<scanty>
i'm a little confused about the TV app, however... do we have bt848 drivers?
<scanty>
drivers*
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
AlaskanEmily has joined #haiku
HaikuUser has joined #haiku
jsto has left #haiku [#haiku]
scott_puopolo[m] has quit [Quit: Client limit exceeded: 20000]
HaikuUser2 has joined #haiku
HaikuUser2 is now known as gregatron5
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
frkzoid has quit [Read error: Connection reset by peer]
frkazoid333 has joined #haiku
x10z has joined #haiku
HaikuUser has joined #haiku
HaikuUser is now known as psychonate
psychonate is now known as nmbosley
gregatron5 has quit [Quit: He didn't quit. He was fired!]
walkingdisaster has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
Maturi0n has joined #haiku
Andrey has joined #haiku
nhtello has quit [Ping timeout: 480 seconds]
Maturi0n_ has quit [Ping timeout: 480 seconds]
Andrey has left #haiku [#haiku]
Andrey has joined #haiku
tuaris has quit [Read error: Connection reset by peer]
bjorkint0sh has quit [Ping timeout: 480 seconds]
Andrey has left #haiku [#haiku]
walkingdisaster has quit [Quit: Vision[]: i've been blurred!]
vdamewood has joined #haiku
nmbosley has quit [Quit: Vision[]: i've been blurred!]
OscarL-x64 has joined #haiku
<OscarL-x64>
a bit late but.... scanty: the moved TV app was used for DVB-T (while it was MPEG2). AFAIK, it only supported cards with the cx32882 chipset.
<OscarL-x64>
It never had analog TV support (nor we had open source bt8x8 drivers)
<OscarL-x64>
I had a really crappy driver for SAA713x based TV/FM cards, but only got it so far as to make the FM tuner work, and output sound via its analog rear jack :-D
<OscarL-x64>
(thus, why in a silly attempt at word play... I called my driver: "I don't C TV". Still available on BeBits archives!)
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<OscarL-x64>
So, we have a text/x-patch filetype (even with diff, .patch, and .patchset as associated extensions). Yet... 99.999% of .patchsets get identified as text/x-source-code
* OscarL-x64
wishes he knew how to fix the sniffer rules.
<erysdren>
i know that the xml mime specification files can have associated weights for each marker of what a file is
<erysdren>
maybe the extensions in the x-patch mime file don't have a high enough weight?
<OscarL-x64>
I *think* they actually don't play any role for files in BFS (possibly other FSes with enouth attribute support).
<OscarL-x64>
IIRC, there's a ticket somewhere to make better use of the "extensions data" if all else fails.
* OscarL-x64
puts on his best Janitor clothing, and starts updating Python 3.11/3.12 to the latest bugfix releases (missed those by a few days)
Begasus has joined #haiku
<Begasus>
g'morning peeps
<OscarL-x64>
Aloha Begasus :-)
<Begasus>
Hola OscarL-x64 :)
<bbjimmy>
I can't update from hrev57579 to hrev57581 on x86_gcc2 Haiku server not found
<bbjimmy>
x86_64 works fine.
<Begasus>
k, updating here, biab (before I can't react due to the noto update) :)
Begasus has quit [Quit: Vision[]: i've been blurred!]
<augiedoggie>
OscarL-x64: increasing the range on the sniffer rules for patches will help a lot i.e. change the 0:100 to 0:500 or so
<augiedoggie>
for the +++ --- @@- rules
<OscarL-x64>
Coolio, thanks augiedoggie! Will have to give that a try.
Begasus has joined #haiku
<Begasus>
re
<Begasus>
might as well delete my noto branches now :)
<OscarL-x64>
AboutSystem gets noticeably wider after the font change :-)
<OscarL-x64>
what was the name of that website you used to past images Begasus?
<OscarL-x64>
That screenshot I linked shows... hrev57578's AboutBox with "Noto Sans Display" at the top. and with "Noto Sans" at the bottom. Middle one is current hrev57581 with "Noto Sans Display".
<OscarL-x64>
(Using "Noto Sans" in latest hrev shows it the same width as in hrev57578)
<OscarL-x64>
Begasus: running on bare metal today, to try to update Python 3.11/3.12, without builds taking me hours :-D
<Begasus>
upgrade is nice :D
<OscarL-x64>
as long as I finish before the ambient temps get too high... this thing gets HOT without frequency scalling :-D
illwieckz has quit [Quit: I'll be back!]
<Begasus>
k, all caught up on reading :)
<Begasus>
nice report/progress again!
<Begasus>
was thinking it missed the ICU transfer, but that's for this month :D
<Begasus>
biab
<OscarL-x64>
StreamRadio almost blew my hears after chaning station... now it blew it self (crash report ):-D
<OscarL-x64>
"looper must be locked before proceeding", danr lazy devs forgeting to properly lock/unlock things... /me shakes head.
illwieckz has joined #haiku
<OscarL-x64>
Oh, thanks Thor! hBlock seems to actually work in helping to reduce the amount of crappy ads.
freddietilley has joined #haiku
higen has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
<Begasus>
re
<andreasdr[m]>
Hi trhere.
<andreasdr[m]>
there.
AlaskanEmily has quit [Remote host closed the connection]
<Begasus>
Hi andreasdr[m]
<OscarL-x64>
'lo andreasdr[m].
<OscarL-x64>
Either I got lucky, or I'm finally getting a bit better at this. Finished updating the patch for 3.12 (+ fix for nightlies). Let's try a parallel build (will likely fail, but... worth a try :-D)
<Begasus>
nice, always a bit tricky, specialy for the larger ones
<andreasdr[m]>
Hi Begasus and OscarL.
<Begasus>
ah, the famous lag! :)
<OscarL-x64>
parallel builds + optimized builds still a big "no no" :-( Oh well...
<Begasus>
a bit overkill OscarL-x64? :)
<OscarL-x64>
yeah. profiler gets confused, you end up with a valid .hpkg... but with less proper optimizations :-D
<Begasus>
heh
<OscarL-x64>
defaulting to the slow build then :-D
<Begasus>
guess expected
<OscarL-x64>
indeed but... one can hope! :-D
<Begasus>
twice as slow, 4 times build slow speed :D
<Begasus>
using -j8 (default) in haikuporter?
<Begasus>
iirc*
<Begasus>
hmm .. maybe could change that to -j12 here to speed up some builds? :)
<OscarL-x64>
don't you have 16 threads on there?
<Begasus>
yeah
<Begasus>
not using them all
<OscarL-x64>
why not? I give them ALL I got, even on the VM :-D
<OscarL-x64>
(unless they fail, as in this case, of course!)
<Begasus>
heh
<Begasus>
k, let's check this
<OscarL-x64>
weird, still got a "Python-ast.gcda:Merge mismatch for function 95" even after doing a `make clean`. /me will do `hp -c python3.12` instead :-(
mmu_man has quit [Ping timeout: 480 seconds]
<Begasus>
not using "make distclean"?
<Begasus>
minus 3 minutes for building texstudio ...
mmu_man has joined #haiku
<OscarL-x64>
Maybe I just forgot to save my last .recipe change.. let's see :-D
<OscarL-x64>
k. `make clean` DID ran this time... will see if I get again that profile error.
<Begasus>
about 4 minutes with -j8
<Begasus>
so some speed improvement
<OscarL-x64>
booh! got some fake cores in there or what? ask for a refund! :-P
<OscarL-x64>
(more seriously, not all builds are easily parallelizable)
<Begasus>
this was also just a quick test, could be in Terminal (without haikuporter) it improves even faster :)
<Begasus>
extracting/initializing/packaging isn't optimized I think with haikuporter
<Begasus>
in Terminal, build time 1 minute :P
<OscarL-x64>
I tend to measure build times with hp just after I did `hp -b package_name` :-D
<Begasus>
heh, that's cheating :P
<OscarL-x64>
one of these days I'll patch haikuporter to do the builds on RAMFS and not under "work-*` dirs next to the .recipes.
<OscarL-x64>
(at least of my personal usage)
<Begasus>
could speed things up pretty well too I guess
<OscarL-x64>
did a quick test with Python 3.13.0a3... indeed quite the speed up.
<OscarL-x64>
(quick build not using haikuporter)
<Begasus>
yeah, seen it multiple times compared with not using haikuporter
<Begasus>
well, for new things I always run it in Terminal anyway
<andreasdr[m]>
:DDD
<Begasus>
18 branches ... time to do some housekeeping there
<OscarL-x64>
from current Python build log: "LIBRARY_PATH=/sources/Python-3.12.2:%A/lib:" <<< wonder what the heck that %A is, or where it comes from.
<OscarL-x64>
darn Python's `make test --help`... stop LIYING to me!
<OscarL-x64>
it says that the "none" resource is the default... actually does whatever it want, like using for 3.12: "resources: all,-audio,-gui,-largefile"
<Begasus>
tss :)
<OscarL-x64>
or plain ALL of them like with 3.13 yeasterday :-)
<Begasus>
maybe that's the %A previous :P
<OscarL-x64>
apparently, some python modules replace %A at the start of a path with the current working directory.
<OscarL-x64>
At least, that's why I see thanks to TextSearch :-/
<Begasus>
lifesaver! :D
<Begasus>
One of those small things that are not appreceated enough
<OscarL-x64>
been using it since it was called TextGrep, I've added the "open exact line in Pe" to it looong ago :-D
<Begasus>
thanks for that! +1
<OscarL-x64>
*TrackerGrep (no wonder I couldn't find it in HaikuArchives :-D)
<botifico>
[haikuports/haikuports] Begasus 7bd01a9 - qbittorrent, revbump switch to new boost version (#10106)
<Begasus>
k, that was in the pipeline for a few days :)
<OscarL-x64>
nice... if *only* I was a pirate!
<OscarL-x64>
that qbitorrent could indeed come in handy (says he while having about 620 GB upload / 600 download on it on another machine :-P
<Begasus>
rofl
<Begasus>
you know there's also amule :P
* OscarL-x64
prefers "KaZaa Lite" (damn, that takes me back a couple of decades!)
<Begasus>
newsgroups when they were still open :D
<Begasus>
eg, not blocked*
<Begasus>
or even BBS(?) :)
<OscarL-x64>
never got to use those (USENET), and I was even late for napster. Morpheous -> KaZaa Lite -> uTorrent -> qBitTorrent. SO many .ISOs shared over the years! brings a tear to my eyes!
* OscarL-x64
waits for another Python test-run to hopefully finish, properly this time.
<phschafft>
;)
<phschafft>
currently writing something on windows 3.11 ;)
<Begasus>
whoot!
<Begasus>
take about going back :D
<phschafft>
;)
<phschafft>
I just remembered how I watched Ice age (as VCD) on my wfw machine.
<Begasus>
never got those here
<phschafft>
it was my only VCD.
<phschafft>
but it was fun, just insert it and it's media player worked fine with that. no extra software required.
<OscarL-x64>
my first VCD was "The Matrix" on 2 discs... in English, with no subtitles. after the 2 watch... I would understand all the dialog, NOICE
<Begasus>
I've seen them working, but never owned one :)
<Begasus>
lol
<OscarL-x64>
(played it on PC, thou)
<OscarL-x64>
I think those were my first... and LAST time using VCD. lol.
<phschafft>
it was... a short living standard.
<OscarL-x64>
welp, these hangign tests are at least consistent... hang reliably... /me disables them manually.
<phschafft>
I just realised, I'm kind of looking at wfw daily at work. as in right now.
<OscarL-x64>
phschafft: do remember reading about SVCDs, but never got one myself.
<phschafft>
After all I made a theme for my Debian 12 with the original icons and stylings (oh how much work that was...)
coolcoder613 has quit [Ping timeout: 480 seconds]
<Begasus>
BeOS style phschafft? ;)
<phschafft>
you want me to do a BeOS style that looks like wfw? ;)
<Begasus>
nah, your Debian icon set :)
* Begasus
doesn't know what wfw even is ...
<phschafft>
you want a Debian 12 icon set for BeOS? ;)
<phschafft>
Begasus: windows for workgroups (3.11)
<Begasus>
ah ... bumps head :D
<Begasus>
did some icon sets in ZETA back then phschafft :P
<phschafft>
:)
<OscarL-x64>
phschafft: hey, you, Debian guy! question for you... it is possible to do a netinstall and somehow make apt use "--no-install-recommends" (I mean... during the install process itself, and not, disabling "everything" and installing the "desktop" afterwards with thouse pesky "recommends")
coolcoder613_32 has quit [Quit: Vision[]: i've been blurred!]
* OscarL-x64
wants a simple xfce/mate desktop, without LO and stuff, but he's also lazy, and would want to do it in just one go of the netinstall :-D
<phschafft>
I never did a netinstall. but all recent installs had it disabled by default. not sure if the default was changed or if I might have just picked the right images.
<OscarL-x64>
tried it last night with netinstall for debian testing... selected destop as mate... ended up with GIMP and LO installed :facepalm:
<OscarL-x64>
first time installing debian thou, will try with the expert options. Can't be worse than installing SuSE 5.1 in 1998, I gather.
<OscarL-x64>
at that time... those package selections were scary, hehe... disable the wrong one... you better start from scratch :-D
<phschafft>
I don't use tasksel.
<phschafft>
while it might be nice for many users, I feel like it's intransparent on what it actually selects.
<phschafft>
but I must say, if you select a *desktop* option I would expect it to select some kind of office software and also some graphic tools. so gimp und lo seem sane to me.
<OscarL-x64>
indeed. Was expecting something more fine-grained, not sure if the "expert install" allows for that, will see.
<OscarL-x64>
*indeed, to your "intransparent" comment.
<phschafft>
I generally just skip that step and then install things manually. but surely that is very expert-y. maybe you still want to use tasksel for stuff like ssh and printing.
<phschafft>
specifically with me working on laptops all the time I want to have as many printer drivers pre-installed as possible. you never know.
<phschafft>
and it's also fun to get somewhere, connect to someone's printer and they're like 'printing on linux? no way!' and you print without even installing a driver. ;)
<OscarL-x64>
yeah. guess I should just bite the bullet, install base system, reboot, and do apt instal <some-desktop> --no-install-recommends. Oh well. Etither that... or back to Porteus :-D
<OscarL-x64>
remindsme... I only ever need "print to PDF", if ever, so cups and co, can go too :-P
<OscarL-x64>
lol at this error: "AssertionError: 1707.909866 not greater than or equal to 1707909865.8281128"
<OscarL-x64>
Finally a `hp --test python3.12` and ends without manual intervention :-D
<OscarL-x64>
results: Total test files: run=438/425 failed=23 skipped=24 resource_denied=10 rerun=23
<OscarL-x64>
Vererict: Good enough for me. Ship it!
mmu_man has joined #haiku
<phschafft>
haha.
<OscarL-x64>
YES! finally a test run that doesn't totally sucks!
<OscarL-x64>
Results (in under "real 4m18,879s"): "Total test files: run=437/424 failed=22 skipped=24 resource_denied=9 rerun=22"
<OscarL-x64>
not having the test runs be a TOTALLY PITA... makes your QA team less miserable.
<OscarL-x64>
(says me after doing QA and QA automation for 7 years)
<OscarL-x64>
Now let's do the same for Python 3.11.8, but... coffee first :-D
<phschafft>
Python 3.11? Python for Workgroups? ;)
<OscarL-x64>
isn't Debian 12 using that? :-D
<phschafft>
hm, maybe. But I thought with python you need to install *all* the versions as they are completly incompatible to each other.
<phschafft>
at least I'm used to have like six or seven versions installed on a desktop and if you try to remove any one of them half of the system is removed with it.
<OscarL-x64>
I only have several installed because I try to keep the Haiku packages reasonably updated. Otherwise... 3.10 is good enogh for a system-wide version.
<OscarL-x64>
in Linux you can (and should, I think), install the version you want, but only for your user/container/whatever, and only pip install things "in there".
<OscarL-x64>
Haiku with its system vs home installs, vendored-packaged, and packaged vs non-packaged... makes it not that much of a pain to have plenty around...
<OscarL-x64>
still, I rather only use `pip3 install --user`, just in case :-D
BrunoSpr has joined #haiku
<OscarL-x64>
I always get the "funnies" crashes when doing .recipe work on Pythons... Tracker crashed with. apparenty, and assertion fail: "getNumAvailable() == 0" near "BPrivate::superblock::getBlock() + 0x8e "
<OscarL-x64>
s/funnies/funniest/
dovsienko has joined #haiku
<dovsienko>
hello all. on Haiku with all the latest updates installed, if you run "ping localhost", the loopback interface counters in ifconfig remain zero:
<dovsienko>
this may, or may be not be related to the fact libpcap can start a capture on the "loop" interface, but no packets arrive
<dovsienko>
so is this a bug or a peculiarity of the network stack design?
<phschafft>
I have no idea, but I know that for example Linux does shortcuts for local traffic. so it could be similar here. but we would need for someone who knows to actually tell.
coolcoder613_nok has quit [Read error: No route to host]
<OscarL-x64>
dovsienko: you'll probably want to "ping" waddlesplash about it, when he gets around (different timezones, and all that).
<dovsienko>
thank you, let's try that, I will stay on the channel today
<OscarL-x64>
he might reply later. In case you need to log out, this channel is logged, so maybe do an CTRL+F for your name on those later on (link to the logs: https://oftc.irclog.whitequark.org/haiku/)
<Begasus>
you could also look into the bugtracker to see if it's a known problem dovsienko
<Begasus>
CTRL-F?
* Begasus
smacks OscarL-x64
<Begasus>
Alt-F! :P
<Begasus>
or rather Alt-f :)
<BrunoSpr>
hello all, do an CTRL+F for your name? What is this doing?
<dovsienko>
finds a string in a web page
<dovsienko>
(or does not, depending on the string and the page)
<BrunoSpr>
ah ok thx
<BrunoSpr>
thought is is some special Vidion feature, sorry
<BrunoSpr>
* Vision
<Begasus>
common shortcut BrunoSpr :)
<BrunoSpr>
yes, thx
BrunoSpr has quit [Quit: Vision[]: Ich wurde gewaschen!]
BrunoSpr has joined #haiku
zard has joined #haiku
* OscarL-x64
troutslaps Begasus wihile double checks to see if his Keymap is set to Win/Lin mode on this install.
* OscarL-x64
is a CUA-keybindings kind of guy.
<OscarL-x64>
Later folks, out for some errands.
OscarL-x64 has quit [Quit: Vision[]: i've been blurred!]
<Begasus>
traitor! :P (I know you read the logs) :)
BrunoSpr has quit [Quit: Vision[]: Ich wurde gewaschen!]
BrunoSpr has joined #haiku
BrunoSpr has quit [Quit: Vision[]: Ich wurde gewaschen!]
comrad has quit [Read error: Connection reset by peer]
comrad has joined #haiku
BrunoSpr has joined #haiku
BrunoSpr has quit [Quit: Vision[]: Ich wurde gewaschen!]
BrunoSpr has joined #haiku
BrunoSpr has quit []
comrad has quit [Ping timeout: 480 seconds]
m_eiman has quit [Read error: Connection reset by peer]
BrunoSpr has quit [Quit: Vision[]: Ich wurde gewaschen!]
BrunoSpr has joined #haiku
FreeFull has quit []
BrunoSpr has quit []
Begasus has quit [Quit: Vision[]: i've been blurred!]
Begasus has joined #haiku
dovsienko has quit [Ping timeout: 480 seconds]
dovsienko has joined #haiku
OscarL has joined #haiku
* OscarL
reads the logs... from Windows... Mmm...
<OscarL>
Heh, we hit a limit on Github: "Contributions to master, line counts have been omitted because commit count exceeds 10,000." :-D
<phschafft>
;)
* OscarL
thinks that his commit graph can be useful next time a psychiatrist asks him about his ups and down. :-P
<OscarL>
(word count on the IRC logs as second source of data, LOL)
<phschafft>
I actually know about some people exporting IRC statistics for their therapist. (keep in mind that I'm a bit into social work.)
<OscarL>
makes sense, actually, at least for those folks self-aware enough of their behaviour patterns :-D
KapiX has joined #haiku
<OscarL>
mood analysis in my logs would be: either excess of "happy emoticons", or... haven't seem him in a while :-P
BrunoSpr has joined #haiku
BrunoSpr has quit []
<zard>
No doubt that would be interesting :P
<phschafft>
x512[m]: does the ELF loader on Haiku allow for a set of functions to be loaded in a (logical) segment with nothing else in there?
<phschafft>
basically meaning that if I pack them in such a (logical) segment and I know the size of it and the start (which is just the symbol with the lowest address) I could figure out which function a pointer points into by comparing the pointer to the list of symbols in that (logical) segment.
<phschafft>
(I'm aware that there are high level functions to do that for me, just wondering here about implementation details)
mr_lou has joined #haiku
KapiX has quit [Quit: KapiX]
* OscarL
nods and "hmm"s, pretending he understood any of that, but found a file containing "Contains the ELF loader", so... why not share a link here:
<phschafft>
wondering if Nephele has some time for me tomorrow.
slendi has joined #haiku
<slendi>
Hello, I am having an issue with haiku where I cannot compile a simple C hello world program, both gcc and clang17 tell me that they cannot find stdio.h. What package do I need to install?
<OscarL>
slendi: try with `pkgman install haiku_devel` ?
<slendi>
let me try that real quick
* OscarL
starts his SLOW VM in the meantime
<slendi>
Worked thanks
<OscarL>
awesome. I assume you're running a nightly image then, right? (because IIRC, beta4 already includes all the needed packages)
<slendi>
I don't know I just went to the website, clicked the big yellow button and downloaded the 64 bit torrent
<zard>
So, you're on beta4
<OscarL>
you'll find the "hrev" number (short for Haiku Revision number) either with `uname -a`, or by openining the AboutSystem from Deskbar's Menu.
<augiedoggie>
OscarL: %A in the library path is the current executable location
<slendi>
hrev56578+95
<OscarL>
augiedoggie: thanks for confirming! just to be clear... is that a "Haiku" thing, or a POSIX/unixy thing?
<augiedoggie>
Haiku is the only place i've seen a dynamic path like that
<OscarL>
augiedoggie: ok. I also saw it used in one of Python's "patched for Haiku" modules, so at least I understood how it was used...
<OscarL>
just seeing that as part of the "./configure" output log got me a bit dissoriented :-)
<OscarL>
slendi: that's beta4, for sure. Weird you needed to install haiku_devel manually. Oh well... as long as it works now.
<win8linux[m]>
Only just found out that Google laid off a bunch of folks who worked on Fuchsia.
<slendi>
When I installed, all optional packages seem to have been messed up I think, they are present in /system/packages but not installed
<OscarL>
win8linux[m]: again?
<slendi>
So when installing I had to remove some files from that directory
<win8linux[m]>
Wonder if any of the ex-Be folks are still there.
<OscarL>
slendi: ah... that might explain it then :-)
<slendi>
Yup
orealis has quit [Quit: yap...]
<OscarL>
slendi: I guess you could do a `pkgman full-sync` to try and make sure your system and the repos are 1:1...
<slendi>
Let me see
<OscarL>
or... perhaps a bit more risky, depending on what you ended up with under /system/packages/...
<slendi>
Lots of upgrades it seems
orealis has joined #haiku
<slendi>
I just deleted like 3 packages from there
<win8linux[m]>
OscarL: No, just found out indirectly from this article on full Chrome no longer coming to Fuchsia:
<OscarL>
you could also try removing /boot/system/packages/administrative/activated-packages, and doing a reboot (that should force re-detecttion/re-activation of those packages)
<OscarL>
win8linux[m]: ah, ok. I was thinking... damn another layoff there... must be only Travis now still there :-P
<waddlesplash>
dovsienko: that's definitely not with latest updates, because I fixed that bug with counters on the nightly builds. loopback packet capture was also fixed since the beta, IIRC
<slendi>
s there a shortcut for switching terminal tab?
<win8linux[m]>
OscarL: Heh, tbh I only knew of Travis as an ex-Be person who worked on Fuchsia. Don't really knew if there were any others.
<win8linux[m]>
s/knew/know/
<slendi>
the fonts got messed up and cannot really navigate it with mouse
<OscarL>
slendi: I thinkg you got hit by a pretty recent bug/regression. Should fix itself after a reboot.
<waddlesplash>
yeah, just reboot
<waddlesplash>
"shutdown -r" from terminal
<nosycat>
Oh yeah, the noto font update. That was scary.
<slendi>
yeah but is there a shortcut? I have my pkgman command on another tab
<waddlesplash>
yes, but I don't remember what it is lol
<waddlesplash>
Alt+Arrowkeys maybe?
<waddlesplash>
or something like that
<OscarL>
slendi hit and hold CTRL+ALT+DEL for a few seconds, that should reboot.
<slendi>
yeah but i dont really want to reboot if pkgman is still working
<OscarL>
right :-D
<slendi>
thats why im asking for the terminal tab switching keybind if there is any
<waddlesplash>
slendi: reboot while pkgman is still working generally shouldn't break anything
<slendi>
oh nvm i managed to click
<slendi>
rebooting
<waddlesplash>
only during the last stage, which lasts a fraction of a second, could it even possibly break your OS
<nosycat>
Good to know!
<OscarL>
For the record... and for NOW (untill I get my patchset finished and merged :grin:)... you switch between Terminal tabs with SHIFT+left / SHIFT+right.
slendi has quit [Read error: Connection reset by peer]
slendi has joined #haiku
<slendi>
I am back
<OscarL>
yay! :-)
<slendi>
CTRL ALT DEL didnt work so I forcefully powered off
<slendi>
I think I lost some configs lol
<slendi>
The Deskbar for some reason moved back to the top right but idk
<OscarL>
that's why I try to "> sync" before reboots... unless I get KDL'ed
<nosycat>
Bummer.
<slendi>
the terminal became a vertical mess so i couldnt really know what i was doing lmfao
<nosycat>
Luckily I was able to type `shutdown` when it happened.
<dovsienko>
waddlesplash: are you talking about the updates that arrive via "pkgman update"?
<waddlesplash>
if you started with beta4, the pkgman update will only give you beta4-channel updates
<waddlesplash>
the fixes I am talking about are on nightlies only
<waddlesplash>
so you will have to switch repositories in pkgman and run a "full-sync"
<dovsienko>
does it mean these bug fixes are going to trickle into beta given some time?
mmu_man has quit [Ping timeout: 480 seconds]
<OscarL>
slendi: at least you now understand why the "beta" in beta4 :-P. Happy "Haikuing"!
<nosycat>
Can't complain, it's been remarkably trouble-free otherwise.
<slendi>
Thanks!
<OscarL>
nosycat: indeed! I had almost forgotten about KDLs, for example, until I switched to nightlies to try some customized drivers :-D
<OscarL>
and building Python, and running its test-suites on nightlies... that *always* gives me "something new to play with" :-/
<nosycat>
:D
<dovsienko>
looks like this is commit 819c510 from 2023-12-29, so it is taking some time. maybe I will risk switching after all...
* OscarL
pre-emtively switched to Noto Sans on beta4 a few days ago... now tries to update.
<dovsienko>
on a separate note, I managed to get the first kernel debugger prompt on system shutdown. the only difference from all previous shutdowns was that there was a couple SSH connections opened at the time
<OscarL>
"pre-emtively" -> preemptively. so Begasus doesn't says I type so badly :-P
<dovsienko>
and the stack trace mentions EndpointManager::Unbind(TcpEndpoint*), so it looks there is a connection. would the complete stack trace be of any interest?
<dovsienko>
sshd also appears in the stack
mmu_man has joined #haiku
walkingdisaster has quit [Quit: Vision[]: i've been blurred!]
* dovsienko
has switched from r1beta4 to nightly and is trying to tell the difference
<OscarL>
dovsienko: uname -a == hrev57581? (just to be sure).
<dovsienko>
OscarL: hrev57581 Feb 13 2024 07:02:22
<OscarL>
sounds about right :-)
<OscarL>
what speeds do people in EU or US get while updating packages from htts://eu.hpkg.haiku-os.org ?
<nosycat>
Much higher lately.
<nosycat>
Something has improved.
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<waddlesplash>
dovsienko: if the crash still happens on the nightly, please do report it, yes. don't recall anything like this
<waddlesplash>
the nightly builds have more assertions and sanity checks, so sometimes KDLs differ or are mor informative there
<OscarL>
nosycat: glad for you! I never got more than an ocassional peaks of 350 KB/s, more usually around 100/200 KB/s if I'm lucky.
mmu_man has quit [Ping timeout: 480 seconds]
<OscarL>
lol, even setting Noto Sans BEFORE doing the update... somehow messes up the system, heh.
<nosycat>
That was the case before a recent update.
<nosycat>
Always thought it was my wi-fi signal, or drivers.
<dovsienko>
waddlesplash: it did not reproduce since, but maybe the stack trace will give a good enough sense of what could go wrong: https://pasteboard.co/tLILpcmRhfK4.png
<OscarL>
nosycat: I get the same even on latest nightlies, via ethernet (while being able to download up to 2 MiB/s from elsewere... oh well).
<OscarL>
another reason for me to want a "LAN package proxy" for Haiku updates :-D
<nosycat>
Sounds like it's on the server side then.
<waddlesplash>
hmm, did ambroff's fixes actually fix this too?
<slendi>
how do i swap the escape and capslock keys? I have the keymap application open, I figured out I can remove keys, but how do I change them?
<dovsienko>
to me it looks like sshd's request to close a socket coincided with some other shutdown-related cleanup and caused a race condition, but I trust you know better
<nosycat>
I'm in Bucharest, dunno if that's relevant.
<waddlesplash>
dovsienko: no idea. I've been doing some work in the TCP module lately but I haven't touched the endpoint hash logic.
<slendi>
nosycat: salut
<nosycat>
Ha ha.
<dovsienko>
let me update the bug then
<slendi>
Figured it out, had to drag and drop
<slendi>
nvm its different for shift
<OscarL>
slendi: Haiku's keyboard/shortcut handling can go from... "less than ideal" to "highly infuriating". YMMV.
<slendi>
yeah i cannot seem to be able to select escape from right click menu
<slendi>
is keymap program the only way to edit a keymap?
* OscarL
still tries to find how to open an app menu with just the keyboard.
<nosycat>
And that setting that lets you swap the Ctrl and Alt keys... let's just say the handling is inconsistent.
<OscarL>
I can complain about Haiku's keyboard handlind for days... wish I could offer solutions instead :-/. Oh well... at least it seems I can keep a python or two updated :-D
<OscarL>
naptime. See you around folks! Have a good one, and have fun on Haiku!
OscarL has quit [Quit: zzzZZZzzz]
v_harkonnen has quit [Ping timeout: 480 seconds]
v_harkonnen has joined #haiku
<dovsienko>
waddlesplash: on a related note, would it be correct to say that after ioctl(SIOCSPACKETCAP) all captured packets will always be delivered with the same length as the packets on the wire?
<waddlesplash>
they should...
<waddlesplash>
slendi: Keymap GUI, but you can also "decompile" keymaps to text, see "keymap" program, and edit that way
<dovsienko>
so, regarding the ifconfig counters, I confirm they increase in general, but Tx counters seem to be the sum of Tx and Rx on the wire, and Rx counters stay zero
<dovsienko>
i.e., ping localhost -c 5 increments transmitted packet counter by 10 and that's it
<dovsienko>
however, libpcap now sees both Tx and Rx packets, which is good
<dovsienko>
both of the above observations apply to the latest nightly revision
mmu_man has joined #haiku
bbjimmy has joined #haiku
Hannah has quit []
<waddlesplash>
hmm, odd. I would expect for localhost that rx may not get incremented at all
<waddlesplash>
oh
<waddlesplash>
dovsienko: actually this is what I'd expect. the RX counter for loopback isn't incremented at all because loopback RX bypasses the normal mechanism (as the packets are already in net_buffers)
<waddlesplash>
so, ping sends a packet (increments counter) -> stack replies to ping (increments counter), in both cases this counts as a "send *to* loopback"
slendi has quit [Read error: No route to host]
HaikuUser has joined #haiku
HaikuUser has quit []
<dovsienko>
waddlesplash: in Linux I am observing an increment of 2 on both Rx and Tx for each ping-pong pair packets, which makes complete sense
<dovsienko>
on OpenBSD the stats are very much asymmetric, but still there's data registering in both directions
slendi has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
<dovsienko>
same in NetBSD (slightly less asymmetrical though)
<dovsienko>
correction: in NetBSD it is perfectly symmetrical (was looking at the wrong interface)
<dovsienko>
same in FreeBSD
<dovsienko>
arguably, even for loopbacks stats should be maintained for each direction independently in order to register different types of errors, but this does not block any of my work
slendi has quit []
tqh has joined #haiku
tuaris has joined #haiku
tqh has quit [Quit: Leaving]
tqh has joined #haiku
<waddlesplash>
dovsienko: what is your work, if I may ask?
nosycat has quit [Quit: Leaving]
x10z has joined #haiku
slendi has joined #haiku
<x512[m]>
phschafft: > x512: does the ELF loader on Haiku allow for a set of functions to be loaded in a (logical) segment with nothing else in there?
<x512[m]>
Yes. You can put a set of functions into separate section and add 2 symbols that indicate begin and end of that section.
<Begasus>
closing down here, cu peeps!
Begasus has quit [Quit: Vision[]: i've been blurred!]
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<slendi>
OK, weird discovery: I think my system is running with framebuffer driver instead of radeon_hd, I looked into driver.c and found my device's PCI ID. Is there a way to disable the framebuffer one and enable the radeon_hd one instead?
<slendi>
Using EFI so probably that's why it's using the framebuffer but I am not sure
zard has quit [Quit: leaving]
jmairboeck has joined #haiku
<tqh>
framebuffer should be the fallback. You should be able to read in the syslog why the radeon driver didn't work.
<dovsienko>
waddlesplash: trying to make practicable improvements to the Haiku module in libpcap. I started at a simple bug in the Hurd module and used Haiku module to cross-reference the logic, then after fixing some bugs at the Hurd end I realised the Haiku module experiences some of the same
<waddlesplash>
ah, interesting!
<waddlesplash>
x512[m]: phschafft: you have to be careful though, GCC reorders symbols within a file these days by default as part of optimizations. you need to use one of the -f to stop it from doing so to make that trick work
<dovsienko>
one obvious space for improvement is that the userland BPF filtering uses a packet size that is too short, hence filters may reject more packets than they should
<x512[m]>
waddlesplash:
slendi has quit [Read error: Connection reset by peer]
<x512[m]>
> As long as the section name results in a valid C variable name, gcc (ld, rather) generates two magic variables: __start_SECTION and __stop_SECTION. Those can be used to retrieve the start and end addresses of a section, like so:
Anarchos has joined #haiku
<waddlesplash>
oh, sections. yes I see
<dovsienko>
now that you say the packets captured via a monitor are always full-size, this makes it easier to get the size right, but one thing I see at the loopback interface is pseudo-Ethernet headers with all-zeroes MAC addresses
<x512[m]>
So you can put functions in separate section and detect that function is inside that section by special compiler-generated symbols.
slendi has joined #haiku
<waddlesplash>
dovsienko: yes, I think that was added because pcap was rejecting the packets otherwise, or something like that.
<waddlesplash>
or something seemed to really want ethernet headers even for loopback
<slendi>
im back haiku decided to put me in the debugger and not return me to the desktop when i tried to exit lol
<Anarchos>
hello
<slendi>
Anarchos: hi
* Anarchos
always has to put mtu=1400 to do ssh. Weird
<dovsienko>
I was going to make a point that it ought to be something like DLT_NULL or DLT_RAW, but surprisingly Linux loopback delivers exactly the same
<waddlesplash>
Anarchos: korli reported that on beta, but it may be fixed on nightly, not sure
<waddlesplash>
dovsienko: yes, I think that's why we started doing it
<Anarchos>
waddlesplash not fixed for me.
<waddlesplash>
odd
<waddlesplash>
I think someone said it only happens with Haiku to Haiku ssh connections? and if there's anything else on the other end it works
<waddlesplash>
either way a tcp trace should hopefully indicate the problem
<waddlesplash>
guess the recent rounds of TCP fixes did not improve matters here
<Anarchos>
yes it is Haiku to Haiku, on local network.
<waddlesplash>
both ends are nightly build?
<waddlesplash>
recent one?
<Anarchos>
hrev57579 and x86_64 for both.
<waddlesplash>
ok. yeah, that's past all the recent fixes
<waddlesplash>
Anarchos: are there any errors/dropped listed in ifconfig stats?
<Anarchos>
Transmit : 0,error, 1 drop
<Anarchos>
with mtu=1600
mr_lou has quit [Quit: Leaving]
<waddlesplash>
does send seem to increment when you try to start a ssh session?
<tqh>
maybe our ssh port needs recompiling or fixing?
<waddlesplash>
drivers using MTUs larger than 1500 only became a default last year
<waddlesplash>
tqh: I kind of doubt it, if changing MTU fixes the problem this sounds like a TCP issue
<waddlesplash>
and it sounds like other TCP issues we've had in the past
<waddlesplash>
or maybe even an IPv4 issue
<tqh>
waddlesplash, if ssh uses mtu and calculates wrong I guess.
<Anarchos>
waddlesplash threre is no send log, only receive, error, dropped
<Anarchos>
and mcasts
<waddlesplash>
log? I mean if the "send" packet count increases
<Anarchos>
waddlesplash where do i find the send packet count?
<waddlesplash>
in ifconfig?
<Anarchos>
the transmit increased, indeed
<waddlesplash>
ok, very good
<waddlesplash>
so it's at least getting that far
<waddlesplash>
then the question is, does the other end receive it?
<Anarchos>
1938 -> 1963
<waddlesplash>
well it sounds like there's traffic going both ways then
<waddlesplash>
so, probably just need to check packet logs to see where/why it stalls out
<Anarchos>
will look at it later. Anyway, how can i trace tcp ?
<waddlesplash>
tcpdump
<waddlesplash>
or Wireshark GUI
<waddlesplash>
though, if it's this easily reproduced, you may not need to do that at all
<waddlesplash>
just open a ticket and I'll take a look myself in the next few days I suppose
<tqh>
we don't do anything strange in the ssh port at least.
jmairboeck has quit [Quit: Konversation terminated!]
HaikuUser has joined #haiku
HaikuUser has quit []
coolcoder613_32 has joined #haiku
<phschafft>
x512[m]: thank you.
<phschafft>
waddlesplash: that is why I thought about having a specific section, so reordering of the functions is no problem. it would only be a problem if addition objects would be added.
<slendi>
How do I let the kernel let me write files to /system/develop?
xet7 has joined #haiku
<Anarchos>
coolcoder613_32 yes ?
<coolcoder613_32>
The IRC log checkboxes don't work in Web+
<slendi>
I want to make install a library I compiled, put the proper directories there but no shot, it cannot do any writing to the filesystem
<coolcoder613_32>
slendi: use /system/non-packaged/lib/ instead
<slendi>
ok
<slendi>
what about header files
<coolcoder613_32>
use /system/non-packaged/develop/headers/
<slendi>
ok
x10z has joined #haiku
<coolcoder613_32>
does that lbrary use ./configure?
<slendi>
No
<slendi>
It has a makefile
<slendi>
it's libvterm
<slendi>
I installed it and cmake seems to find it
<coolcoder613_32>
We have libvterm
<coolcoder613_32>
use pkgman install libvterm_devel
<slendi>
it tells me no matching packages found when i search for it
<slendi>
is it in a repo i dont have added?
<coolcoder613_32>
In HaikuDepot or pkgman?
<slendi>
pkgman
xet7 has quit [Remote host closed the connection]
<coolcoder613_32>
pkgman search vterm?
<slendi>
yes
<coolcoder613_32>
Paste from the terminal to bpa.st
<slendi>
Welcome to the Haiku shell.
<slendi>
~> pkgman s libvterm
<slendi>
No matching packages found.
<slendi>
~> pkgman s vterm
<slendi>
No matching packages found.
<slendi>
~>
<slendi>
oops
<slendi>
meant to paste the link sorry
<coolcoder613_32>
Why pkgman s? use pkgman search
<slendi>
Is it not the same?
coolcoder613_nok has joined #haiku
<coolcoder613_32>
not AFAIK
xet7 has joined #haiku
coolcoder613_nok has quit [Read error: Connection reset by peer]
<slendi>
coolcoder613_32: It is the same
coolcoder613_nok has joined #haiku
<coolcoder613_32>
can yoiu search for other things?
<coolcoder613_32>
like.. digger?
<coolcoder613_32>
try pkgman s digger
<coolcoder613_32>
and paste the output of pkgman list-repos
coolcoder613_nok has quit [Ping timeout: 480 seconds]
x10z has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<dovsienko>
waddlesplash: upon a closer look at the "receive a packet" call I found that the data has a Ethernet framing (whether real or not), but the sockaddr structure is filled differently
<dovsienko>
when capturing on Ethernet, it fills the sockaddr with sdl_family == 4 (AF_LINK), just as the comment in /system/develop/headers/posix/net/if_dl.h says
<dovsienko>
when capturing on loopback, the AF == 1 (AF_INET), and decoding the sockaddr as sockaddr_in yields correct sin_len == 32 and sin_addr set to the IPv4 address
<dovsienko>
do you know if this is this intentional too?
<slendi>
coolcoder613_32: Yes digger shows up, I only have Haiku and Haiku Ports repositories for r1beta4
<slendi>
sory I was inactive, I am battling with luajit in neovim build
<dovsienko>
it is acceptable if loopback pretends to be an Ethernet, but it would be nice if it pretended better
<dovsienko>
alternatively, it could just lose the Ethernet pseudo-header and let the receiver see whether that's IPv6 or IPv4 from sockaddr.sa_family
<dovsienko>
currently it seems to be neither of these two cases
x10z has joined #haiku
nmbosley has joined #haiku
<waddlesplash>
dovsienko: hm, we may have a bug here then. you may want to open a ticket