TheEricExperiment has quit [Remote host closed the connection]
TheEricExperiment has joined #haiku
countryboy has joined #haiku
<countryboy>
hello
dasouth has joined #haiku
dasouth has quit []
dasouth has joined #haiku
dasouth has quit []
roiredX has joined #haiku
<roiredX>
good day
countryboy has left #haiku [#haiku]
dasouth has joined #haiku
Coldfirex has quit [Remote host closed the connection]
dasouth has quit []
countryboy has joined #haiku
Saijin_Naib has joined #haiku
Saijin_Naib has quit []
CatButts has joined #haiku
countryboy has left #haiku [#haiku]
johndrinkwater has joined #haiku
bitigchi has joined #haiku
johndrin1water has quit [Ping timeout: 480 seconds]
KapiX has joined #haiku
<nephele[m]>
juliencenri: to be honest, i would implement users akin to linux namespaces, and keep posix users for certain processes as such
<nephele[m]>
though, we can do either, currently almost all services run as root anyway
<shaka444[m]>
Both BePDF and DocumentViewer not working after hrev55399
<waddlesplash>
yes, I need to submit patches for those, or someone else can
Dusan_Krehel has left #haiku [#haiku]
<repetitivestrain>
"...I think BBitmap memory being freed depends on a running BApplication or something like that..." well it is definitely being freed somehow, because subsequent bitmap allocations seem to re-use memory used by the bitmap that was freed
<repetitivestrain>
but the memory doesn't show up as freed in ProcessController
<waddlesplash>
memory accounting as displayed in ProcessController etc. is kind of deceiving and we need to work on this
<waddlesplash>
it only displays actually used pages, we don't have anywhere that indicates how many pages are reserved (which can include swap reservations)
<waddlesplash>
so you may see 90% free, but you really have >10% of memory in a "reserved" state
<waddlesplash>
and, unlike Linux, we don't overcommit except on explicit request, so you actually can derive useful information from the reservations value
<waddlesplash>
I don't know if there's a ticket about this but there really should be
<repetitivestrain>
btw, is it normal for bitmaps to show up as part of the application's memory usage?
<repetitivestrain>
because i think they're on the App Server side
<waddlesplash>
they're on both
<waddlesplash>
shared memory
<repetitivestrain>
ah, so processcontroller will display them as part of both? thanks
<waddlesplash>
no, ProcessController should display actual pages used in the system
<waddlesplash>
so this should only be counted once
<repetitivestrain>
hmm, interesting, so it should be counted as part of the app server?
<repetitivestrain>
thanks
Coldfirex has joined #haiku
<waddlesplash>
what do you mean?
<waddlesplash>
if you are looking at overall system memory usage, of coursed shared memory should only be counted once
<waddlesplash>
if you are looking per-application then indeed it can be double-counted
<repetitivestrain>
i meant per-application in the process controller
<repetitivestrain>
at least, I think, basically I have a BView that does double-buffering
<repetitivestrain>
each buffer is a BBitmap, and there's a view attached to the back buffer
<waddlesplash>
it will most probably appear on both
<repetitivestrain>
when a buffer is flipped, the view attached to the back buffer is detached, the two buffers are exchanged, the view is attached to the new back buffer, and invalidate is called
<waddlesplash>
I'm not sure processcontroller accounts for cloned areas correctly here, yes
<repetitivestrain>
(also please tell me if there's a smarter way to do double buffering, thanks)
<repetitivestrain>
also thanks
<waddlesplash>
you probably can just have BBitmaps and draw them on the view
<repetitivestrain>
yeah that's what I'm doing right now I think
<waddlesplash>
or, better, just use SetViewBitmap and let app_server draw it for you
<repetitivestrain>
SetViewBitmap kind of results in an unpleasant tiling effect
<repetitivestrain>
but thanks, I'll look into it
<waddlesplash>
possibly can be disabled, I think?
<waddlesplash>
anyway, you mean you are using B_BITMAP_ACCEPTS_VIEWS?
<repetitivestrain>
yeah I am
<waddlesplash>
worth noting this is slightly expensive, it requires an extra thread for every bitmap you set it on
<waddlesplash>
on the app_server side that is
<waddlesplash>
and accompanying ports
<repetitivestrain>
thanks, I think that is managable
<repetitivestrain>
the real problem though, is the bitmap getting re-allocated every time the view being double-buffered is resized (and please, tell me if there is a smarter way to do this)
<waddlesplash>
SetViewBitmap has an "options" parameter which is where B_TILE_BITMAP is set
<repetitivestrain>
and I noticed that the memory consumption reported in the processcontroller doesn't shrink when the view is shrinked
<waddlesplash>
you can just set that to 0 and no tiling will happen I'm pretty sure
<repetitivestrain>
hmm, I will try that
<repetitivestrain>
it doesn't grow, so i'm sure nothing is being leaked
<repetitivestrain>
I think I missed the options parameter
<repetitivestrain>
by the way, would it be reasonable to have an app-server side double buffering mechanism?
<repetitivestrain>
on X-windows, I would have used the X double buffering extension, not sure if there's anything comparable in other APIs
<waddlesplash>
probably not, most things don't do a lot of direct drawing here
<repetitivestrain>
hmm... okay, thanks
<waddlesplash>
OpenGL is of course double buffered internally (or can be anyway) independent of app_server
<waddlesplash>
I think app_server does internal double buffering?
<repetitivestrain>
for individual views? you mean BeginViewTransaction?
<repetitivestrain>
does that double-buffer internally? if so, it would be nice
<waddlesplash>
no, for the whole screen
CatButts has quit [Ping timeout: 480 seconds]
<repetitivestrain>
ah, okay, that doesn't quite fit my use-case, but thanks
<waddlesplash>
view transactions are for command batching, I believe
<repetitivestrain>
yeah it seems that way, but having commands draw to a back buffer and then showing the buffer when ending a transaction would be a way to implement it
<waddlesplash>
repetitivestrain: pretty sure SetViewBitmap clones the bitmap
<waddlesplash>
"Once bitmap has been passed in and this method returns the caller may safely delete the object."
<waddlesplash>
I'm less familiar with this API, but I think that is what this is aying
<repetitivestrain>
SetViewBitmap doesn't clone the bitmap, I just tried
<waddlesplash>
PulkoMandy will probably know more than me here
<repetitivestrain>
it probably increases a reference count, or something
<repetitivestrain>
(speaking from the POV of not knowing how bitmaps are implemented server-side)
<repetitivestrain>
bright side is, setting options to 0 makes the tiling go away
TheEricExperiment has quit [Quit: Going offline, see ya! (www.adiirc.com)]
<repetitivestrain>
i confused it with followFlags at first
<waddlesplash>
what is your use case that you need to draw on bitmaps and double buffer them, anyway?
<repetitivestrain>
well i'm porting Emacs
<waddlesplash>
if you are porting some toolkit, won't it be drawing its own bitmaps completely?
<repetitivestrain>
it has a very silly internal architecture
<repetitivestrain>
basically, it expects to be able to draw everywhere, at any given moment
<repetitivestrain>
so right now I run the emacs event loop in a separate thread
<repetitivestrain>
and send SIGPOLL to it whenever some notable event happens in an app/interface kit thread
bitigchi has quit [Remote host closed the connection]
<repetitivestrain>
naturally, this results in a lot of flickering, so the emacs text display mechanism (redisplay) provides an interface for supporting double buffering
<waddlesplash>
you don't have to do that, I think
<waddlesplash>
you can just lock the looper and draw from the main thread
repetitivestrain has quit [Quit: Vision[0.10.3]: i've been blurred!]
repetitivestrain has joined #haiku
<repetitivestrain>
uhm sorry there, I tried to hit alt+w (emacs for cut) and forgot it was close
<repetitivestrain>
anyway, I do call LockLooper to draw from the Emacs thread
<repetitivestrain>
basically, when some Emacs code modifies the content of the display, it calls redisplay
<repetitivestrain>
redisplay then calls platform dependent functions through an abstraction layer called the redisplay interface, to draw text and other things that might have changed
tbcr has joined #haiku
<waddlesplash>
also, I think mmu_man had an existing xemacs port at some point
<repetitivestrain>
so if redisplay thinks a portion of the screen has been cleared, it calls RIF->clear_frame_area with a rectangle describing the area to clear, and in this function, I lock the looper, and fill the rectangle specified
<repetitivestrain>
well, XEmacs is also known for extreme flickering on X
<repetitivestrain>
and the internal architecture has changed a lot, after all, the fork happened in the 90s
<repetitivestrain>
i could get away without double buffering, it just results in an extreme amount of flickering
<nekobot>
[haiku/haiku] 05709fe3f747 - headers: Fix B_CURRENT_IMAGE_SYMBOL for GCC2.
HaikuUser has joined #haiku
HaikuUser has quit []
beaglejoe has quit [Quit: Leaving]
Dusan_Krehel has joined #haiku
KapiX has quit [Quit: KapiX]
KapiX has joined #haiku
<roiredX>
bye all
roiredX has quit [Quit: Vision[0.10.3]: i've been blurred!]
<Coldfirex>
Im used to using Google Play Music via a web browser while working. It doesnt work in Web+. Does anyone have an alternative that works in Web+ by chance?
HaikuUser has joined #haiku
HaikuUser has quit []
andreasdr has joined #haiku
tuaris has joined #haiku
tqh has joined #haiku
gouchi has joined #haiku
kikadf has joined #haiku
Begasus has joined #haiku
<Begasus>
'lo peeps
<B2IA>
(mrentropy) Howdy
<Begasus>
Hi mrentrophy
<andreasdr>
Hi
sg94 has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit [Quit: Vision[0.10.3]: i've been blurred!]
<nephele[m]>
Is there a way to dump clipboard contents for inspection?
<nephele[m]>
(for html clipboard and such)
<nephele[m]>
seems clipboard with -o for saving to BMessage, hm
<nephele[m]>
message can show a flattened bmessage as text, but it does not show me the actual content, only the sizes, any way to dump this directly aswell?
<sg94>
it's been a while but iirc Clipdinger in the HaikuDepot should give you what you're looking for
<Begasus>
hi andreasdr nephele[m]
KapiX has quit [Quit: KapiX]
<andreasdr>
:)
Begasus_32 has joined #haiku
<nephele[m]>
sg94: it doesn't even shot the text/html section unfortunately :(
Animortis has joined #haiku
<Animortis>
Anyone who knows c++ wanna look at my noob pastebin and tell me why getline isn't working?
<nephele[m]>
hi begasus
<Begasus>
Checking libjxl on the new changes for haikuporter :)
<Not-5726>
[haikuports/haikuports] Begasus pushed 1 commit to master [+1/-0/±0] https://git.io/JuAnT
<Not-5726>
[haikuports/haikuports] Begasus 90501c4 - libjxl, new recipe (#6096)
Begasus_32 has quit [Quit: Vision[0.10.3]: Gone to the dogs!]
<B2IA>
(mrentropy) Dang. Web+ is having a problem with my Wordpress site. Still don't know what the correlation is with the "finished" message on the left, and the blue bar on the right.
mmu_man has quit [Ping timeout: 480 seconds]
<PulkoMandy>
probably all random because the code is all broken :>
ClaudioM has quit [Quit: leaving]
<B2IA>
(mrentropy) Ah, okay. That sort of explains it. I wasn't sure which one (if either) was telling the truth, or if they were indicating different things.
<nephele[m]>
I dunno, I think it just shows finished when the "main" first network request finished, or something
<nephele[m]>
I finally managed to get my text/html extracted, that was too damn complicated... had to hack on Kottan to get it to render it, and even then it had no way to copy it out of it (ouch)... so i just user puts() and will deal with this some later day
<nephele[m]>
and i guess, make a patch for kottan to display text/html and text/plain atleast... maybe just all text/*
Anarchos has joined #haiku
mmu_man has joined #haiku
oldleather has joined #haiku
vdamewood has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<Anarchos>
hello mmu_man
Vidrep_64 has joined #haiku
<oldleather>
hi all
<andreasdr>
Hi
<Begasus>
g'night peeps
Begasus has quit [Quit: Ik ga weg]
<oldleather>
I've got a networking issue with R1 Beta 3 32-bit... looks like the same sort of thing that's going on in Bugs 14268 and 14521
<oldleather>
Qualcomm Atheros AR9462 wireless card not showing up in the network preferences window... getting the "Unable to attach hardware (HAL status 3)" message in syslog
<oldleather>
Just wondering if there's a fix for this, or what I could do to help move toward one
<andreasdr>
BTW: I need help from haiku core devs. I need RWLock downgrade and upgrades. Means upgrading a readLock to writeLock and downgrading a writeLock to readLock. :) Ill subscribe to mailing list then and ask for it via mail.
<waddlesplash>
andreasdr: you probably don't need it, you can just unlock-relock
<nephele[m]>
Heh, there is a lot of html for copying code from github... jeez
<andreasdr>
@waddlesplash: Ok. I can try. And maybe think of it. Might depend on existing code there.
<Anarchos>
andreasdr good luck !
<andreasdr>
Thanks. My mate I am working with on our 3d engine is on vacation. Perfect time to try to continue DRM port.
<andreasdr>
:)
jmairboeck has quit [Quit: Konversation terminated!]
<andreasdr>
AFAIK OpenBSD did a new DRM port. Might be worth a look if they have improved code organization. But I dont expect much change in functionality as RADEONDRM is stable for lots of years. So whether I look or not, might not do a big change.
Maturion has quit [Ping timeout: 480 seconds]
<andreasdr>
@Waddlesplash: May I ask what could be done on Haiku to improve compilation speed?
<waddlesplash>
what's wrong with it now?
<waddlesplash>
it's pretty good so far as I can tell
<waddlesplash>
maybe not quite so fast as Linux, but not so far behind, I think?
<andreasdr>
Ok. Let me tell you some benchmarks:
<andreasdr>
Compiling TDME2(my engine), which is 6MB huge in source code takes
<andreasdr>
45s on Linux
<andreasdr>
32s on FreeBSD
<andreasdr>
1m50s on Haiku
<andreasdr>
1m50s on Windows
<andreasdr>
And even lot more time on OpenBSD
<waddlesplash>
what model of CPU?
<andreasdr>
AMD Ryzen 3950X
<andreasdr>
With 16 cores
<andreasdr>
I compile with -j30
* Anarchos
has a funny bug of two processes blocked on a socket : one in writing, the other in reading. Still wonder how to debug that
<Anarchos>
andreasdr compiles with -j17
<Anarchos>
andreasdr or -j16
<andreasdr>
Just asking. I know that Haiku kernel is not that much matured like the FBSD one. Also Haiku kernel is Microkernel whereas Linux/FBSD is monolithic.
<Anarchos>
haiku is hybrid kernel
<andreasdr>
Oh Ok. Tell me more please.
<waddlesplash>
yes, try with -j16 and see what changes, you are probably CPU-bound not I/O-bound so scheduler dynamics are a massive factor here
<waddlesplash>
and no, Haiku is not a microkernel, it's a modular monolithic kernel
<andreasdr>
Ok. I try in next sesssion.
<andreasdr>
Oh cool.
<andreasdr>
I thought it was micro kernel.
<Anarchos>
andreasdr drivers are linked against kernel and run in kernel space
<andreasdr>
Ah ok. Makes sense.
<waddlesplash>
also worth noting that Haiku nightlies build with KDEBUG_LEVEL=2 which does have noticeable performance implications
<andreasdr>
Cool. Learned something.
<andreasdr>
Ok. Ill try -j16 and tell you about it.
<andreasdr>
Ah Ok. But beta3 does not have KDEBUG_LEVEL=2
<andreasdr>
?
<waddlesplash>
correct, it's using KDEBUG_LEVEL=1, this disables the worst of the performance-impacting stuff
orealis_ has joined #haiku
<Anarchos>
waddlesplash first time i heard about that parameter !
<andreasdr>
Nice.
orealis has quit [Ping timeout: 480 seconds]
orealis_ is now known as orealis
<Anarchos>
waddlesplash may i ask for a strategy to find why my socket is blocking two processes in read and in write at same time ?
<waddlesplash>
not sure...
bitigchi has quit [Remote host closed the connection]
<Dusan_Krehel>
waddlesplash: my report bug, one question. r u change standart C library on musl?
<Dusan_Krehel>
waddlesplash: how can i check change?
<waddlesplash>
I replaced some functions that were from glibc to musl, yes
<andreasdr>
BTW: About daily usage. Is there any work planned on Mail? It does not seem to work for me.
<andreasdr>
I heard I can play now YT videos in WebPositive. Thats cool already.
<andreasdr>
Or I even tested If I remember right.
<Dusan_Krehel>
without/with = anothers version of somone library?
<andreasdr>
Also there is no Email client that worked for me.
<Dusan_Krehel>
updates =
<andreasdr>
Wait Ill give Haiku a shot right now. :) Lets have some fun.
andreasdr has quit [Quit: Page closed]
mmu_man has quit [Ping timeout: 480 seconds]
HaikuUser has joined #haiku
HaikuUser has quit []
HaikuUser has joined #haiku
Vidrep_64 has joined #haiku
<Not-5726>
[haikuports/haikuports] waddlesplash pushed 1 commit to master [+0/-0/±1] https://git.io/JuAyX
<Not-5726>
[haikuports/haikuports] waddlesplash 16f4231 - BePDF: Bump to pick up columnlistview fix.
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<oldleather>
If this is not the right venue to ask about the network issue, where should I go?
<waddlesplash>
oldleather: you can add a comment on the ticket. but I kind of doubt much will happen anytime soon with it; FreeBSD WiFi development has stagnated
<oldleather>
I'm not a developer, but I would like to help resolve this issue, even if all I can do is test and provide info
<waddlesplash>
and OpenBSD WiFi drivers don't support this family of devices at lal
<oldleather>
ah, I see
<oldleather>
the device does work under FreeBSD 13.0-
<oldleather>
RELEASE
<waddlesplash>
well, then that does indicate the problem is somehow on the Haiku side
<waddlesplash>
what and where are another matter
mmu_man has joined #haiku
<oldleather>
I'll look at a couple more things and see if I can determine which ticket is the more appropriate one (guidance here would be helpful)
<oldleather>
and then I'll leave a comment with system info, and attach a syslog
HaikuUser has quit [Quit: Vision[0.10.3]: i've been blurred!]
andreasdr has joined #haiku
<andreasdr>
Hi
<andreasdr>
With -j16 on Beta3 it takes 1m30s
<andreasdr>
How to change resolution? I started to get into boot loader with shift and space, but it did not work. I am using EFI.
<andreasdr>
Resolution is too low with my NVIDIA so I could not do anything with Vision Window which was not in screen space.
<waddlesplash>
spam spacebar for EFI
<andreasdr>
My votes would be for daily usage: Mail should work also and what would be nice would be USB video. I ever wanted to do a USB driver. But not yet done.
<andreasdr>
Ok. I try to spam spacebar next time I boot Haiku.
Vidrep_64 has quit [Quit: Vision[0.10.3]: i've been blurred!]
<andreasdr>
Or does nightly recognize my 1060 NVIDIA?
<andreasdr>
Oh wait I can add this to your "Haiku Sprint" post. Email at least.
tqh has quit [Quit: Leaving]
dburkart has joined #haiku
robobox has quit [Quit: Leaving]
The_Ringmaster has quit [Quit: Page closed]
gouchi has quit [Remote host closed the connection]
<B2IA>
(mrentropy) I find it easier to hold the spacebar down right after the drive light starts blinking. It should be late enough to avoid a keyboard error, and early enough before the boot process starts.
bitigchi has joined #haiku
kikadf is now known as Guest4
<andreasdr>
Is there maybe a config file where I can tell it should always stop at Bootloader?
<andreasdr>
But I can try to hold spacebar dow
<andreasdr>
n
ClaudioM has joined #haiku
WoC has quit [Remote host closed the connection]
WoC has joined #haiku
dburkart has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bitigchi has quit [Quit: Lost terminal]
AlaskanEmily has joined #haiku
HaikuUser has joined #haiku
DKnoto has quit [Ping timeout: 480 seconds]
HaikuUser has quit []
Skipp_OSX has joined #haiku
Coldfirex has quit [Remote host closed the connection]
<jt15s[m]>
I'm going to write up a list of suggested tasks for the Haiku Code Sprint - should I make this as a blog post or do it like the GSOC ideas page (i.e. a static page in the community folder)?
<waddlesplash>
none of the above
<waddlesplash>
easy tasks should be tagged with "easy" on Trac, or whatever we come up with on HaikuPorts
<waddlesplash>
otherwise, coding sprints are mostly for either fresh contributors (who will presumably start with an "easy" task) or existing contributors (who have some idea of what to do already :P)
<Vidrep_64>
I was using a Radeon HD7570 for awile on Haiku
<Vidrep_64>
awhile
<andreasdr>
Ok. Its even newer.
<andreasdr>
Seems to be AMDGPUDRM class.
<jt15s[m]>
waddlesplash: Alright then. I'll probably write up a blog post though announcing the Code Sprint just so we can spread the word
dcatt has joined #haiku
Vidrep_64 has quit [Quit: Vision[0.10.3]: i've been blurred!]