dpirate has quit [reticulum.oftc.net dacia.oftc.net]
dpirate has joined #haiku
HaikuUser has joined #haiku
Maturi0n_ has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
HaikuUser is now known as KENZ
OscarL has joined #haiku
<OscarL>
pairisto[m]: simple one in "new" style API: add-ons/kernel/drivers/power/acpi_thermal/acpi_thermal.c, simple one in legacy driver API: add-ons/kernel/drivers/misc/poke.cpp
Maturi0n has quit [Ping timeout: 480 seconds]
AlaskanEmily has joined #haiku
KENZ has quit [Quit: Vision[]: i've been blurred!]
KENZ has joined #haiku
<pairisto[m]>
thanks guys
<pairisto[m]>
* thanks guys!
KENZ_ has joined #haiku
<OscarL>
pairisto[m]: np! just beware that if you use the new style API, you'll need to reboot if you recompile/reinstall the driver you're developing/testing.
<OscarL>
legacy API drivers seem to be properly reloaded if you remove and reinstall them.
KENZ_ is now known as KENZ__
KENZ__ is now known as KENZ___
KENZ___ has quit []
<OscarL>
with the "new" ones... your device properly dissapears from /dev/ (if you published something) if you remove the binary, but if you replace it with a new one (without rebooting), the OLD code is still being used, instead of the one with your latest changes.
<pairisto[m]>
is there a preference when developing drivers?
<OscarL>
(I got bitten by that pretty hard :-D)
<OscarL>
pairisto[m]: I'm not the best person to answer that :-/
<OscarL>
I would assume the new API is the prefered one, but for my use cases, I stick to the older one (easier for me to test things out... as I suck at programming, I do that a lot :-D)
<OscarL>
Be sure to reach to your mentors via email. I don't think ScottMC or korli visit here too much.
<pairisto[m]>
forgot I didn't send that out to the mailing list, thought I did and no one replied 0_0
tuaris has quit [Read error: Connection reset by peer]
<Guest1520>
Is it possible to set a keymap up so some keys move when Ctrl/alt/meta are held down?
<Guest1520>
The goal is Dvorak with qwerty shortcuts (Ctrl+c in the same place)
<OscarL>
Guest1520: I've just tried with the Keymap preflet... first I typed "c" (no quotes) on the "Sample and Clipoard" input box....
<OscarL>
then I hold one of the modifiers (physical ALT, but Ctrl on Haiku, as I have them switched around)... and with that hold down, I drag-and-dropped that "c" from the input to the place where the "x" key is.
<Guest1520>
But then it doesn't change back when you let go of the modifier key right?
<OscarL>
result: I get an x if I type just "x", and I get a "c" if I do "Ctrl+C" (alt+c on my keyboard).
<Guest1520>
What you describe works for shift but not for ctrl
<OscarL>
at least that's what happens on that "test" input box.
<augiedoggie>
are you in windows/linux shortcut mode?
* OscarL
is.
<augiedoggie>
hm, dunno
* augiedoggie
goes back to playing with tree-sitter
<Guest1520>
More poking around it looks you can assign different keys for what haiku calls option and control but not for command
<OscarL>
I see... trying to do the same with CMD (physical CTRL), it replaces the original letter in both hold/no-hold states :-( (tried replaceing "<" with "O")
orealis has quit [Read error: Connection reset by peer]
<OscarL>
not sure if that's a Keymap preflet bug, or a wider limitation.
orealis has joined #haiku
<OscarL>
as holding "cmd" DOES changes the visible keys for the numeric keypad, but not for the "main" portion of the keyboard.
orealis has quit [Read error: Connection reset by peer]
orealis has joined #haiku
<OscarL>
Weee! we have Python 3.12.0.beta2 running on Haiku :-D
<OscarL>
"Python 3.12.0b2 (main, Jun 10 2023, 00:07:20) [GCC 11.2.0] on haiku1"
* augiedoggie
is amazed that OscarL isn't sick of dealing with python
<OscarL>
heh :-D
<augiedoggie>
i hate touching python recipes
<OscarL>
one of these days, I'll burn them all!
<waddlesplash>
we should really make some kind of scriptlet/template for all that gunk that you see in every recipe
<waddlesplash>
scriptlets that live in the HaikuPorts repo, that is, instead of in HaikuPorter
<OscarL>
I'll try to help with that, but first I need to gain more confidence making changes and understanding the different use cases. I'm slow and rusty, so... might take a while :-D
<Guest1520>
New question, first time using haiku on real hardware, an old Asus eee PC, I've shut it down 3 times so far and only once did it turn off successfully
<Guest1520>
The other times the screen goes black but it stays on until I hold the button for 10 seconds, after waiting 15 minutes or so to make sure it's not shutting down normally
dby has joined #haiku
<OscarL>
weird, on my Atom based netbook, Haiku is the only OS that reliable turns it completely off. Maybe a wonky BIOS/ACPI thing?
<OscarL>
you can try playing with some boot options, like disabling ACPI, or APM, etc. Hold SHIFT when booting up to get to the boot loader menu.
<erysdren>
on that eMachines PC from about 2013 it rarely shuts off with Haiku for some reason. will check the bios later
bbjimmy has joined #haiku
bbjimmy_64 has joined #haiku
bbjimmy_64 has quit []
bbjimmy_64 has joined #haiku
<Guest1520>
Yep disabling acpi works... How do I make that permanent?
<OscarL>
Setting "acpi false" in the kernel setting file.
<OscarL>
kinda weird that the "kernel" file is under kernel/drivers, but... ¯\_(ツ)_/¯
<OscarL>
speaking of weird... these pystone.py results are nuts: https://bpa.st/GR534
bbjimmy_64 has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
(first one is 3.8, actually)
<Guest1520>
Thanks for all the help
<OscarL>
no problem! :-) Happy "Haikuing"
<OscarL>
"Note The default start method will change away from fork in Python 3.14." then I guess my change to make it use "spawn" instead of "fork" might be not so bad after all. (build was failing for 3.12 otherwise, and multiprocessing tests always gave problems with fork/forkserver)
<OscarL>
"--enable-optimizations" seems to hurt more than it helps, at least on this pystone.py benchmark (tested on both 3.11 and 3.12). Good for me, as not enabling optimizations, cuts build times from ~35 to <8 minutes.
<Begasus_64>
nice OscarL :)
<OscarL>
I might just disable them for 3.10.12, if I get similar results.
<Begasus_64>
giving up on swift ... too many hurdles :/
<OscarL>
disable it on a PR, and see if someone steps up :-D
<Niklas[m]>
<x512[m]> "https://discuss.haiku-os.org..." <- I just came here to report exactly the same. Hm, that's strange
<nielx[m]>
I am trying to give it a kick
<Begasus_64>
OscarL, been thinking about it, same for fife, can't be installed, can't get it to build ...
<nielx[m]>
well except for users trying to create posts or comments that did not reach the server before it went down
<nielx[m]>
Discourse is restarting
Begasus_32 has quit [Quit: Vision[]: Gone to the dogs!]
<Begasus>
that was fast PulkoMandy :) (just did a quick check on the cmd's, probably could use some further patching, but at least it's there now, instead of collecting dust on my repo) :)
Begasus_32 has joined #haiku
<PulkoMandy>
I think pairisto[m] can make use of this, so let's have it ready :)
<Begasus>
yeah, was waiting for maybe him creating a PR himself :)
<nielx[m]>
And the forums are back!
<OscarL>
time to complain on the forums about the lazy admins! :-P (j/k, j/k)
<KENZ[m]1>
hello, folks. I finally found here to touch you.
<OscarL>
KENZ[m]1: hello there!
* OscarL
applauds KENZ[m]1's efforts on porting FF, and wishes him the best of luck on that endeavor!
<Begas_VM>
Hi KENZ[m]1 :)
<Begas_VM>
+2 nielx[m] :)
<KENZ[m]1>
OscarL: thanks a lot. Firefox build system heavily relies on python, thanks your work on it.
<Begas_VM>
nuking the build, let te buildmaster take care of that one (done enough stress builds the last days) :)
<KENZ[m]1>
Begas_VM: hi! I saw you tackling updating Swift on Haiku. I'm looking for it because latest Swift improves C++ interop much. I want see how it works well on Haiku API.
<KENZ[m]1>
(I'm afraid I just saw you gave up though...)
<Begas_VM>
pushed what I had so far in a PR to disable it KENZ[m]1
<Begas_VM>
for llvm part there should be another patch done in one file, not that hard
<Begas_VM>
but still ran into the dependency for python2.7 later on :/
<andreasdr[m]>
KENZ: Good morning, thanx for the upcoming FF port <3 :)
<Begas_VM>
there is a "disabled" 4.* recipe intree, maybe you could have a look there (or jump to 5.* straight ahead) :)
<Begas_VM>
hi andreasdr[m] :) (peeps have been talking about FF for about as long as BeOS was around) :)
<OscarL>
I used to run BeZilla from a ram drive, to speed up its startup time, lol.
<KENZ[m]1>
Begas_VM: oh, I misunderstood you on IRC log. sorry, sorry it will be disabled for now..
<KENZ[m]1>
andreasdr: good morning, I found there are (at least) some to fix on my side (non-EGL part). Working firefox is not on my sight yet...
<KENZ[m]1>
But I don't lose my motivation working on it. So please wait for long.
<Begasus>
fixing things in the process is nice too KENZ[m]1 :)
<KENZ[m]1>
Begasus: Yes, I end up digging the Debugger Kit because spending time on it is the best to do...
<Begasus>
yeah saw it, great to get some work done (although I wouldn't know where to look for)
<Begasus>
one of the missing peaces with lazarus iirc
<KENZ[m]1>
Waiting hours for my RAM and SSD swapping data is waste of my time... I need to speed up it for final result.
<KENZ[m]1>
Maybe it helps debugging other large scale projects, maybe LibreOffice, WebKIt?
<KENZ[m]1>
s/Maybe it helps debugging other large scale projects, maybe LibreOffice, WebKIt?/Maybe it will help debugging other large scale projects, maybe LibreOffice, WebKIt?/
<Begasus>
there is some loop in you comments KENZ[m]1 :)
xet7 has joined #haiku
<Begasus>
OscarL, no luck in getting further with mozc?
<OscarL>
until last time I've tried, no. Be my guest!
* OscarL
hands it over to Begasus.
<Begas_VM>
grabbed any patches from other OS to try? (or just clean source)
<OscarL>
clean sources only.
<Begas_VM>
ah k, will give it a spin
<OscarL>
+1
<OscarL>
I don't want to point fingers but... korli broke it.
<OscarL>
:-P
<OscarL>
mmm, python build log too large, and gets cut off from Terminal :-/
<OscarL>
is there a way to make Terminal/bash keep a larger buffer?
<Begasus>
patches welcome :P
<Begasus>
cmd > filename ? :)
<OscarL>
funny guy.
<OscarL>
guess I'll resort to that if there's nothing better.
<Begasus>
well ... it should work :P
<Begasus>
ps, don't think we need cmd:g++ for targetAchitecture (x86_gcc2 is marked broken anyway)?
<Begasus>
don't think we need the portPackageLinksDir variable too
<Begasus>
hmm ... first one we probable do need (being 32bit)
<OscarL>
After changing CFLAGS to BASEFLAGS (following a comment on ./configure), adding -O3 to OPT, and disabling "--enable-optimizations" actually makes 3.10.12 faster. Seems we were running with the default -O0 before, even when "--enable-optimizations" was used :-/
bjorkint0sh has quit [Remote host closed the connection]
<OscarL>
From Python's ./configure: "XXX Shouldn't the code above that fiddles with BASECFLAGS and OPT be merged with this chunk of code?"
bjorkint0sh has joined #haiku
<OscarL>
no wonder I ended up with a debug build the other day, just by removing `export OPT=""` from the recipes :-D
tqh has joined #haiku
gouchi has joined #haiku
gouchi has quit [Remote host closed the connection]
tqh has quit [Quit: Leaving]
<OscarL>
354 tests OK. 35 tests failed. 20 tests skipped. Good enough for me.
<Begas_VM>
PulkoMandy (from the arm_* package): provides: arm_none_eabi_newlib_x86 = 2.1.0
<PulkoMandy>
The recipe clearly provides 4.3.0 here…
<Begas_VM>
or that one didn't get build on buildmaster?
<OscarL>
Mmm, after running Python's test... some how the following dir appeared: haikuports/dev-lang/python/work-3.10.12/sources/system/packages/administrative
bjork1intosh has joined #haiku
bjorkint0sh has quit [Ping timeout: 480 seconds]
mmu_man has quit [Ping timeout: 480 seconds]
gouchi has joined #haiku
<Begasus>
after tests the dir isn't cleaned?
<OscarL>
notice the location, it's inside "sources" instead of one level above (as it should)
<Begasus>
ah right
<OscarL>
or inside boot/, but not parallel to Python-3.10.12. Something got screwed :-D
mmu_man has joined #haiku
<Begasus>
you broke it? :P
<OscarL>
nah... This test suite is bonkers.
<OscarL>
being able to repeat test runs without hanging, or borking the system is the objective for me (at least it now doesn't requiers multiple gigabytes to run :-D)
<OscarL>
then we can see about what fails, and what can we fix.
<OscarL>
for now, this tells me that -O3 didn't break things compared to older builds, so... that's good.
<Begasus>
progress :)
<OscarL>
"test_python_worker_129869æ" <<< I remember seeing this last year. That "æ" seems suspicious (and probably the cause of it not finding that "test_python_worker_129869" temp dir.... wonder where that "æ" comes from).
tqh has joined #haiku
<OscarL>
(several test fail similarly)
<Begasus>
good luck tracking that
<OscarL>
yeah... will probably need lots of coffee, and a good sleep first, before trying to find that one :-)
<Begasus>
no luck on mozc here (tried fidling with newer source too)
<OscarL>
I don't think the upstream mozc repo has Haiku support.
<Begasus>
nah :/
<OscarL>
and figuring out how to update hanya's version... seems like a task for hanya :-D
<Begasus>
will add an issue later there
<OscarL>
+1
<Begasus>
not sure he's still active on Haiku :)
<OscarL>
maybe not, but not much else we can do, other than disable the recipe, and let him know about that issue, no?
<Begasus>
well as long as python3.7 is around it could stay?
<OscarL>
doesn't builds anymore, at least on my system, even after re-enabling six for 3.7
<Begas_VM>
it's still available in the depot
<OscarL>
can't mark the recipe disabled without doing a rev-bump?
<Begas_VM>
it will vanish from the depot iirc
<OscarL>
then I guess we can ignore it for the Python 3.7 removal thing... and let it linger until something else breaks it.
<Begas_VM>
right, still building atm here (97/693)
<OscarL>
I was getting "std::bad_alloc" and segfaults. maybe it needs more memory than I have to build :-)
<Begasus>
atm only using about 750 MiB
<OscarL>
"352 tests OK. 32 tests failed. 24 tests skipped." and the good thing is I didn't have to kill any hanging test.
<Begasus>
better compared to before?
systwi has joined #haiku
<OscarL>
yup.
kinkinkijkin has joined #haiku
<Begasus>
+1
<OscarL>
well, had to skip 4 more test, but... at least it is more well behaved.
jjido has joined #haiku
<Begasus>
wouldn't be supprised if those are disabled on other OS's too
<OscarL>
Also... no more "æ" issues. That seems to happen only when setting the tests' "--tempdir" to the RAMFS under /system/var/shared_memory/
<OscarL>
no idea if the issue is with ramfs, haiku, or python there :-P
selfish has quit [Read error: Connection reset by peer]
systwi_ has quit [Ping timeout: 480 seconds]
selfish has joined #haiku
<OscarL>
wanna give "scipy-1.6.3" a go? I know you love sci stuff !
* OscarL
hides
jjido has quit [Quit: My laptop has gone to sleep. ZZZzzz…]
townsfolkPravda has joined #haiku
<Begas_VM>
still hunting this arm thingy :)
<Begas_VM>
newlib should be good with a small change
bjork1intosh has quit [Remote host closed the connection]
bjork1intosh has joined #haiku
<Begas_VM>
and mozc still building on the other laptop
<Begas_VM>
what's up with scipy?
<OscarL>
needs to provide "non _x86" package name.
<OscarL>
I guess we could just make the change and let the buildmaster deal with it.
<Begas_VM>
ah, could have a look, maybe not today (should dive into those non x86 provides at one time)
<OscarL>
no hurry.
<Begas_VM>
not rushed :P
<OscarL>
libreoffice 6.4 comes first! :-P
<Begas_VM>
lol
<OscarL>
"bzr (Python 2.x)" sigh... who uses that stuff anyway? :-D
diver1 has joined #haiku
Diver is now known as Guest2634
diver1 is now known as Diver
tqh has quit [Quit: Leaving]
Guest2634 has quit [Read error: Connection reset by peer]
<Begas_VM>
you in a bit OscarL :P
<Begas_VM>
bugger, other laptop overheated :/
<OscarL>
ouch.
<Begas_VM>
yeah, almost at 200 objects ...
<Begas_VM>
temp is too high atm here
<OscarL>
place laptop in freezer, or move further north :-P
<OscarL>
cmd:${targetU}_ld vs cmd:${targetU}_ld$secondaryArchSuffix... I think my "fake_hp" would have been good enough to show that without building the recipe.
<OscarL>
I should clean it up and share it with you one of these days.
<Begasus>
yeah! would be helpfull to see if I can trace those spots easier
Begasus_32 has quit [Ping timeout: 480 seconds]
<OscarL>
aaaand, now the test run hangs on its last test, sigh... Enough Python for one day.
<Begasus>
yeah, guess you should take a break :)
<OscarL>
getting sleepy already, near 10 AM here :-D. Off I go. Later Begasus!!! Have a good weekend!
<OscarL>
Anyway, nice to see that mozc only required that small change. No idea why it was failing to build on my PC (either with UnicodeErrors, or just segfaults/std:bad_alloc).
kescher has quit [Quit: Bye]
HaikuUser has joined #haiku
kescher has joined #haiku
<HaikuUser>
hi
<OscarL>
hello HaikuUser.
HaikuUser has quit []
_Dario_ has quit [Ping timeout: 480 seconds]
mmu_man has quit [Ping timeout: 480 seconds]
Begas_VM has quit [Quit: Vision[]: i've been blurred!]
mmu_man has joined #haiku
<Begasus>
Just got back home OscarL (in time to close down) :)
jmairboeck has quit [Quit: Konversation terminated!]
<Begasus>
didn't manage to get a build for scipy, but got 2 new python ones in the pipeline (pyproject-metadata and python-meson)
<Blendie>
But you can get it on all major platforms
<KENZ[m]1>
Sorry for very slow progress...
Blendie has quit [autokilled: This host violated network policy. Mail support@oftc.net if you think this is in error. (2023-06-10 23:38:02)]
<KENZ[m]1>
I think the point of the article is that choose your browser to keep diversity of the web.
<KENZ[m]1>
Firefox is losing momentum more and more, but it is maximum user share as the (non-macOS, non-Chrome) final product, so choosing Firefox is logical to against the dominance.
<KENZ[m]1>
But choosing WebKit browsers is good enough not to accelerate the dominance.