<Skipp_OSX>
I'm trying it your way but it is breaking shortcuts so far
<Skipp_OSX>
I refuse to believe I can't cast a &char to a (uint32*)
<Skipp_OSX>
OMG it was
MisthaLu has quit [Quit: Leaving]
mmu_man has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
mmu_man has quit [Ping timeout: 480 seconds]
<Begasus[m]>
k, on next KDE framework update I need to switch QCH path so it's automatically picked up in QtAssitant :)
hightower2 has quit [Remote host closed the connection]
hightower2 has joined #haiku
vdamewood has joined #haiku
<Begasus[m]>
and move them into a seperate package :)
MisthaLu has joined #haiku
jmairboeck has joined #haiku
Anarchos has joined #haiku
HaikuUser has joined #haiku
HaikuUser is now known as _-Caleb-_
HaikuUser has joined #haiku
<_-Caleb-_>
Hi all
<Anarchos>
_-Caleb-_ hello
_-Caleb-_ has quit []
jmairboeck has quit [Ping timeout: 480 seconds]
<Begasus[m]>
Hi __-Caleb-_ Anarchos
nephele_xmpp has joined #haiku
<nephele_xmpp>
good day
<Begasus[m]>
Hi nephele_
<nephele_xmpp>
how are you Begasus[m] :)
<Begasus[m]>
good here nephele_ , finetuning the frameworks a bit for the next update
<Begasus[m]>
how's it going there?
<nephele_xmpp>
I'm fine. also a bit happy to have resolved my gpu troubles now. Turns out the gentoo wiki was right, for some reason linux overclocks the gpu on it's own. I clocked it down to manufacturer specs and now i have no more crashes .-.
<Begasus[m]>
progress! +1 :)
<nephele_xmpp>
I suppose so xD
<nephele_xmpp>
Anyway, want to rework some of my patches later
<nephele_xmpp>
gerrit was really active this week
<Begasus[m]>
yeah, saw some pushes passing by in the logs also, and comments here and dev channel :)
DKnoto has quit [Ping timeout: 480 seconds]
oevl has quit [Ping timeout: 480 seconds]
nephele has joined #haiku
<nephele>
push the begasus :D
DKnoto has joined #haiku
<Begasus[m]>
:P
<Begasus[m]>
saw I had outdated Qt documentation, need to update from 6.8.1 to 6.8.2 for that :P
<nephele>
Begasus[m]: what does nheko use for notifications
<nephele>
does it draw in d-bus?
<Begasus[m]>
there is a seperate framework they use for that iirc, didn't push that as afaict uses dbus (and don't see any push notifications for it)
<Begasus[m]>
not a framework, it's a library: kunifiedpush
<nephele>
what's that?
<nephele>
i think that is something else. It sais NeoChat uses it :)
<nipos>
That used to work perfectly fine in the past.Now trying to find out when the problems started and what might have caused them
<nephele>
nipos: I doubt you will find anything
<nephele>
you triggered an Assertion in virtual memory code, this probably means there was something wrong with the code already, and now we are catching it earlay
<Begasus[m]>
yeah, not using it when in dark mode, then I use NeoChat
<nipos>
Ah,okay,then I should better try to find out what was wrong with the code instead of going through Git history
<nephele>
maybe try running Keymap with libroot debug?
<nipos>
Should I move the enum inside the class and leave the rest as I have done,or should I better pass the font face enum and re-add the two unrelated bools?
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<nephele>
nipos: oh actually there was a keymap commit jus recently
<nephele>
maybe related
<nipos>
MALLOC_DEBUG=gp fixes the crash,so it's probably an application bug,not a kernel one.But why does it crash the whole OS,not just the app? :O
<nephele>
it does not. Assertions are voluntary crashes :)
<nephele>
that is the "this seems fishy, better if we don't continue here" state
<nephele>
Ideally the VM should not get into this situation, but it can. assertions are there to figure out when the code can get into an invalid state by checking early
<nephele>
for the beta releases we can be a bit less paranoid with assertions
BrunoSpr has joined #haiku
<nipos>
The recent commit you mentioned was done after hrev58590 where the bug was already present.And it still is,I ran an update.
<nephele>
ah okay :)
HaikuUser2 has joined #haiku
HaikuUser2 has quit []
nephele has quit [Quit: Vision[]: i've been blurred!]
nephele has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
HaikuUser2 has joined #haiku
BrunoSpr has quit [Quit: Vision[]: Ich wurde eingeweicht!]
HaikuUser2 has quit []
Nasina has quit [Read error: Connection reset by peer]
HaikuUser has quit [Quit: Vision[]: i've been blurred!]
wicknix has quit [Remote host closed the connection]
<nipos>
I started by commenting out the whole function except the return at the bottom,then I reenabled more and more in it.Enabling that line makes it crash again.
<nephele>
Maybe it does not check is file exists?
<nipos>
I think it's more likely a type mismatch
<nipos>
status is a status_t and file is a BFile.The code does status = SetTo(file); - Looks wrong somehow
<nephele>
but it looks like this takes a path, and not a BFile
<nipos>
Ah,that's it then,thanks
<nipos>
Adding another SetTo that takes a BFile should solve it,I guess
<nipos>
Wait no
<nipos>
The implementation you linkes takes the path,constructs a BFile and calls SetTo again with the BFile,so that must also be defined somewhere
<nephele>
It calls file.SetTo not SetTo
<nephele>
so BFile.SetTo()
<nephele>
below it is for BDataIO, maybe BFile implements that so that variant gets called
<nephele>
maybe you can find something in there that crashes it. You can set breakpoints in the debugger instead of deleting lines though, and then single step
<nipos>
> It calls file.SetTo not SetTo
<nipos>
No,it also calls SetTo in line 77
<nipos>
And it never gets converted to a BDataIO,so I don't see why it would call the function below
<nipos>
Let me verify that: I commented out everything in the BDataIO SetTo except the return B_OK; - If it still crashes,that's not the correct code to look for the mistake
<nipos>
Seems you were right,that actually solves the crash.Always when you think you finally understand at least a little bit about C++,something like that happens :/
<nipos>
nephele: Do you have an idea what went wrong at the line I sent above?Would be cool to get it solved,otherwise I'll open a ticket now since I don't know any solution
wicknix has quit [Ping timeout: 480 seconds]
<nephele>
feel free to open a ticket, but you can put a breakpoint there with the debugger, and at that point create a debug report
<nephele>
we can then view the state the variables are at at that time
<nipos>
How can I do that?
<nephele>
Which part?
<nipos>
The only situation where I normally use the Debugger is if an app crashes and I get that crash window
<nephele>
You can just open it from the menu
<nephele>
and then debugger launches the app for you
<nephele>
or you can use "Debugger <path to my app>" on the commandline
<nipos>
Yes,I have opened Keymap with it.Now how do I add a breakpoint?
<nephele>
can you see the source code view?
<nipos>
Source file unavailable
<nipos>
I used the repo version.Should I use my own built version instead?
<nephele>
Darn. Maybe you have to recompile it with debug symbols. but no clue, this is black magic to me when Debugger has or has no source files .-.
<nephele>
but anyway, in the souce view left are these black dots on the white strip, click one, the line will turn red
<nephele>
that is a set breakpoint
<nipos>
Ah ok
<nephele>
if you run it after that it will run up untill the breakpoint
<nipos>
There's no haiku_debuginfo package.Would have been too easy
<nephele>
no clue how this is supposed to work, this has always been a point of annoyance to me
<nipos>
Self-built one also doesn't work,and no,I don't want to rebuild the whole thing with debuginfo.My computer isn't the fastest one as you may know
<nephele>
maybe waddlesplash has some info on how this is all supposed to work
<nipos>
I'll link the file and line I think causes the crash in the ticket and waddlesplash may know a lot better than me what is the reason for it crashing
<waddlesplash>
nephele: nipos: userland behavior is totally irrelevant here, I mean it may be wrong but it still should not cause a kernel crash
<waddlesplash>
so the fact that it does t happen with libroot_debug isn't relevant
<waddlesplash>
we need to fix this in the kernel anyway
<waddlesplash>
I'll check and see if I can reproduce later
<nipos>
Ah good,thanks
<nephele>
waddlesplash: doesn't it make debugging simpler to figure out how exactly userland provokes this assertion?
BrunoSpr has joined #haiku
<waddlesplash>
only if userland causing it is unreliable
<waddlesplash>
if userland can reliably cause it then not really, all the interesting debugging happens after the syscall
DKnoto has joined #haiku
<BrunoSpr>
evening all...
<nephele>
alright
<nephele>
for my control colors thing i need to re-calculate the gradient so it matches the old one. The new values I have give a too high contrast from top to bottom
<nephele>
I guess I need a pen and paper and properly solve this :D
<Begasus[m]>
old-school! :)
<nephele>
Begasus[m]: pfft, I used a RPN calculator during my education a lot
<nephele>
just solving stuff on paper is sometimes much nicer
<nephele>
to be honest I've not found any application for a computer that comes close to the utility of pen and paper for solving math, *despite* a keyboard beeing a much faster and more precise inpout
<nephele>
don't even have an RPN calculator on Haiku :(
<nephele>
Skipp_OSX: is the deskbar Menu opener also a button?
<nephele>
The rendering of deskbar with the menu color there also changes if the gradient is fixed
<nephele>
or, well, the part that uses the gradient anyhow
<nephele>
I'm begining to think this is too many moving parts. Maybe you are right and we should "for now" do panel => control color. And i fix the gradient stuff in a future patchset
<nephele>
low cpu and ram usage would be something we can massively improve, also by starting to cache *anything* :P
Nasina has joined #haiku
<BrunoSpr>
anyone an idea? Clicking on an icon in the folder will not select the folder, I have to click on the text to select the folder... was it like this before? I don't think so
<pol>
Cache on HDD?
<pol>
No, works here icon/label regardless.
<nephele>
cache in memory for starters, not re-requesting all network ressources each time
<BrunoSpr>
open the home folder and click on the icon works for you?7
<pol>
Testing new nigtlies?
<BrunoSpr>
this is hrev58610 nightly
<BrunoSpr>
64bit
<pol>
Yeah, Icon mode/list mode/mini mode.
<pol>
But, I think the code is somehow selective. It's not a perfect square it's more like a vactor icon where some click around it won't select it.
nephele has quit [Quit: Vision[]: i've been blurred!]