixman_home: most of the implementation problems I've heard of have been from NICs in computers to switches on a local network (which, incidentally, is also, seemingly,the biggest use-case, since it lets people get more bandwidth without upgrading cables) 18:39:24 unixman_home: i have various aquantia (or how it's really spelled) working on gigabyte motherboards with a driver from ports, not that i could really "recommend" it, didn't do any real bandwidth testing 18:40:36 s/MicroTik/MikroTik/ up there. 18:41:16 yuripv: have that port been updated to support the newer NICs? 18:41:40 I will be going direct from the cable modem to my router. Any switches will be after the router. As long as the router gets full rate, I will be happy. 18:41:51 not sure, both my nics (5gb and 10gb ones) worked with the version from an year ago 18:43:26 https://github.com/Aquantia/aqtion-freebsd/commit/c664598a0f65a562e726c19b29835bce9183a036 was committed after the last update, so it looks like no. 18:43:27 Title: adds aqc107 with AQ_DEVICE_ID_0001 · Aquantia/aqtion-freebsd@c664598 · GitHub 18:43:32 last update to the port, I mean 18:44:14 Planning to get a switch that can take a 10Gbit SFP to put between the router and the LAN. Overkill for now, but ISP is talking about eventually providing 10Gbit to the home, so I am planning ahead. 18:44:35 unixman_home: Is the router some sort of DOCSIS thing? 18:45:02 Yeah, it is DOCSIS cable modems. 18:45:45 Yeah, NBaseT for DOCSIS modems has come up a lot in a thread on a forum I'm on. 18:45:52 FTR, this is not my current ISP. I will be adding the other ISP and will be paying for two connections until I get this working. 18:46:36 They're apparently all the exact same Broadcom chipset, but reportedly don't mix well with Intel NICs. 18:46:36 They just got their service out to my neighborhood here in the sticks. 18:47:11 Sounds like "fun" debdrup. :D 18:48:25 I don't know about you, but I tend to view "We're Intel who dual-licenses code as both BSD and GPL" with a bit more respect than "We're Broadcom and are going to buy up every competitor and stop releasing software updates to push customers onto newer products".. 18:49:01 Same. Especially the BSD license part. 18:49:08 * unixman_home > back_to_work 20:25:00 Are versions and release channel of packages in ports / pkg same across presently supported bsd versions? Like if I am using FreeBSD 12.x-RELEASE / 13.x-RELEASE and 14.X-CURRENT is the ports or pkg (both quarterly or latest same? 20:25:15 Sorry if the question is noobies. I am still trying to understand how FreeBSD is different than Linux 20:25:22 ***noobish 20:27:13 acommonman: checkout the table here for instance https://www.freshports.org/devel/dbus/ 20:27:14 Title: FreshPorts -- devel/dbus: Message bus system for inter-application communication 20:27:44 meena, thanks. let me check 20:29:13 meena, thanks 20:29:15 got it 20:29:20 that explains 20:33:23 Just love how FreeBSD handles the separation of the system and the ports/ pkgs 20:35:54 acommonman: Absolutely. 20:37:14 I mean its something that is a total mess in a lot of LTS linux distros like Debian / Redhat / SUSE 20:40:23 One thing I appreciate even more than that is that if you just use programs like they are as they're installed, the config gets removed - but if the config gets changed, it gets left behind when the software is removed. 20:40:57 This lets you install software, configure it, deinstall it, and use what's left to create a custom image that installs exactly as you want it. 20:44:20 Ahh. That's interesting 21:11:44 is there any fast way to setenv a variable with ! without escaping all of them ? 21:11:51 I tried quotes, double quotes, it still hangs on ! 21:11:58