ChanServ changed the topic of #haiku to: Open-source operating system that specifically targets personal computing. | https://haiku-os.org | Nightlies: https://download.haiku-os.org | Bugtracker: https://dev.haiku-os.org | SCM: https://git.haiku-os.org/ | Logs: https://oftc.irclog.whitequark.org/haiku | Matrix: #haiku:matrix.org | XMPP: #haiku%irc.oftc.net@irc.jabberfr.org
nipos has left #haiku [Disconnected: Replaced by new connection]
nipos has joined #haiku
bjorkintosh has joined #haiku
xet7 has joined #haiku
<nekobot> [haiku/haiku] waddlesplash pushed 2 commits to master [hrev57897] - https://git.haiku-os.org/haiku/log/?qt=range&q=ff890d04b398+%5E917dcdfe7599
<nekobot> [haiku/haiku] 4448758b3e1e - app_server & Interface Kit: Rework cursor reference management.
<nekobot> [haiku/haiku] ff890d04b398 - app_server: Acquire the lock in GlobalFontManager::GetStyle().
<nekobot> [haiku/haiku] waddlesplash pushed 2 commits to master [hrev57898] - https://git.haiku-os.org/haiku/log/?qt=range&q=587c5ecdac2d+%5Eff890d04b398
<nekobot> [haiku/haiku] 61f7048ea38d - RAMFS: Properly set node->ops in ramfs_get_vnode.
<nekobot> [haiku/haiku] 587c5ecdac2d - kernel/fs: Check for missing vnode fields in get_vnode.
diver1 has joined #haiku
Diver has quit [Ping timeout: 480 seconds]
Diver has joined #haiku
diver1 has quit [Ping timeout: 480 seconds]
SLema has quit [Read error: Connection reset by peer]
SLema has joined #haiku
AlienSoldier has joined #haiku
<AlienSoldier> juat upsated my haiku 32bit install. the Haiku "about"never load and crash.
<AlienSoldier> *just
<AlienSoldier> **just updated
SLema_ has joined #haiku
SLema has quit [Read error: Connection reset by peer]
SLema_ has quit []
<AlienSoldier> from sysinfo Kernel name: kernel_x86 built on: Jul 30 2024 06:02:46
SLema has joined #haiku
<AlienSoldier> qmplay 2 still close automatically on going full screen
<AlienSoldier> ~> /boot/system/data/deskbar/menu/Applications/QMPlay2
<AlienSoldier> [30 Jul 2024 23:17:38.974] QGuiApplicationPrivate::processMouseEvent: Got NaN in mouse position
<AlienSoldier> [30 Jul 2024 23:17:39.023] QMPlay2 crashed (SIGSEGV)
<AlienSoldier> Segmentation violation
AlienSoldier has quit [Quit: Vision[]: i've been blurred!]
diver1 has joined #haiku
Diver has quit [Ping timeout: 480 seconds]
mmu_man has quit [Ping timeout: 480 seconds]
flag has joined #haiku
flag_ has quit [Ping timeout: 480 seconds]
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
OscarL has joined #haiku
OscarL has quit [Read error: Connection reset by peer]
OscarL has joined #haiku
OscarL has quit [Read error: Connection reset by peer]
OscarL has joined #haiku
<OscarL> seems that the x86_64 haiku builder went down.
<OscarL> (while doing the nightly build)
axeld has joined #haiku
Begasus has joined #haiku
<Begasus> g'morning peeps
<OscarL> hello Begasus.
<Begasus> Hi OscarL, still up I see :)
<OscarL> yup. messing around with some VMs (trying to make them behave better on my local network).
<botifico> [haikuports/haikuports] Begasus pushed 1 commit to master [+1/-2/±0] https://github.com/haikuports/haikuports/compare/c5cb2faf6f12...b771ded643fb
<botifico> [haikuports/haikuports] Begasus b771ded - mpvqt, bump version (#10775)
<Begasus> updated gettext here yesterday, that cycling dependency in there is giving me a headache :P
<OscarL> now at least I can connect *to* a Haiku VM (via ssh/sftp) from other machines (but still not from other VMs on the network, lol).
<OscarL> ooof... yeah not easy when deps start to eat their own tails :-)
<Begasus> haven't been fidling with that here yet
<Begasus> plasmatube kept me busy for a while yesterday, but as it is "atm" it's pretty good (give and take some annoyence) :)
<Begasus> with that update on mpvqt it doesn't open the videos in an external window anymore
TMM has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
TMM has joined #haiku
<OscarL> "Issue with Python 3.10 package"... no, it isn't.
<Begasus> heh
<Begasus> still not sure what's going on with that
<Begasus> OscarL, what's the stats on those 2 PR's atm?
<OscarL> korli already tried rebuilding pyqt_sip. Waddlesplash suggested a rebuild of Qt libs on 32 bits.
<Begasus> I'm not going to check a revbump on 32bit for Qt5 :P
<Begasus> I "might" trigger a rebuild on the buildmasters (as suggested on the topic)
<Begasus> atleast rebuilding it here on 64 is "acceptable" in speed and time, but not on the other laptop :)
<Begasus> anyway, qt6 should already been good for gcc changes
<OscarL> re my 2 open PRs... 10750 (python3.10) should be OK, but not sure if my patchset changes were reviewed (no one said much on that :-D).
<botifico> [haikuports/haikuports] Begasus pushed 1 commit to master [+0/-0/±2] https://github.com/haikuports/haikuports/compare/b771ded643fb...7694f68eca5b
<botifico> [haikuports/haikuports] OscarL 7694f68 - python3.10: patch set re-work. (#10750)
<OscarL> the discussion there regarding locks... "completely unrelated" as jmairboeck put it. And the mention of "testing (meson test) on nightlies"... no idea)
<Begasus> let's take it from here and see where it leads us, thanks!
<OscarL> welp, that's one way to do it, heh. If they don't like it... should be more explicit next time :-D
<Begasus> for some I also take some time for other to review the PR, I even ask for it, when nothing pops up I presume no one objects :P
<OscarL> Regarding "python_pkgconfig"... it is still how I think it should be. korli seems to disagree. So... stale mate :-)
<botifico> [haikuports/haikuports] Begasus pushed 1 commit to master [+0/-0/±1] https://github.com/haikuports/haikuports/compare/7694f68eca5b...3017c1fb8671
<botifico> [haikuports/haikuports] OscarL 3017c1f - python_pkgconfig: recipe cleanup/fixes. (#10744)
<Begasus> the ECM recipe is kinda in the same situation, provides only data and could be "any"
<Begasus> but it needs Qt for the build ...
<OscarL> I usually conceede to korli, but... listing indirect dependencies is something I really do not like.
<Begasus> don't use the kde frameworks then :P
<OscarL> (borgbackup doesn't directly calls cmd:pkg-config. python-pkgconfig does. pretty clear cut case to me, even if the later .recipe might needs improvements)
<OscarL> Begasus: noted! :-D
<OscarL> Aiming for zero open PRs soon Begasus?
<OscarL> would be cool, same with chopping down some good % of the open issues :-)
<Begasus> I guess it's clean enough atm OscarL in regards or the PR's
<Begasus> means we're good on "most" things? :P
<Begasus> for the issues, I've been going over them quiet some times already, I still have to pull myself from not pushing "close" on some of them :D
<OscarL> heh
<OscarL> 10717 should be closable. mandoc one... I started working on it at least (but you know how I "rotate" among a few things)
<Begasus> heh
<OscarL> $settingsDir/bashrc one... should be easy enough to fix (if I weren't so lazy).
<OscarL> the one about bash_completions... I at least have some idea about how to fix that one too.
<Begasus> k, let's see if x512[m] could confirm
illwieckz has quit [Ping timeout: 480 seconds]
<Begasus> x512[m], if you have the time, could you check https://github.com/haikuports/haikuports/issues/10717 ?
<OscarL> 8806 (the one about "non _x86" package names)... I'm tempted to close it, given that only "pandas" remains on the list, and that is a no go for now.
<Begasus> go for it :)
<OscarL> ok.
freddietilley has joined #haiku
<OscarL> Begasus: some days ago I tried installing pip for Python 3.13 without any pkgman intervention... went surprisingly smoothly.
<OscarL> Maybe we should drop the pip recipe after all, and avoid having to keep it updated for each new Python version.
<OscarL> IIRC, I just used the "python3.13 -m ensurepip --altinstall" (altinstall only installs commands with version suffix, ie: pip3.13 instead of pip/pip3))
<OscarL> "--user" can also be used, which is great.
dby has quit [Quit: Vision[]: i've been blurred!]
<OscarL> Beats having to add even more dependencies for newer pip too :-)
freddietilley has quit [Quit: WeeChat 4.2.2]
freddietilley has joined #haiku
<nipos> Good morning. I retried booting Haiku using a Nightly Version now. The error is still the same as with Beta4, but the Bootloader gives some serial output now. Here's what I got: https://paste.bka.li/Auwhtd Looks like something is cut off at the beginning, unfortunately.
<Begasus> maybe nuke the other lua stuff too then OscarL :)
<Begasus> g'morning nipos
<nipos> Now I also tried a 32bit Image of the same Nightly version and that one boots almost completely and crashes at the end with a totally different error: https://paste.bka.li/qfWm8a
<OscarL> nipos, FWIW, serial log doesn't seems cut off to me (I have serial output from my VM... first line is: "options=0")
<nipos> Ah, that's fine then :) I expected something in front of that options=0
<nipos> Limiting memory to 4GB in the boot loader makes 32bit Haiku also boot fine, just like the 64bit version (where too much memory causes a totally different error message). But that can't be the actual solution? Attempting to compile huge applications on 24 cores but with only 4GB memory is going to be fun xD
DKnoto has quit [Quit: Leaving]
<Begasus> heh
<Begasus> I can imagine nipos :)
<Begasus> nice setup though! :D
<nipos> I hoped I could avoid messing around with the hardware, but could physically removing one or two of the memory chips help?
<OscarL> only "fail to boot with lots of RAM" bug I remember is: https://dev.haiku-os.org/ticket/12896, but it showed different sympthoms.
<nipos> Begasus: It's my build server that I normally use to build OpenIndiana packages, but I wanted to experiment with big Haiku stuff where my desktop computer isn't sufficient
<nipos> Fujitsu Primergy RX300 S7
<Begasus> could only dream of it here nipos
<Begasus> pretty pleased with this one though :)
<OscarL> I still dream of getting a Ryzen 5 5600GT some day.
<Begasus> we could start crowd funding OscarL, but knowing postal service up there ... :D
<OscarL> heh :-)
<nipos> The machine is really powerful, builds LibreWolf/Firefox in less than a hour if I remember right :D
<Begasus> wOOt!
<Begasus> plasmatube playing 15min in a row, no issues so far :D
* OscarL wonders how a "beowolf" cluster of 3 Atom netbooks + a Phenom II X4 would perform (most likely worse than just the Phenom :-D)
<nipos> Hardware support on Haiku is quite okay, besides the memory. Another issue is that it doesn't detect the RAID controller, but USB storage should be fine as long as I don't use it too much. Only this 4GB memory limit drives me crazy
<OscarL> understandably so.
<nipos> Hit another bug at shutdown
owad has quit [Quit: No Ping reply in 180 seconds.]
owad has joined #haiku
vdamewood has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<Begasus> running with 16GB fine here
<Begasus> 64bit* (R1B4)
<nipos> Also had it running fine on 16GB a while back. This thing has 132GB, however
<OscarL> I think you should open a ticket for that KDL, nipos (and also for the RAM issue, given it still happens with nightlies)
<nipos> Removing one 16GB RAM doesn't help
<nipos> I'll open tickets later, but I want to get something to work first :)
tuaris has quit [Remote host closed the connection]
<OscarL> "bisecting" 4-132 GB of RAM sounds "fun" :-P
<nipos> Yeah, and that machine takes ages to initialize the BIOS and RAID Controller before even attempting to boot Haiku
<nipos> Still no look with 32GB removed
<Begasus> fresh coffee :D
<OscarL> I'd try first the smallest >4GB size I could, just in case (if that fails... saves lots of tries).
<nipos> That would be 32GB, because one RAM for each CPU is required... Let's try that...
HaikuUser has joined #haiku
<HaikuUser> Hello! With the latest updates, mednafen works slow like hell! Do you know why this happens?
<HaikuUser> By slow I mean the emulator has something like 1fps. Until the latest haiku updates, it worked perfectly.
<nipos> 32GB works. Maybe I can get at least 64GB to work, that should be sufficient for my work
<OscarL> HaikuUser: if the only difference is with Haiku's hrev... you should report the issue over https://dev.haiku-os.org/ (ideally, detailing the last version that worked OK for you, and the first where you noticed the slow downs)
<Begasus> right, not seeing anything related in the recipe
<OscarL> HaikuUser: if you haven't cleared the /system/packages/administrative/ directory, you might just be able to boot an older Haiku version to test things (just select an older "state" from the boot menu.
<OscarL> HaikuUser: just in case, more info about boot loader menu here: https://www.haiku-os.org/docs/userguide/en/bootloader.html
<HaikuUser> ok, Ill try what you suggest.
<HaikuUser> thx
<OscarL> no problem! :-)
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
<Begasus> stress testing here, several apps open and building "dialog" in 32bit (qemu) :D
HaikuUser has joined #haiku
<HaikuUser> ok, so I tried to keep SHIFT at the booting process and I had no boot options. The os started normally
DKnoto has joined #haiku
<OscarL> HaikuUser: if your PC uses UEFI boot, the SHIFT method won't work. You need to "spam" the spacebar key for UEFI.
<OscarL> (hit SPACE repeatedly between "power on" and before the Haiku boot icons show up)
<HaikuUser> ok, I'll try with the spacebar
HaikuUser has quit []
<OscarL> guide is pretty clear about that but... oh well :-)
<diver1> nipos: the crash with 32bit Image seems to be related to a graphics driver, it should boot if you force vesa
diver1 is now known as diver
<nipos> I did this.
<nipos> Without forcing VESA, it resized the iRMC window before crashing a few times (trying different resolutions, probably), but it showed the same error in serial output
<nipos> Also, with limiting memory to 4GB, 32bit booted just fine without forcing VESA
<OscarL> I see "vesa: acc: vesa.accelerant" on https://paste.bka.li/qfWm8a, and the other KDL is around intel_cstate, not intel_extreme.
<nipos> It has a Matrox graphics card, so there should be nothing related to intel_extreme
<Begasus> k, dialog still buils ok with gcc2 :)
<OscarL> nipos: indeed (I just wasn't sure to which of the two crashes diver referred to, and as the first is indeed using VESA already, I thought he might have just misread the intel_cstates on the second)
<nipos> 64GB also worked fine here, btw. I'm currently trying with 80GB (5x16GB) and I guess it worked, since the fans slowed down, but now the Wifi suddenly fails to connect so I can't access the iRMC :/
<OscarL> BTW, hello theere diver, and thanks again for all your Haiku-related work over the years!
<Begasus> yeah! one of the founders of the KDE frameworks! :D
<OscarL> nipos: so... 64 GB Ok, 80 GB and interrupts handlings gets borked and you start losing devices? (thinking about that NMI panic).
<OscarL> (I'm of course, just talking out of my rear end... specially at 5:30 AM :-P)
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
<nipos> No, 80GB seems to also work, but my Wifi dongle is now going crazy, so that I can't access it anymore. That doesn't have anything to do with Haiku, since I can normally access the iRMC without any OS running
<nipos> 96GB I already tried, there starts the kernel panic
HaikuUser has joined #haiku
<nipos> The NMI panic isn't related to RAM at all. That even happens with the limit to 4GB active. It always happens, but only on this device
<HaikuUser> Ok, so in hrev57609 mednafen works great, but in hrev57898 is slow (1fps). What can I do next?
<HaikuUser> can I go back with the os package state by default to the hrev57609? How can I set this version as default?
erysdren has joined #haiku
<OscarL> HaikuUser: https://www.haiku-os.org/guides/daily-tasks/updating-system.html (see the "downgrading" section).
<OscarL> HaikuUser: Also, would be ideal, if you could open a ticket on https://dev.haiku-os.org/ detailing what you have found (attaching /var/log/syslog for both "fast" and "slow" versions might help).
<HaikuUser> I want to do that but I'm not sure I can
<HaikuUser> I'm not so technical
<nipos> I have to go to work now, will do further tests and write the tickets at the evening
<OscarL> be well nipos!
<HaikuUser> thank you for your help!
<OscarL> HaikuUser: no problem.
<Begasus> cu nipos
<diver> OscarL: Begasus hi there :)
<Begasus> he's alive!!! :)
<Begasus> so to add to OscarL's comment, thanks for all the groundwork done over all the years! (and being a great mentor over that period) !
<Begasus> still can't code, but learned a lot :D
<OscarL> I used to smack coworkers on the head, point to diver's bug reports and tell them... "see? why can't you right tickets like that?"
<diver> np :) btw Begasus could you take a look why suitesparse package is so large? Perhaps it needs to be stripped
HaikuUser has quit [Ping timeout: 480 seconds]
<OscarL> s/right/write/ (getting too late here :-P)
<diver> OscarL: hah :)
<Begasus> heh
<Begasus> I'll have a look diver
<diver> thanks)
<OscarL> similar with commit messages... used Haiku's one as examples... "see? I'm not making things up... 'changed things' is NOT how proper devs do it!"
<Begasus> lol
<Begasus> OscarL, not committing that suggestion you made for "dialog" :P
<Begasus> could pull out x86 for secondary though
<phschafft> morning.
<Begasus> err nope :P
<OscarL> Begasus: nothing seems to use lib:libdialog anyway (not on-tree at least)
<Begasus> Moin phschafft
<phschafft> :)
<OscarL> and for the cmd:dialog... only one should be more than enough. Just sayin'! :-P
erysdren has quit [Ping timeout: 480 seconds]
<Begasus> OscarL, there is only "one" :P
<Begasus> diver, we're way of on releases for suiteparse it seems :)
<OscarL> misread the cmd:dialog_config$secondaryArchSuffix one.
illwieckz has joined #haiku
<Begasus> heh
<Begasus> diver, package is already stripped with defineDebugInfoPackage?
<Begasus> maybe moving the documentations will reduce the size
<Begasus> ah bah ... sci-libs :P
<Begasus> I blame most of them on miqlas :P
<OscarL> Alright... time for me to emulate those little cats on phschafft's picture.
<OscarL> See you around, nice fellows. Have a great one!
<Begasus> cu OscarL
OscarL has quit [Quit: zzzZZZzzz]
<Begasus> Alt-F12 doesn't work :P
* phschafft wonders if OscarL has a tree for that.
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
<Begasus> diver, 2.40MiB chopped of moving docs to _devel package, don't think more can be done to reduce it
<botifico> [haikuports/haikuports] humdingerb pushed 1 commit to master [+1/-1/±0] https://github.com/haikuports/haikuports/compare/3017c1fb8671...edaa8281c0cb
<botifico> [haikuports/haikuports] humdingerb edaa828 - yt-dlp: update to 2024.07.25
<Begasus> devel:libtbb >= 2018.5
<Begasus> tss ... should build that with latest one here :P
<BlueSky76> Not Haiku specific, but not totally OT either ;-) I encountered the problem of generating random point positions within a circle while writing a game. Found this great video on the topic. Thought it might be of interest to at least some of you. Here's the link: https://www.youtube.com/watch?v=4y_nmpv-9lI
<x512[m]> Why video, not article?
<BlueSky76> x512: Because that's what I found first and worked for me ;-) I'm sure there are many written articles about the topic to be found on the internet.
<BlueSky76> In general, I too prefer written articles to videos. But in this case I was eager to find a solution and used the first resource I found which was this video. And it worked for me ;-)
dalme_ has joined #haiku
EnriqueMG[m] has joined #haiku
<EnriqueMG[m]> I guess a random point based on polar coordinates has a bad distribution
JakeSays1 has joined #haiku
freddietilley has quit [Ping timeout: 480 seconds]
freddietilley has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
JakeSays has quit [Ping timeout: 480 seconds]
<phschafft> fun. yt directly suggests the numberphile video on the same topic.
<phschafft> also note that using some of those ways to map random numbers into circles is actually used to test the generators.
gouchi has joined #haiku
<Begasus> k, this requirement is wrong for glib2: gettext$secondaryArchSuffix
<phschafft> what I feel is missing on the video is that is missing in most of them: they often look at the speed. but they don't look into the stability.
<phschafft> for example how much entropy does the random function actually return. how much does sqrt(), sin(), cos() or any of the other functions remove, ...
<phschafft> also those functions add noise to the problem, which might be amplified in later stages.
<Begasus> so many unanswered questions ... tsss :D
<Begasus> like glib2 already requires libintl (expected), but why is the base package gettext required (doesn't make sense, and not required/no policy warning when not included)
<Begasus> maybe diver knows ;) (can't go back far enough in history to found anything usefull :)
<Begasus> found/find*
zard has joined #haiku
<Begasus> with "fixed" glib2 package: https://bpa.st/FHKA
<Begasus> a LOT better ;)
zard has quit []
catcream has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
AlienSoldier has joined #haiku
catcream has joined #haiku
zard has joined #haiku
<Begasus> And another fix landed in KDE :) fix for plasmatube :D https://0x0.st/XONr.png
<dovsienko> rpcap-protocol.h:145:9: error: unknown type name 'uint8'
<dovsienko> did the C headers reshuffle recently?
tuaris has joined #haiku
<zard> Well, gcc got updated around 3 weeks ago.
<zard> I don't know it that affects you, but it did expose a bug in HaikuWebKit...
* zard is working on fixing that bug :)
dqk_ has joined #haiku
AlienSoldier has quit [Quit: Vision[]: i've been blurred!]
dqk has quit [Ping timeout: 480 seconds]
DKnoto has quit [Ping timeout: 480 seconds]
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
DKnoto has joined #haiku
bbjimmy has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
erysdren has joined #haiku
nosycat has joined #haiku
DKnoto_W has joined #haiku
HaikuUser has joined #haiku
linuxmaster has quit [Read error: Connection reset by peer]
HaikuUser has quit [Remote host closed the connection]
linuxmaster has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit [Remote host closed the connection]
HaikuUser has joined #haiku
DKnoto has quit [Ping timeout: 480 seconds]
illwieckz has quit [Quit: I'll be back!]
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
bbjimmy has joined #haiku
mmu_man has quit [Read error: No route to host]
dby has joined #haiku
freddietilley has quit [Quit: WeeChat 4.2.2]
TMM has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
illwieckz has joined #haiku
TMM has joined #haiku
B2IA has quit [Read error: Connection reset by peer]
B2IA has joined #haiku
DKnoto_W2 has joined #haiku
erysdren has quit [Quit: Konversation terminated!]
linuxmaster has quit [Ping timeout: 480 seconds]
linuxmaster has joined #haiku
DKnoto_W has quit [Ping timeout: 480 seconds]
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
<waddlesplash> nipos: NMI panic should be continuable
<waddlesplash> uint8 & etc. are defined in SupportDefs include chain
<waddlesplash> nipos: thanks for that serial log. I may be able to try and debug this just using the memory address ranges specified
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
xet7 has quit [Quit: Leaving]
<nekobot> [haiku/haiku] waddlesplash pushed 1 commit to master [hrev57899] - https://git.haiku-os.org/haiku/log/?qt=range&q=ffd9482847d8+%5E587c5ecdac2d
<nekobot> [haiku/haiku] ffd9482847d8 - FAT: Extend volume size limit
juanjo has joined #haiku
xet7 has joined #haiku
<waddlesplash> dunno who's been reading the commit log, but: today's nightly build has IP/UDP/TCP receive checksum offload support (if you are using a driver from FreeBSD that also supports it)
<Begasus> another one died for Haiku, first kdevelop, now rkward also (need qt6webengine) :/
<x512[m]> Checksum offload support is a very little problem compared to small TCP window problem.
<x512[m]> HaikuPorts package download is very slow because of this.
<waddlesplash> yes, we do need to get around to that
<waddlesplash> we should probably only do it in the case where SACK is supported as I commented though
<x512[m]> Maybe other Haiku users are located near HaikuPorts servers so they are not so suffer with slow download.
<x512[m]> For me it is ~300 KB/s.
<HaikuUser> hey, could you tell me why my input filter for LockWorkstation is not working? https://github.com/cafeina-software/LockWorkstation/tree/master/src/input_filter as soon as it is fixed, I could make a PR to HaikuArchives
<waddlesplash> why do you need an input filter?
<HaikuUser> bc it has to prevent any non authorized user to call team monitor and kill the screen locker app
<HaikuUser> if anyone is able to do that, the app does not make sense at all
<HaikuUser> I don't know. I tried to resurrect the application, but it seems nobody cares...
<waddlesplash> ah, I see
<waddlesplash> tbh, we should just add support for this directly in app_server
<waddlesplash> and input_server. but your change does make sense for now
<waddlesplash> what's not working specifically?
<HaikuUser> it does not filter correctly all events of Cmd Alt Del, sometimes filters it once, but after that does not filter all attempts, if at all
<waddlesplash> hmm
<HaikuUser> or Ctrl Alt Del, depending on the keyboard
<waddlesplash> I think Ctrl+Alt+Del may have special handling in input_server
<waddlesplash> but the inconsistent behavior seems strange, maybe a bug
<HaikuUser> The app is being updated to be able to login using a real account (in addition to a fake app account), and it allows to set it a rotating background image to make it look good
<HaikuUser> of course the system account has to be created in order to be used
<waddlesplash> hmm
<waddlesplash> how does screen_blanker's password window handle this?
<waddlesplash> I just tested, Ctrl+Alt+Del does nothing
<dovsienko> waddlesplash: yes, that seems to be the one, I tried to search in the commit log, but could not
<HaikuUser> screen_blanker or somewhere else seems to handle it correctly
<HaikuUser> but I couldn't replicate it
<waddlesplash> hmm
<HaikuUser> there is a screen_saver input filter also
<waddlesplash> it doesn't seem to use a filter
<waddlesplash> yes, but it doesn't look like it handles keyboard events
<waddlesplash> Ctrl+Alt+Del is handled directly inthe input server keyboard device
<waddlesplash> so, it can't be filtered
<diver> x512[m]: somehow package download speed is pretty decent over here, about 7-10 Mb/s, it is much slower in vbox tho
<x512[m]> diver: Ping?
<waddlesplash> HaikuUser: so, actually, TeamMonitor does show up, it's just underneath the screensaver windows
<waddlesplash> so how does the screensaver accomplish that?
<x512[m]> The problem is current Haiku TCP implementation is slow if ping time is big. It can't saturate bandwidth.
<HaikuUser> perhaps it uses some window flags to get over everything
<HaikuUser> if Team Monitor is showing up, then, could any canny user blindly and guessing, use the keyboard to kill processes until it kills the screen locker app?
<diver> x512[m]: ping s3.wasabisys.com
<diver> 64 bytes from 38.27.106.12: icmp_seq=0 ttl=56 time=116.771 ms
<diver> PING wasabisys.com (38.27.106.12): 56 data bytes
<HaikuUser> under the risk of crashing the system, ofc
<waddlesplash> HaikuUser: it shouldn't
<waddlesplash> diver: are you in VMware?
<waddlesplash> (and using NAT?)
<waddlesplash> (or a virtual network inside VMware)
<diver> VMware, but this ping output is from the host.
<diver> lemme check from Haiku
<waddlesplash> I think VMware proxies TCP
<waddlesplash> so that's why your speeds are so good, it's not really Haiku's TCP
<HaikuUser> it would be bad if r5 is relased with a buggy tcp
<diver> it is using NAT
<HaikuUser> *b5
<waddlesplash> HaikuUser: the TCP isn't too buggy anymore, it just doesn't implement window scaling fully
<waddlesplash> so we just wind up with window full -> wait for ACK -> window full -> wait for ACK
<waddlesplash> which can be slow if ping time is slow
<waddlesplash> I need to look at a TCP bug today, maybe I'll look at this too
<diver> about same ping latency from inside Haiku
<diver> ping s3.wasabisys.com
<diver> 64 bytes from 38.27.106.106: icmp_seq=0 ttl=128 time=124.759 ms
<diver> PING wasabisys.com (38.27.106.106): 56 data bytes
<diver> 64 bytes from 38.27.106.106: icmp_seq=1 ttl=128 time=118.986 ms
<waddlesplash> Haiku's TCP only implements the send side of SACK not the receive side
<waddlesplash> I don't remember the initialization sequence, if we know the remote SACK support before we determine window size/scale, then that would be good
tijon has joined #haiku
<tijon> I have no idea how good of a place this is to ask but i'll give this a shotdoes anyone know what would be the best and intended way to load a HVIF RDEF format off a Resource file as a BPicture? (intended for BPictureButton)
<waddlesplash> why do you need a BPictureButton specifically?
<waddlesplash> BButton accepts BBitmaps
<waddlesplash> most of the things in the GUI that have icons on buttons use BButton
<tijon> I thought it would be a good way to load a HVIF RDEF
<x512[m]> diver: Can you try to use bridge networking mode (no NAT)?
<diver> the following speed is in bridge mode
<diver> ~> pkgman ins libreoffice
<diver> 2% libreoffice-24.2.3.1-3-x86_64.hpkg 6.05 MiB/250.88 MiB 1.40 MiB/s
<diver> ~> the following speed is in NAT mode
<diver> ~> pkgman ins libreoffice
<diver> 6% libreoffice-24.2.3.1-3-x86_64.hpkg 13.74 MiB/244.13 MiB 4.00 MiB/s
<waddlesplash> tijon: the best way to load a HVIF is just to rasterize it to a BBitmao
<waddlesplash> BBitmap
<tijon> I didn't know that BButton had actual implemented bitmaps, thank you very much waddlesplash!
<waddlesplash> :)
<waddlesplash> read the headers (or the docs)! :)
<x512[m]> > the best way to load a HVIF is just to rasterize it to a BBitmao
<x512[m]> That is sad.
<x512[m]> Vector is lost.
<waddlesplash> well most of the time you are not going to need the vector again
<waddlesplash> we should introduce a BIcon class though
<diver> so indeed NAT is way faster over here
<waddlesplash> so that we can pass it around and handle HiDPI magically
<x512[m]> Also it makes remote desktop slower.
<waddlesplash> remote desktop really needs a major overhaul, it's just not designed for anything but LAN
<waddlesplash> anything that draws to BBitmap will be unbearably slow because we don't multiplex and we don't compress them
<x512[m]> It it better to draw HVIF on app_server side.
<x512[m]> Ticket about remote desktop BBitmap's already exist: https://dev.haiku-os.org/ticket/15901
<waddlesplash> does the remote protocol compare bitmaps locally and avoid sending them if not necessary?
tqh has joined #haiku
<gordonjcp> well
<gordonjcp> Vienna is very very hot today
<diver> waddlesplash: ramfs still crashes here :(
<waddlesplash> I saw your message, yes
<waddlesplash> at least it's an assert failure now
<x512[m]> It may be an idea to render bitmaps on client side of remote desktop, but it will be problematic to handle disconnections.
<waddlesplash> shouldn't be
<x512[m]> Some mechanism will be needed to report that BBitmap content is lost.
<waddlesplash> keep track of all bitmaps attached to the remote and delete them on disconnect
<x512[m]> Applications should continue running even after disconnect. We are not X11.
<waddlesplash> oh
<waddlesplash> yes, sure
<x512[m]> It is currently working fine.
<waddlesplash> x512[m]: ultimately we will probably need to render bitmaps on both neds
<waddlesplash> ends
<waddlesplash> both server and client
<waddlesplash> that way, reconnects work fine, and also the server can look at the bitmap content quickly
<x512[m]> Yes, that can work.
<waddlesplash> this will also make it possible to remote-in to a running session
<waddlesplash> i.e. start on a real display, then connect via some remote, and so on
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
yann64 has joined #haiku
nosycat has quit [Quit: Leaving]
xet7 has quit [Remote host closed the connection]
xet7 has joined #haiku
tijon has quit [Quit: Vision[]: i've been blurred!]
orealis has quit [Read error: Connection reset by peer]
Anarchos has joined #haiku
orealis has joined #haiku
HaikuUser has joined #haiku
HaikuUser2 has joined #haiku
HaikuUser2 has quit []
HaikuUser2 has joined #haiku
<HaikuUser2> hi
HaikuUser2 is now known as MegaTrond
<zard> o/
<MegaTrond> why there is too many python versions and packages on haikudepot?
<MegaTrond> it is confusing... why not have the latest version and packages installed using PIP?
<zard> I think we have multiple Python versions because some packages might depend on different versions
<zard> And then if a package depends on a Python library, then I think it needs to be put on haikudepot as well
<MegaTrond> I see, thanks for the info...
MegaTrond has quit []
HaikuUser2 has joined #haiku
HaikuUser2 is now known as MegaTron
<MegaTron> hi, installing NumPy using pip3.12 gives me an error: FAILED: numpy/_core/_multiarray_umath.cpython-312.so.p/src_umath_string_ufuncs.cpp.o c++ -Inumpy/_core/_multiarray_umath.cpython-312.so.p -Inumpy/_core -I../numpy/_core -Inumpy/_core/include -I../numpy/_core/include -I../numpy/_core/src/common -I../numpy/_core/src/multiarray
<zard> Ah, I think that's one reason that we would put Python libraries on HaikuDepot:
<zard> They sometimes require extra work to get working
<zard> I suspect numpy is difficult to port or it would probably have been done already
<MegaTron> zard, it is necessary to port them each version is available? that's a hard work...
<zard> Eh, just update whatever patches need to be updated. Hopefully nothing needs to be changed
* zard wonders if patchsets are shared among Python versions or if there is a lot of code duplication
<MegaTron> I think it is the compiler, it would not have the latest version of C++
<MegaTron> it was compiling... and give me that error
<zard> What version of gcc do you have?
<MegaTron> 13.3.0
<MegaTron> 10-08-2023... that's OLD :)
<zard> I suspect that's new enough, though
<zard> I think its newer than the gcc that comes with some Linux distros :D
dalme_ has quit [Read error: Connection reset by peer]
<zard> Did it give you a detailed error?
<MegaTron> ninja: build stopped: subcommand failed.
<MegaTron> error: metadata-generation-failed
<MegaTron> ../numpy/_core/src/umath/string_buffer.h:501:5: error: expected primary-expression before ')' token
<zard> Tip: use a paste-bin for long error messages (>3) lines, e.g. https://bpa.st
<zard> Ah, but looks like you got the heart of it there
* zard tries to see what's on line 501
<zard> Not seeing anything obvious (maybe I don't have the right source code for your version)
<zard> Anyway, I certainly won't have the time to fix it
<nipos> waddlesplash: Thanks for looking into my issues :) I can now confirm that with 80GB RAM the system boots fine (5x16GB), and if I insert a sixth RAM chip (96GB) the error I sent starts to appear. No difference between 96GB, 112GB and 128GB
<MegaTron> I think NumPy,Pandas,Matplotlib are the most used packages on python...
<Begasus> MegaTron, numpy is borked
<MegaTron> zard, thanks for your help
<Begasus> pandas isn't in a state to build newer versions
<Begasus> matplotlib should still be good (I think) :)
<nipos> 80GB should be enough for now, but if I can assist finding the issue, I'd be happy to do so. Did you get all three different errors? And should I still open tickets, if you're already aware of them?
<Begasus> OscarL is our man there :)
<MegaTron> Thanks Beagsus
<MegaTron> Thanks Begasus
zard has quit [Quit: leaving]
MegaTron has quit [Quit: Vision[]: i've been blurred!]
<Begasus> been able to install spyder the other day with pip, so that sh....
<Begasus> tl :P
HaikuUser has quit [Ping timeout: 480 seconds]
<waddlesplash> nipos: there are already tickets about this sort of error I think
<waddlesplash> however, I don't know that the users who opened them are still active
<Begasus> zardshard, problem with "pip install" is that it extract and uses tmp directories, so it's hard to patch anything in that process :)
<Begasus> waiting for build package gettext_libintl_devel-0.22.5-1 to be deactivated
<Begasus> split it up a bit further ... big wip still :)
<nipos> As a last question regarding this device... I guess there's no magic trick to somehow get the RAID controller to be detected?
<nipos> I already tried switching between IDE and SCSI mode, in both cases it doesn't show up. I have three unused hard disks that would be preferrable over the USB stick. But the USB stick should do as well if there's no other solution
dalme_ has joined #haiku
yann64 has left #haiku [#haiku]
Nephele has joined #haiku
DKnoto_W2 has quit [Ping timeout: 480 seconds]
DKnoto_W2 has joined #haiku
<Nephele> hi :)
* Nephele is back from renovating, on a new standing desk
<Begasus> wb nephele
* Begasus is closing down for today
<Begasus> cu peeps!
Begasus has quit [Quit: Vision[]: i've been blurred!]
clee_ has joined #haiku
Anarchos has quit [Ping timeout: 480 seconds]
tijon has joined #haiku
<tijon> I am a bit lost on the technicalities of resources in Haiku. Is compiling rdef into rsrc using rc and using resattr to embed compiled inside binary the correct way?
<tijon> Trying out this method seems to result in nothing on xres of the binary for me
clee has quit [Ping timeout: 480 seconds]
<Nephele> It sounds correct to me, but I've not tried this myself to verify this
<tijon> I have seem to found the cause of the issue: resattr
<tijon> highly likely that it is user error on my side
<tijon> although trying out different options in resattr brings me to the same result everytime: nothing being embedded, or at least visible by binary and xres. Can someone who has used this mysterious thing confirm if it is user error or a bug?
<tijon> Nevrmind, I got fooled by the internet once again. I tried out the write mode in xres and it worked. Sorry for the spam
tijon has quit [Quit: Vision[]: i've been blurred!]
<Nephele> :)
<dovsienko> in case anybody is interested, VirtualBox mouse integration does not work in Haiku with VirtualBox 7.0.20 because the virtual USB times out and reconnected in a loop. downgrading to 7.0.18 works around that
<dovsienko> could be entirely a VirtualBox bug
tijon has joined #haiku
* phschafft waves.
* phschafft also waves specifically to nephele.
axeld has quit [Ping timeout: 480 seconds]
tqh has quit [Quit: Leaving]
gouchi has quit [Remote host closed the connection]
<nekobot> [haiku/haiku] waddlesplash pushed 3 commits to master [hrev57900] - https://git.haiku-os.org/haiku/log/?qt=range&q=9888c3ce8386+%5Effd9482847d8
<nekobot> [haiku/haiku] e86e0ee4fd0f - TCP: Fix operator precedence in SACK in _PrepareSendSegment().
<nekobot> [haiku/haiku] 04468d614b2a - TCP: Don't send extra duplicate ACKs with SACK enabled.
<nekobot> [haiku/haiku] 9888c3ce8386 - TCP: Skip updating round-trip time if it's <= 0.
<Nephele> hi phschafft
<Nephele> dovsienko: if you want you can report a ticket on the bugtracker, but in general i don't think many haiku devs use virtualbox
<phschafft> just showed stephanj some terminal abstraction.
<Nephele> oh?
juanjo has left #haiku [Disconnected: Hibernating too long]
<B2IA> (AGMS) Thanks for that VirtualBox tip, dovsienko. Could also use PS2 mouse instead of USB, though that's painful.
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
DKnoto has joined #haiku
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
tijon has quit [Quit: Vision[]: i've been blurred!]
Nephele has quit [Quit: Vision[]: i've been blurred!]
DKnoto_W2 has quit [Ping timeout: 480 seconds]
<dovsienko> USB is for the virtual tablet, which copies the pointer position from the host screen, which is very convenient
<dovsienko> just realised I am not sure the error was for the tablet, should have compared the output of "listusb" between the versions. it spams the serial console anyway.
Chai-T-Rex has quit [Remote host closed the connection]
ChaiTRex has joined #haiku
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
AlienSoldier has joined #haiku
dalme_ has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
<AlienSoldier> facebook messenger can detect that our internet connection is cut and it resume when we reconnect to the network (i do that by disabling DHCP and reactivating it). Would it be possible to have an OS level notification that internet connection is cut (even if no cable is removed) and a setting that allow for an automatic DHCP reset if that criteria is met?
<AlienSoldier> i don
<AlienSoldier> ti don't have this problem with my cable ISP but i do a lot with my satellite ISP.
AlaskanEmily has joined #haiku
<dovsienko> a shell script and a cron job could do this, why would it be necessary to have an OS-level notification and who would need to receive it?