<phschafft>
was thinking about GUI designs yesterday.
<Jookia>
yeah?
<phschafft>
diffent styles of how a GUI could be designed, and how things might interact.
<phschafft>
was wondering about windows with subwindows.
<phschafft>
and that got me thinking about when running a virtual machine on a host how the user interface of the guest should interact with that of the host.
<Jookia>
i'm not a very big GUI fan. most my GUI interest would be in trying to expose it as a command line :D
<phschafft>
what I very much hate is how bad mouse support is after all those years on general TUI applications.
<phschafft>
like, why I can't just click into a spot on a bash (-> readline) prompt to continue editing at that location.
<phschafft>
I mean, in GUI applications a single click into a text entry area will set your cursor position. why is that not universal on TUIs?
<Begasus[m]>
neat, fixed the icons for Artikulate :)
lesderid has quit [Remote host closed the connection]
lesderid has joined #haiku
andreaa has joined #haiku
dovsienko has quit [Quit: Leaving]
<andreaa>
(^.^)/
lesderid has quit []
lesderid has joined #haiku
<andreaa>
^ ^
<andreaa>
O.O
andreaa has quit [Quit: cu later]
Begasus has joined #haiku
Begasus is now known as Guest26323
Guest26323 is now known as Begasus
<Begasus[m]>
141 packages so far for gear 25.08.1, still building (some of the major ones still in the pipeline) :)
moparisthebest has quit [Ping timeout: 480 seconds]
zelectric has joined #haiku
politebot has joined #haiku
moparisthebest has joined #haiku
Anarchos has joined #haiku
<Anarchos>
hello
BrunoSpr has joined #haiku
<Jookia>
phschafft: i'm not too sure where to start, it's a very frustrating topic for me. i don't want to give an angry or upset answer. GUIs are great for most people but absolutely terrible for a very small percentage of people with disabilities with vision or motor functionality. that's probably my primary gripe
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<phschafft>
Jookia: if you feel like it doesn't belong into the channel OR you fear that you might not hit a friendly tone, feel free to send it to my query.
BrunoSpr has quit [Quit: Vision[]: Ich wurde eingeweicht!]
<Jookia>
i mean, i think i'm not the most optimist person about this topic and have fairly extreme views about this. i just think my core problem is that GUIs have become mandatory for interacting with computers instead of a nice to have
bbjimmy has joined #haiku
<Jookia>
i enjoy a good GUI, i don't think GUIs need to go away or they're inferior. i just think maybe we need to tone it down a bit
* phschafft
seems to be a person somewhere in the middle. plus into the position that he can make things better at least for the software he writes himself. ;)
<Jookia>
there's a semi-forgotten tribe of blind linux users who use linux with just the command line as the GUI has basically fallen apart for them over the years
<Jookia>
i got an email today from someone asking for a better search engine because google is blocking lynx
<phschafft>
hm.
<Jookia>
i use to think this was an indictment of the state of linux GUI screen reader support
<Jookia>
but if you think about 'how would i use a computer if i couldn't see', your first thought isn't really 'i would love to navigate through a maze of arrows and buttons and fields to accomplish my computer tasks'
<phschafft>
to me part of that just sounds like GUIs should be improved by having more sematic infrmation around so software like screenreaders can work nicely.
<Jookia>
maybe? i don't know how much i agree with that any more. screen readers are really bad to use
vdamewood has joined #haiku
<phschafft>
I mean, for example if you have a simple dialog, then the text and buttons could be marked as such and basically the screenreader could read out the text, and then allow you to select a button without the need to navigate it.
<phschafft>
just like those very old text menus: for ok press 1, for abort press 2.
<phschafft>
with the buttons basically not being buttons but rendered actions. and those actions could also be rendered in a different way.
<phschafft>
such as a list.
<Jookia>
there's a lot of different ideas and things to try to make a more accessible world
* phschafft
nods.
<Jookia>
it's a very under-researched topic because all of the effort has been put in to allowing to navigate existing applications
<phschafft>
are you blind (or similar) yourself?
<Jookia>
no, i have 20/10 vision
<phschafft>
I mean over at the Icecast channel we always have this fun when the deaf and the blind discuss media streaming.
<Jookia>
i don't have any good solutions here outside probably inventing a new type of accessibility thing
* phschafft
nods.
<Jookia>
but even then i'm not sure any of them could beat the efficiency of a command line
* phschafft
enjoys inventing new wheels... em...
<Jookia>
it's not because the command line is special or anything, just because its kind of a discoverable environment
<phschafft>
is it?
<Jookia>
if you know commands you can type command --help or man and get a lot of documentation/an overview of how to use a tool fully
<Jookia>
i'm not saying it's a good discoverable environment, but it certainly gets you in to a system of directly talking to your computer
<Jookia>
if you're blind you don't really care what a 'button' is
<Jookia>
but at the same time if you squint, we could probably build GUIs to be usable with some kind of intent or menu system
<phschafft>
hm.
<phschafft>
I personally feel that command line is less discoverable in many parts as GUIs as it is unclear if something is data or structural.
<Jookia>
yes, that's because the command line is text based and kind of trashy
<phschafft>
to attach to your example: some software use -h for help, others use -h to destroy your data.
_whitelogger has joined #haiku
<phschafft>
hm.
<Jookia>
(you'd still see the text for the lines and stuff)
<Jookia>
half the battle is that screen readers make accessibility testing an absolute nightmare compared to having the logic implemented in to the application
<phschafft>
hm.
<Jookia>
if we could agree on having some kind of keyboard/line mode for applications that would be cool. but i'm not sure
<phschafft>
for the input side: I feel like all input devices should be first class, but also all should be optional.
<phschafft>
naturally a keyboard is a good thing if you want to enter text, so is a graphic tablet if you want to draw.
<Jookia>
definitely
<phschafft>
but it would be good if they would all have the same level of support, and with some generic fallback in case they are not around.
<Jookia>
i don't think the problems of accessibility come from having to shrink your application down to a line of text, i think it's that we don't have the tools to help devs do that
<phschafft>
such a fallback might be improved by user settings and (key)bindings and user provided scripting and such. sure.
<Jookia>
if you've ever tried to develop for a screen reader you'd quickly see that it's not a fun, understandable or easy to use process
* phschafft
nods.
<Jookia>
so i'm not sure. maybe i'll propose a TTS api for haiku or something
<Jookia>
i plan to try some experiments on dosbox-x sometime to see if i can make a new GUI for it
<phschafft>
hm.
<Jookia>
but there's also the other argument that applications should expose their data structures as APIs and we should have generic GUIs for things
<Jookia>
i don't think that's a feasible approach or a good one
<phschafft>
I really like to have stock*, like stock dialogs or stock controls.
<phschafft>
in the application you only say 'I want an ok button' (easy for the dev) and all the tools can figure out that this control is a button that follows the defined 'ok' logic.
<Jookia>
yes, semantic UIs like that are good
<Jookia>
these solve another problem: blind people can't make GUIs easily
<Jookia>
being able to say 'ask the user for something' and have the UI toolkit just do that for is very helpful
<phschafft>
one of the things I discussed before in here was that I think there should be a way to set a (window) title not to a string but to the-title-for-a-resource.
<phschafft>
e.g. you say window.setTitleFor(somefile)
<Jookia>
yeah, i don't know what a good solution is. i've spent years angry at this instead of being productive. don't be like that
<phschafft>
and you get some template that renders a title based of that, but software could also inspect and figure out that this window is about that resource.
<Jookia>
haiku being a GUI-based system also has some interesting advantages
<phschafft>
the more stock stuff we have the less points we have to fix to improve things.
<Jookia>
one big problem with screen readers in linux/unix is the console subsystem, boot messages, etc
<Jookia>
you also have to fight your audio stack to work with the VT subsystem and GUI subsystem
<phschafft>
hm.
<phschafft>
(will be back in a few minutes, quick run to the kitchen)
<Jookia>
so i don't know, i'm in kind of a burned out situation. it would be nice to hack on haiku to try and do something about this, but... i don't know how the culture is about it
chilledfrogs has joined #haiku
<zelectric>
chilledfrogs, now that's a nice nick I must admit
OscarL has joined #haiku
<OscarL>
waddlesplash: results of testing +/9637 PS4 + cstates.patch, on Atom N450, AMD E-450, and Phenom II X4: https://bpa.st/J7YQ
<phschafft>
Jookia: with a smaller OS one can actually change things easier.
<waddlesplash>
OscarL: so the E450 actually tries to activate it but it doesn't work for some reason. Oh well
<OscarL>
seems so, yes.
<waddlesplash>
oh actually I have an idea
<waddlesplash>
need to add some more logging for it though
<OscarL>
The Phenom having invariant TSC means that if we had a P-States driver (ie: "acpi_cpufreq") it would have a chance of NOT messing up the system time, right? (/me tries to find some hopium :-D)
kevans91_ has joined #haiku
<Begasus[m]>
done, 165 packages (only need to check which ones failed to build and fix those) :)
<Begasus[m]>
OscarL Hola :)
<Begasus[m]>
For the record, not doing a release for this update, we're good in the depot for them :)
<OscarL>
Aloha Begasus[m]! Keeping yourself busy, I see :-)
<Begasus[m]>
automated update/build, so not really :P
kevans91 has quit [Ping timeout: 480 seconds]
<Begasus[m]>
but got the script fixed so it doesn't reset the REVISION in each recipe for like in dev-libs, and only touch the ones it need to :)
tuaris has joined #haiku
<OscarL>
Begasus[m]: re gcc 13.4.0 + gfortran patches.... did you tried running gfortran by itself and not just from scipy build? (to test for the presence or not of those ugly warnings, testing with the simple "touch foo.f90 && gfortran foo.f90" should suffice)
<Begasus[m]>
OscarL no, only with the scipy build
<Begasus[m]>
nuked my attempts after I didn't see any progress here, I'll best leave it to the experts :D
Anarchos has quit [Quit: Vision[]: Gone battling for revolution !]
<Begasus[m]>
ps, ninja is updated :)
<zelectric>
I wish I were a ninja
<Begasus[m]>
upstream* :)
<waddlesplash>
OscarL: yes for Pstates
<OscarL>
Begasus[m]: were those clean builds (make clean and/or hp scipy -c)? I ask because I kinda wondered if maybe meson was somehow caching the flags.
<waddlesplash>
also, weird thing to try, on the Atom where booting stalls with the new idle driver: try pushing the power button after the stall and see if booting continues
<waddlesplash>
it may just stall again after a bit, only printing a few more lines of log, but it would be an interesting test
BrunoSpr has joined #haiku
<OscarL>
waddlesplash: the one that "froze" after loading the new idle driver was the AMD E-450. and I did tried hitting the power button a couple of times. still froze, so I took a picture of the on-screen debug info.
<Begasus[m]>
cleaned builds
<OscarL>
Begasus[m]: alright. No clue then :-D
<OscarL>
really odd that gfortran thing, as Adrien's last suggestion looks like it *should* work (given that "-Wno-multichar" is removed from HAIKU_CC1_SPEC)
<waddlesplash>
OscarL: it didn't progress further when pushing the power button you mean? ok
<OscarL>
yes, no new lines at all after several pushes.
<OscarL>
thus decided to just screenshot it, and forcibly power down (via holding down the power button for a while). I was kinda hoping to kick it in the acpi events, but seems thing was really frozen :-D
<nekobot>
• korli (07cd7343): llvm21: trigger build on x86…
<Begasus[m]>
k, let's testrun this before pushing
* OscarL
wonders where he left the Atom N2600 "half-netbook".
<Begasus[m]>
lol
<Begasus[m]>
so much space you lost it? :P
<OscarL>
so many (old, or parts of) computers... too bad that even if I could magically add up their CPU power... would still be slow as a snail :-D
<swiftbanana[m]>
Ah, yes, "diy all-in-the-keyboard", I know of that technology
<OscarL>
swiftbanana[m]: :-D
<OscarL>
found it... without RAM :-( (I think I took the 2 GB it had to put in the E-450 all in one).
<swiftbanana[m]>
("all-in-one", eh, moar like "all-in-the-screen" =p)
<OscarL>
indeed. I don't see any keyboard or mice in there, where's the "all" in that, hmm?
* OscarL
slaps some random 1 GB sodimm in the "raspberry Pi 400 we have at home" netbook, and prepares to either laugh, or cry.
* Begasus[m]
smirks
<Begasus[m]>
is that even a word? :P
PetePete has quit [Ping timeout: 480 seconds]
duanin2 has joined #haiku
<Begasus[m]>
k, let's try this calligra-25.08.1 ...
<Begasus[m]>
not in my list for gear* atm :P
Begasus has quit [Quit: Vision[]: i've been blurred!]
BrunoSpr has quit [Quit: Vision[]: Ich wurde eingeweicht!]
<Begasus[m]>
heading out, dogschool, cya peeps!
OscarL has quit [Remote host closed the connection]
<B2IA>
(whitepaperkat) hi
<B2IA>
(whitepaperkat) hi
Aedil has joined #haiku
PetePete has joined #haiku
mmu_man has quit [Ping timeout: 480 seconds]
OscarL has joined #haiku
<OscarL>
welp, that was a trip. The N2600 refused to boot from the sata-to-USB device, but luckly had an install there that I could update to tests waddlesplash's x86_acpi_cstates.
<OscarL>
Seems the N2600 already supports C-States, so the x86_cstates driver get loaded/used (still got some log lines from x86_acpi_cstates).
mmu_man has joined #haiku
<OscarL>
blacklisting the x86_cstates driver, results in the same sympthoms as with the AMD E-450. system freezes after loading x86_acpi_cstates.
<waddlesplash>
hmm
<waddlesplash>
well that's interesting at least, because I had thought the E-450 problems might be AMD-specific
<OscarL>
(that was when trying to run "cpuid > foobar.txt")
<waddlesplash>
oh, that's bad
<waddlesplash>
stack trace, pls
<OscarL>
subsequent runs of that worked OK.
<waddlesplash>
did it only happen on that machine?
<OscarL>
took a picture of the KDL. yes, only in the N2600, and only once.
<Habbie>
once? or once per boot?
<OscarL>
only in one occasion.
<waddlesplash>
OscarL: does KDL shortcut work when machines are hung?
<Habbie>
i should really figure out sysrq on the chromebook
<Habbie>
.. is sysrq remappable?
<waddlesplash>
PRntScrn works too
<waddlesplash>
I fixed that a while back
<Habbie>
ok, i have neither, but this tells me something could be done :)
<waddlesplash>
yeah the keycombo is just programmed into the ps2 driver
<waddlesplash>
you could change it
<OscarL>
That machine has a borked keyboard, and the PS/2->USB adapter I used with a "working" keyboard is not great (sometimes machines seem to have problems initializing it)
<waddlesplash>
OscarL: pushed a new version of the change that may fix init on the other machines. But it will likely just hang there too so don't bother trying it yet
<OscarL>
ok. will try to upload logs/cpuid for the N2600 anyway (syslog has a lot more x86_acpi_cstates lines than on the other machines).
<OscarL>
shall I create a ticket for the KDL around lock.cpp, right?
<waddlesplash>
yes
<waddlesplash>
the real problem however is likely in some other code, unfortunately
<waddlesplash>
this ASSERT just tracks symptoms not causes
<OscarL>
tty driver was involved. (exact command I've tried was: "cpuid ~/Desktop/New\ Folder/CPUID-AtomN2600.txt", which was wrong, because I forgot to put ">" after cpuid :-D)
<waddlesplash>
hm
<OscarL>
will upload image asap.
<waddlesplash>
that may be what got woken up but I doubt it's the cause
<OscarL>
tried reproducing it after the "co"ntinuation, but didn't triggered again.
<waddlesplash>
that assert basically always happens due to races
<waddlesplash>
and the races tend to be very rare
<nekobot>
• korli (f2a566fc): llvm21: trigger build with -j2 on x86
<waddlesplash>
OscarL: if you can get it to happen again, even once, I will prepare a patch you can use to try and track down the culprit
<OscarL>
Guess I can try to see if I can fix the keyboard issues on that machine to make using it less annoying. I don't use it much, as I only have one monitor (and the netbook has no screen :-D), and I don't have a working LAN anymore.
<waddlesplash>
it could happen on other machines with similar performance
<waddlesplash>
it's more likely on nightly builds than beta
<waddlesplash>
... actually it won't happen on beta at all
<OscarL>
then maybe I should run whatever patch you can come up to try and catch it?
<waddlesplash>
problem is that the patch may affect timing too unfortunately
<waddlesplash>
I really should try and figure out a more generic way to catch these problems
<waddlesplash>
unfortunately it's not very easy
<OscarL>
I see. welp... will let you know if I see it again, at least.
<OscarL>
Seems the Atom N2600 has invariant TSC, unlike the older N450/N455.
<waddlesplash>
right, so it can use the newer cstates module
<OscarL>
waddlesplash: this is the "previous_syslog" from the N2600, with x86_cstates disabled at boot, trying to forcing it to use x86_acpi_cstates (frozen after last line): https://0x0.st/KbV_.txt
<OscarL>
https://0x0.st/KbVp.txt <<< and this one is a clean syslog where x86_cstates gets loaded/used, but still has lots of x86_acpi_cstates lines
mmu_man has quit [Ping timeout: 480 seconds]
<OscarL>
funny because it has "KERN: found cpuidle module: power/cpuidle/x86_acpi_cstates/v1" at some point, but the driver isn't actually loaded.
<waddlesplash>
yeah it gets discarded if the other module succeeded
<OscarL>
cpuid dump for the Atom N2600, in case it might be interesting: https://0x0.st/KbVt.txt
mmu_man has joined #haiku
<waddlesplash>
OscarL: uploaded new patchset with a fix for the "FFH" method
<OscarL>
to try first on the E-450 (and maybe then on the N2600 if I disable x86_cstates), right?
<waddlesplash>
yes
PetePete has quit [Ping timeout: 480 seconds]
<OscarL>
alright... let me prepare things up (and get a coffee first :-D).
<swiftbanana[m]>
Is it hard to do a rebranded build of Haiku? I am writing a retrotech-themed mystery game for a group of friend, and I am contemplating the idea of using a modified (renamed, reskinned, etc.) Haiku as the diegetic fictional OS
PetePete has joined #haiku
<waddlesplash>
it will be difficult to strip out all identifiers
<waddlesplash>
but since the Haiku name/logo is a trademark, the default builds don't use it
<waddlesplash>
so if you build from source there won't be any logos already
<waddlesplash>
but of course it'll still have all the regular icons and the like
<waddlesplash>
replacing all that would not be difficult exactly, just very annoying
<waddlesplash>
tedious, mostly
<swiftbanana[m]>
Yeah, of course ah ah
<swiftbanana[m]>
Well, we’ll see later with my co-organiser if we want to go to the "icons" level of details or not ah ah =‘)
<waddlesplash>
well, guess that answers whether it ever happens in the wild lol
<waddlesplash>
well, I'll have to investigate how to fix that
<OscarL>
KDL still active in the N2600 at the moment, in case you want me to type something in there.
<waddlesplash>
nope
<waddlesplash>
It'll take me some time to come up with a potential solution though
<waddlesplash>
and will involve changing the bootloader
<OscarL>
k, killing it then, and off to test the N450, just in case.
<OscarL>
N450 also got stuck right before the rocket icon (forgot to enable on-screen debug... rebooting to enable that).
<OscarL>
"can't find matching cpu_ent for acpi cpu 2" (same line repeats with cpu 3, 4, 5, 6, and 7)
<waddlesplash>
is that how it ends?
<OscarL>
will take a picture. Ends after: "using x86 ACPI idle" / "found cpuidle module: power/cpuidle/x86_acpi_cstates/v1" / "ignoring MTTRs due to PAT support".
<waddlesplash>
okay
<waddlesplash>
probably just stalling at the same location then
<OscarL>
cpu fan at 100% :-D
<OscarL>
at least does more than before :-) (didn't tried to load it in the N450)
<OscarL>
this "bluetooth sharing" is sloooooow. :-(
<OscarL>
alright... let me cook things up a bit :-)
<OscarL>
no wonder the build feels slower.... the CPU is getting throttled down due to temps :-(
Chai-T-Rex has joined #haiku
ChaiTRex has quit [Ping timeout: 480 seconds]
<Halian>
:(
<OscarL>
alright... let's see where then N450 blows out now :-D
<OscarL>
waddlesplash: same output as before, but we got a new line now before boot stops: "acpi_cstate_ffh_enter: cpu 0 state 1". (and CPU fan turned into a 747 at lift off)
<waddlesplash>
well, we get there at least
<waddlesplash>
doesn't make sense that we just hang at that point
<OscarL>
will test on the E-450 now.
ADS_Sr has joined #haiku
<waddlesplash>
OscarL: itll be the same for cstates
<waddlesplash>
but if you can try and reproduce the assert failure, that would be good
<OscarL>
E-450 freezed in the same place as before, no new lines. Will try on the N2600 (the one that got the assert ealier today).
<OscarL>
enabled ALL safe mode options (because last attempt also turned the video signal off... monitor was showing "incompatible mode", and a single press of the power button turned machine off).
<OscarL>
image still up, but froze after "loaded driver /boot/[...]/realtekwifi". This makes no sense.
<OscarL>
can boot normally from the hdd install (hrev59042+1+dirty from early in the morning... I think it was PS4).
<waddlesplash>
so, it's the old instability again
<OscarL>
dunno, never had this loss of video signal so consistently.
rexbinary has joined #haiku
<OscarL>
rebooted the N2600 again, no issues with hrev59042+1+dirty (possibly PS4) so far.
<waddlesplash>
maybe related to the other patch you applied somehow?
<waddlesplash>
that patch should only cause KDLs though
<waddlesplash>
it shouldn't cause any other problems
<waddlesplash>
and it should only turn things that were "silent" issues before, into KDLs
FreeFull has quit []
<OscarL>
do PS9 and/or the "kernel-thread-unblock-assert.patch" require an updated haiku_loader package? (I'm still using the one I got from the build checker bot)
<OscarL>
will try downloading the new one the bot made for PS9, just in case.
Halian has quit [Ping timeout: 480 seconds]
<OscarL>
still the same thing (no video signal after loading vesa and some wlan stuff) with updated haiku_loader, and after disabling both x86_acpi_cstates and x86_cstates, just in case.
PetePete has quit [Ping timeout: 480 seconds]
PetePete has joined #haiku
<waddlesplash>
OscarL: no, didn't need a loader update in the end
<waddlesplash>
really weird that it has no video signal
<waddlesplash>
I don't know if that's just random or if something I changed actually affected that
<waddlesplash>
is it actually "no signal? Or just black?
<OscarL>
but seems system was still on. ctrl+alt+del didn't auto rebooted, and a gentle press of the power button did turned the machine off, but not immediatly as before.
<waddlesplash>
really weird
<waddlesplash>
too many problems to try and debug here :(
<OscarL>
I'm using an external monitor via VGA port. Monitor reports "incompatible mode", as if signal was just messed up / out of range.
<waddlesplash>
try blocklisting the VESA driver
<waddlesplash>
and see if it falls back to framebuffer instead
janus has quit [Quit: janus]
Halian has joined #haiku
<OscarL>
gets into a loop of "app_server: Failed to initialize virtual screen configuration: No such file or directory" and "ieee80211_notify_scan_done"
<OscarL>
(with both vesa and vesa.accelerant disabled at boot menu)
<waddlesplash>
ah well
<waddlesplash>
but you said the N2600 had other strange instabilities before
<OscarL>
switched to UEFI mode, but it booted the install from the HDD (PS4) and not the one from the pendrive.
<waddlesplash>
so, maybe this is just related to that
<waddlesplash>
you may need to switch boot partitions in the bootloader for that
<OscarL>
I have seen weird things in this machine, yes. Got the scary red text from boot loader "KDL" once too.
whitepaperkat has joined #haiku
whitepaperkat has quit []
whitepaperkat has joined #haiku
whitepaperkat has quit []
whitepaperkat has joined #haiku
absyn-th has joined #haiku
erysdren has joined #haiku
<phschafft>
erysdren: !! my saviour is here!
* phschafft
offers erysdren a dark chocolate cookie with a hint of orange as bribe.
<OscarL>
ps9+patch definitively a no go on the N2600. Will try with PS9 alone, I guess.
mmu_man has quit [Ping timeout: 480 seconds]
<glassnerves>
do you guys use Haiku on bare metal daily?
<Habbie>
i do some days
mmu_man has joined #haiku
<Habbie>
i recently went camping and took nothing but my phone and my haiku laptop
<OscarL>
waddlesplash: seems the N2600 just hates booting from USB for some reason. Same issue with plain PS9 (and no "kernel-thread-unblock-assert.patch"), but once I installed those packages in the HDD install... it booted without problems.
<OscarL>
so I guess I will re-do PS9+kernel-thread-unblock-assert.patch, and install in the HDD next.
<phschafft>
when I used to be camping I took... a tent with me. not eletronics. they don't help against the rain much.
<Habbie>
i did also bring a tent :)
whitepaperkat has quit [Ping timeout: 480 seconds]
<OscarL>
I'm guessing that's why I had so much issues with this machine in the past... (trying different boots from USB), and the one from hdd... just seem to work fine.
whitepaperkat has joined #haiku
<swiftbanana[m]>
If you can’t shelter yourself from rain with electronics then you didn’t bring enough electronics
<OscarL>
waddlesplash: booted PS9+kernel-thread-unblock-assert.patch from the HDD on the N2600... and seems I managed to trigger that same ASSERT waiter.thread = __null again.
<OscarL>
sc looks pretty much the same as before (minus diffs in tread ids and such).
<OscarL>
waddlesplash: tried "co", now got: "PANIC: recursive_lock 0xffff... unlocked by non-holder thread!"
<waddlesplash>
no other KDL?
<waddlesplash>
the second one sounds interesting
<waddlesplash>
but I'd hoped we would get a different KDL for the first one, my new asserts should have uncovered the cause
<whitepaperkat>
who keeps logging in as my user name
<whitepaperkat>
meh it could be just my crap connection
<OscarL>
nothing else so far... just booted, tried "cpuid ~/Desktop/New\ folder\foobar.txt" (again without the ">" after cpuid as earlier today) and triggered the assert right away.
<waddlesplash>
OscarL: it's very odd the assert is occurring without triggering the new ones
<waddlesplash>
but at least it's reliable I guess
<waddlesplash>
so hopefully I'll be able to come up with some other patch. that'll have to wait till later though, out atm
<OscarL>
next "co" panics with _mutex_unlock() failure: thread 295 is trying to release mutex 0xffff.... (current holder 265)"-
<OscarL>
odd...
<OscarL>
next "co" sent me back to desktop for a brief moment... and then the screen turned all black :-/
<OscarL>
heh, seems it was just the screen_blanker, damn :-D
* OscarL
disables the screensaver.
<OscarL>
as before, repeated calls to cpuid work without issue. will try to trigger it again after a reboot.
<OscarL>
"PANIC: never can get here", hahaha.
<waddlesplash>
OscarL: the recursive lock first KDL means state is corrupted so the other Kdls are uninteresting
<waddlesplash>
no need to upload screenshots
<OscarL>
you don't tell me what I can or can't do!!!! :-D
<OscarL>
(talking to the KDL)
whitepaperkat has quit [Quit: Konversation terminated!]
whitepaperkat has joined #haiku
politebot has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
this is the weirdest KDL I've seen so far.
<OscarL>
cpuid after a reboot first worked, but a second call sent me in to this "never can get here" panic :-D
<OscarL>
tried "co", now got ASSER FAILED on kernel/treadh.cpp:1211 "thread->state == TREAD_STATE_FREE_ON_RESCHED"-
<OscarL>
on next boot... had to call cpuid repeatedly, but in the end hit that same water.tread == __null assert failure. At least seems easy enough to trigger.
<OscarL>
note to self... don't enable Power saving mode on the N2600.
mmu_man has quit [Ping timeout: 480 seconds]
mmu_man has joined #haiku
linuxuser01 has joined #haiku
<OscarL>
enabling "Power saving" seems to bork the input server at some point (power button still triggered a clean shutdown).