<kallisti5[m]>
Honestly, the more I look at build master the more I think it needs rewritten. It makes a ton of crappy assumptions
<waddlesplash>
it needs a rearchitecture, maybe
<kallisti5[m]>
However, I hate python.. so it's a good project for someone other than me 😁
<waddlesplash>
kallisti5[m]: use an IDE
<waddlesplash>
pypy is ur friend
<kallisti5[m]>
vim, this is the way.
<kallisti5[m]>
waddlesplash: a potential design exists where buildmaster dumps it's data to a private mongodb.. then someother microservice parses the data and presents it for frontend apps
<kallisti5[m]>
Tldr; getting the data out of buildmaster as quickly as possible, then making it pretty separately. That architecture design smells though
<kallisti5[m]>
Buildmaster really needs a solid refactor at minimum. It's too many duct-tape shell scripts
<kallisti5[m]>
Anyway. I won't rant there anymore. I'm not doing extensive python... So that makes my position simpler 😁
Skipp_OSX has quit [Read error: Connection reset by peer]
Skipp_OSX has joined #haiku
euandreh has quit [Quit: WeeChat 3.5]
euandreh has joined #haiku
x10z has joined #haiku
x10z has quit []
Jupp_S has joined #haiku
Jupp_S has quit [Remote host closed the connection]
Andros has joined #haiku
tuaris has joined #haiku
Vitto has joined #haiku
Begasus has joined #haiku
<Begasus>
g'morning peeps
Begasus_32 has joined #haiku
<Anarchos>
Begasus hi
<Begasus>
Hi Anarchos
<Anarchos>
there are some memory leaks : when i cross-compile the nightly, after that my memory is still almost used by kernel
<nekobot2>
[haiku/haiku] 46bbf334f635 - intel_extreme: parse VBT device configs, use this to skip unused ports
Gautam has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Anarchos>
interesting : a haiku_loader.pxe_ia32 is able to download/untar a haiku-netboot-archive built in generated.x86_64, but this one fails to find the remote boot disk
<Anarchos>
when all is done on 32 bits, it works.
dvn has quit [Quit: bye]
<Anarchos>
though there is no pxe_ia64
dvn has joined #haiku
Gautam has joined #haiku
Vitto has quit [Quit: Vitto]
Vitto has joined #haiku
Vitto has quit [Ping timeout: 480 seconds]
<nephele>
maybe the pxe support is older than 64bit?
<Anarchos>
nephele sure, but i am not even sure a haiku_loader.pxe_ia64 is needed
<Anarchos>
its sole task is to retrieve and lauch haiku-netboot.tgz
<nephele>
maybe the payload expects to launch in long mode
<nephele>
ah nvm
<Anarchos>
nephele no idea
<Anarchos>
nephele do you know the difference between @nightly-raw and @haiku-minimum ?
<Anarchos>
nephele now , i am atrying to remotely boot a haiku-image
<Anarchos>
a nightly-raw, 32bits
mmu_man has joined #haiku
Anarchos has quit [Remote host closed the connection]
<nekobot2>
[haiku/haiku] 459f1bcd8d68 - intel_extreme: irq hotplug support for IceLake and newer PCH.
<Begasus>
heading down here, will keep an eye on buildmaster later, cu peeps!
Begasus has quit [Quit: Leaving]
<extrowerk>
I have noticed lately on different OS platforms: more and more program behaves similarly at start: they doen't show text during the startup, but the underlines symbolizing the links are already visible, even the mouse cursor knows where the text supposed to be, because it switches to selection tool cursor, but the text still needs time to load. Is this the so called react UI or something similar?
<phschafft>
checked. also checked how it works. which is actually not a bad design.
<phschafft>
what I however sadly see to often is that people have no idea about change mangement meaning they break things all the time. which is also part of the problem.
x10z_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
eroux__ has quit []
eroux has joined #haiku
<x512[m]>
extrowerk: nheko crashes a lot for me.
<extrowerk>
phschafft: give me a link about that in the manual.
<extrowerk>
x512[m]: it was working for me 1 week ago, but now it behaves strange on my side too.
<phschafft>
in my make(1) 'If the -j option is given without an argument, make will not limit the number of jobs that can run simultaneously.'
<extrowerk>
phschafft: yes, this is true, thanks.
<phschafft>
not sure if we have the same version of make and/or the manual, however at least over here it is documented.
<extrowerk>
it is documented here too.
<extrowerk>
is the qt support in qt6 supposed to work from now on?
<extrowerk>
i mean the opengl support, sorry.
<ermo[m]>
the '-j' parameter is likely designed to be used in conjunction with the load-limiter behaviour.
<ermo[m]>
i.e. don't spawn more jobs until the load has fallen below X
<ermo[m]>
so the idea would be to do e.g. "gmake -l $((ncpu + 2)) -j"
<ermo[m]>
sorry, s/ncpu/nproc
<extrowerk>
ermo[m], this examples doesn't makes any sense to me, they are completely unrelated t omy use cases. I can assume it is useful for somebody, but i think gnu tools wasn't developed with the simple user in mind.
ClaudioM has joined #haiku
<extrowerk>
so i built some qt opengl demos, but they doesn't working. sadly.
<ermo[m]>
extrowerk: out of nothing but idle curiousity, what is your use case? What are you trying to accomplish?
<extrowerk>
configure && make && make install
<ermo[m]>
let me rephrase: How many threads would you like to use?
<ermo[m]>
as many as your CPU has?
<extrowerk>
i am planing to buy an IBM T42 ThinkPad or retro activities.
<extrowerk>
ermo[m]: i have 4 threads, so i wanted to write "make -j 4", but my keyboard somehow did not recognized the "4" keypress
<extrowerk>
so it went with infinite processes, which quickly exhausted the system.
<ermo[m]>
and then you were ... surprised that `-j` by itself spawned infinitely many threads...?
<extrowerk>
exactly
<ermo[m]>
ok, so, my use case is centered around long-running compilations while you may be using the system for other things
<ermo[m]>
let's say you have 4 threads available. Doing `gmake -j -l 3.0` would ensure that whenever the system load is over 3.0, gmake wouldn't start new compilation jobs, thus likely leaving at least one thread free to service you
<ermo[m]>
If, in addition, you were to use (the equivalent of) `nice -n20` and `ionice -c3`, you could have a compilation that only uses spare CPU and I/O capacity.
<extrowerk>
ermo[m]: i don1t see the point of spawning infinite processes, even if i try to build a huge source tree. My optimum is either the thread number of the cpu, or 1 less. For heavily boost dependant sources i use core-number.
<extrowerk>
Why would anybody attempt to use -j?
<ermo[m]>
Take it up with the GNU developers.
<extrowerk>
ermo[m]: i see.
<ermo[m]>
This is how it works. If you want to change that, contact the people who maintain the GNU make software and make your suggestion.
<ermo[m]>
Be aware that they may not see things from your perspective.
<ermo[m]>
And best of luck =)
<ermo[m]>
Meanwhile, I would just personally work around it.
<extrowerk>
ermo[m]: thanks for the explanation, it seemingly have some use case, but i genetically dislike any g*u made/licensed software, so i don't really care too much about them
<extrowerk>
i am thinking nowadays about how a so called hash* (a bash and terminal replacement) should look like and how it is supposed to work. (* hash = HAiku SHell)
<extrowerk>
They say if somebody don't know u*ix they will try to recreate it badly. I think this is a real demotivating opinion.There supposed to be newer form of user-computer communication than this.
<ermo[m]>
meanwhile, you could also simply use e.g. `zsh` or `mksh`, both of which are in HaikuPortS?
<ermo[m]>
But there are tons of shell projects already.
<ermo[m]>
Odds are you will find one of the existing options amenable?
<extrowerk>
ermo[m]: i think i used to work on both of those ports. Also on fish, rc and dash and hatever else we have in the ports tree. While i think fish is thi nicest of those, i still think none of the shells are useble at all for the enduser. Using them is like a timemachine back to the eternal '70s. But i don't have a PDP. I have a trillion times faster computer than anything back then. I expect those cpu cycles to work for me.
<ermo[m]>
`rc`? The Plan 9 shell?
<ermo[m]>
Neat.
<extrowerk>
fish is _currently_ impossible to run on Haiku, because they insist to use locale_t.
<ermo[m]>
sounds like you're waaay ahead of me
<extrowerk>
i have even did the ksh port, i think
<extrowerk>
ahead? in what terms?
<ermo[m]>
You know more about which shells are available in Haiku and have thought about concerns related to existing shells.
<ermo[m]>
I merely dabble in Haiku and use the stuff that's there.
<ermo[m]>
I couldn't find the `rc` shell fwiw.
<extrowerk>
i spent many time with this topic, so it is clear i know the current state of the available shells. I wanted to know them to see what they provide and how to see their shortcomings and man, they seems to be obsolate for everyday human-computer communication.
<extrowerk>
ermo[m]: i have a wip locale port for rc
<extrowerk>
i think it is a port of the plan9 for unix package
<Anarchos>
waddlesplash you would be glad to learn that pxe boot is still working on bare metal
<extrowerk>
ermo[m]: i do ports for Haiku since many years, if you will spend more time with haiku and with some specific ports then you will learn plenty things.
<waddlesplash>
nice
<Anarchos>
waddlesplash i had to tweak one or two things (extend one buffer size, disable io-apic) stuff like that
<waddlesplash>
do submit your changes and I'll see about including them
<ermo[m]>
> if you will spend more time with haiku (...)
<ermo[m]>
I certainly plan to. I'm currently in the process of re-rigging my hardware KVMs to ensure that I also have the ability to switch to a running Haiku OS box for comparison purposes.
<Anarchos>
waddlesplash well i had to disable io-apic in the config/settings/kernel/drivers/kernel , so i don't have to hit spacebar at boot to do it....
<Anarchos>
waddlesplash i could'nt come with another mean.
<waddlesplash>
still, maybe we can figure out why IOAPIC causes problems
<waddlesplash>
but if there are missized buffers then that's the place to begin
<Anarchos>
waddlesplash ioapic needs acpi , and do search-module for it, which is forbudden while gkernelStartup is true
<Anarchos>
waddlesplash the buffer was totally unrelated to the ioapic problem
<waddlesplash>
yes, I figured
<waddlesplash>
search-module for ACPI shouldn't occur. it should already have been loaded
<Anarchos>
waddlesplash i just could'nt understand , why in a normal setup (id non pxe) the ioapic can initialize
<waddlesplash>
Anarchos: because ACPI is pre-loaded, or at least should be.
<Anarchos>
waddlesplash i added acpi in uild/jam/images/NetBootArchive, i don't know if the order of the addons is relevant or not
<ermo[m]>
extrowerk: nice =)
<waddlesplash>
it shouldn't be. however it may be needed to be placed in other locations
<ermo[m]>
And now please stop pinging me! ^^
* ermo[m]
is going afk to read a bit
<waddlesplash>
if you got things to work in QEMU then I can take a look myself
<Anarchos>
waddlesplash yes, netbootarchive seems to be rather old for an image definition
<waddlesplash>
if you push whatever changes you have already
<waddlesplash>
indeed
Atomozero has joined #haiku
<Anarchos>
waddlesplash i can load haiku-netboot in qemu, but i don't know how to make qemu communicates via network, to reach the remote disk server on the host
jmairboeck has quit [Quit: Konversation terminated!]
<Anarchos>
waddlesplash but i can confirm that in a gcc2hybrid setup, i could boot by pxe, and after that install haiku on hard drive from the remote image :)
<Anarchos>
which i am rather proud of !
<extrowerk>
waddlesplash: during using Haiku my usb mouse got missing in action. unplugging the usb dongle (it is wireless) and replugging it result in the following log in the syslog: "KERN: usb_hid: no handlers for hid device"
<extrowerk>
does it rings any bell?
<waddlesplash>
weird
<waddlesplash>
any other messages?
<waddlesplash>
it means that the device is reporting itself as something other than a mouse, probably, and usb_hid doesn't know what to do with that
jason123santa has left #haiku [Error from remote client]
<extrowerk>
waddlesplash: KERN: usb_hid: error waiting for report: General system error
<waddlesplash>
yeah, that's the real error
<waddlesplash>
reboot?
<waddlesplash>
or ... hmm we shouldn't even get that error
<Anarchos>
waddlesplash during my attempts i encounter the very rare 'boot loader death land' :)
<Anarchos>
when i compile the pxe loader (or was it haiku-netboot ?) with debug symbols
<waddlesplash>
extrowerk: ah, the device was probably removed
<extrowerk>
waddlesplash: i pulled the dongle and inserted into a different usb port to see if one usb port got messed up or all of them, but seemingly all of them.
<extrowerk>
but i pulled the dongle only after it went unresponsive.
<extrowerk>
i have this: 046d:c52b /dev/bus/usb/0/0/0 "Logitech, Inc." "Unifying Receiver" ver. 2411
<extrowerk>
it is paired with a Logitech MX Anywhere 3