<avanspector[m]>
Hi, I'm porting an from linux application, are POSIX errors positive by default or not? For example in libroot
<avanspector[m]>
Hi, I'm porting a linux application, are POSIX errors positive by default or not? For example in libroot
<Begasus[m]>
morning peeps
<augiedoggie>
not
<Begasus[m]>
here it is :) g'evening augiedoggie :)
<augiedoggie>
there is a posix_error_mapper library/flag if you want positive though
<augiedoggie>
howdy Begasus[m]
<Begasus[m]>
ah my bad :)
HaikuUser has joined #haiku
Begasus has joined #haiku
HaikuUser has quit []
ChaiTRex has quit [Remote host closed the connection]
ChaiTRex has joined #haiku
<Begasus[m]>
k, let's see how this works with the latest push to haikuporter :)
<Begasus>
error: invoking cmake with -DCMAKE_BUILD_TYPE=RelWithDebInfo without debug info packages specified
<Begasus>
that works :)
<Begasus>
error: invoking cmake with -DCMAKE_BUILD_TYPE=Release with debug info packages specified
<Begasus>
that too +1
AlienSoldier has quit [Quit: Vision[]: i've been blurred!]
<Begasus[m]>
needs testing on meson build later
diver has joined #haiku
<Begasus>
test
<Begasus>
(don't mind me for a bit)
<Begasus>
nice, gitlab shows fine now with WebPositive +1
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
erysdren has joined #haiku
janking has joined #haiku
<janking>
good morning to ya all
dalme has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
<Begasus>
morning janking
<phschafft>
avanspector[m]: why is that important?
<janking>
:)
diver has quit [Quit: Leaving.]
<erysdren>
good morning janking
<phschafft>
mau erysdren.
Yoke has quit [Quit: Vision[]: i've been blurred!]
<janking>
:) erysdren
HaikuUser has joined #haiku
HaikuUser is now known as Yoke
<Yoke>
morning ya'll hope you're all having a good day
<Begasus>
Warning: POLICY WARNING: "/packaging/kosmindoormap24/lib/libKOSMIndoorRouting.so.1" needs library "libRecast.so.1", but the package doesn't seem to declare that as a requirement
<Begasus>
nice :)
<Begasus>
Hi Yoke
<Yoke>
heya Begasus, hope you're doing well
<Begasus[m]>
yeah :)
kevans91 has quit [Ping timeout: 480 seconds]
_-Caleb-_ has left #haiku [#haiku]
diver has joined #haiku
<Yoke>
any of yous been woken up at 4am by a ladder falling on a neighbor's roof? Cause that's what happened to me this morning and I've been up since :/
<dovsienko>
what prevented you from going back to sleep then?
<Begasus[m]>
not the ladder, dogs were up at 4am :)
<Yoke>
dovsienko: I have problems falling asleep beacause of some of my meds, nevermind the fact I'm a light sleeper either way
<dovsienko>
a classic move would be to get out of bed to get cold, then to get back into the bed, get warm and fall asleep
<dovsienko>
anyway, one time I was sleeping after/before work, and the police decided to look for something or someone in the street where I lived at the time
<Begasus>
so far for sleeping then dovsienko :)
<dovsienko>
so they slowly flied a police helicopter at low altitude around 3 o'clock in the morning
<dovsienko>
and after they left, I went back to sleep, as well as most of the street, I suppose
<Begasus[m]>
💣 nice :)
<Yoke>
that sucks, reasonable thing to happen but man that just sucks
<dovsienko>
at least there were no gunshots
<Yoke>
that's....a fair point
<erysdren>
true
<Yoke>
thankfuly I live in a place where gun ownership (class1/2, not RIF, or air@>12ft-lbs) is fairly rare
<Yoke>
honestly the classification system is probably enough to tell where I'm from XD
bjorkintosh has quit [Ping timeout: 480 seconds]
<MonniTheCat>
Here gun ownership is so common that they have to have spare Santa Clauses for every Christmas...
<MonniTheCat>
Spent whole morning figuring out why my 64-bit Haiku keeps crashing... Found two revisions that broke different things...
<dovsienko>
I've been thinking about what would be the simplest next improvement to Haiku networking in libpcap space
<dovsienko>
it could be pcap_setdirection(), if internally network packets already have some associated metadata
<dovsienko>
it could be even simple enough for a student to do as one of the next projects
<dovsienko>
it would not solve a significant practical problem, but would be a good learning material
<dovsienko>
two more complicated projects could be implementing sending of packets (which would unblock tooling such as tcpreplay and possibly Scapy) and kernel-mode BPF (which would make packet filtering faster)
<MonniTheCat>
Last thing on my TODO list is to figure out how much is missing from implementing IPv6 routing...
<dovsienko>
also there was an emulator (Hercules?) that wanted to send packets via libpcap. usually if there is a call pcap_inject() or pcap_sendpacket() in the source code, that's it
janking has quit [Quit: Vision[]: i've been blurred!]
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
<Yoke>
MonniTheCat, excuse the fuck outta me but... You guys have to have SPARE SANTAS?! what?!
_-Caleb-_ has joined #haiku
zard has joined #haiku
erysdren has quit [Quit: Konversation terminated!]
<MonniTheCat>
<Yoke> "Mika Lindqvist, excuse the..." <- Someone said that in Iceland there is 13 Santas, but here we have thousands... Some get arrested for DUI, some just hang around too close to Russian border...
HaikuUser has joined #haiku
dalme_ has joined #haiku
HaikuUser has quit []
dalme has quit [Ping timeout: 480 seconds]
diver has quit [Quit: Leaving.]
diver has joined #haiku
diver has quit [Ping timeout: 480 seconds]
illwieckz has quit [Ping timeout: 480 seconds]
illwieckz has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
mmu_man has joined #haiku
Skipp_OSX has joined #haiku
imrahil has joined #haiku
MisthaLu has joined #haiku
janking has joined #haiku
wicknix_ has joined #haiku
wicknix has quit [Ping timeout: 480 seconds]
Begasus has quit [Quit: Vision[]: i've been blurred!]
linuxmaster has quit [Ping timeout: 480 seconds]
janking has quit [Quit: Vision[]: i've been blurred!]
kevans91 has joined #haiku
bjorkintosh has joined #haiku
x10z has joined #haiku
x10z has quit []
<mbrumbelow>
Good morning
<mbrumbelow>
Good Afternoon
<mbrumbelow>
Good evening!
<dovsienko>
and good night? (Truman Show)
<mbrumbelow>
Does it make sense to have a post on the main website for a "state of the union" report concerning the Haiku project as a whole? This report would outline the progress the project has made and where we are today. We do have monthly Haiku Activity & Contract Report(s) authored by Dr. @waddlesplash, but I am thinking of a yearly review for Haiku, apps in Haiku Depot, etc...
<mbrumbelow>
Thoughts?
<waddlesplash>
PulkoMandy does something like this for LinuxFR
<waddlesplash>
however a lot of it is kind of just summarized activity reports
<dovsienko>
would it be possible to state which high-level problems would need to be solved to make Haiku R1 possible?
<mbrumbelow>
Thinking it would be some of the content from your monthlies and some of the items from the forum post "Announcement: New/updated in HaikuDepot (2024)"
<PulkoMandy>
the latest yearly report was rejected by linuxfr because it was too long. So I do it every 3 months now
<mbrumbelow>
🙁
<mbrumbelow>
Something very simple, probably one page.
<dovsienko>
"yearly report, volume 1 of 4"
<mbrumbelow>
@dovsienko
<mbrumbelow>
Pretty much for what was done this year.
<mbrumbelow>
@dovsienko: The high level problems are time, effort and commitment. Not necessarily in that order.
<dovsienko>
I mean, in terms of features that R1 must have, rather than in terms of how to get it done
<dovsienko>
e.g. BFS resizing, at least offline
<Skipp_OSX>
If you'd like to do that sure, the R1B5 release did already recap work done since R1B4
<Skipp_OSX>
It's always good to have more information on what has been done though so if you'd like to do a year end recap that would be helpful.
<mbrumbelow>
@Skipp_OSX: I might just author a draft over the weekend.
<Begasus[m]>
Fresh Fedora 41 install in VM, updates ... almost 5GiB! whoot!
<MonniTheCat>
waddlesplash: For some reason I'm getting garbage in _user_create_pipe() ... flags should be 0 when called from user code, but it gets random values...
<waddlesplash>
do you have a mismatch between kernel and libroot?
<waddlesplash>
this sounds like you have a libroot version from before the extra argument was added to create_pipe
<MonniTheCat>
<waddlesplash> "this sounds like you have a..." <- That would only mean it fails to build, but jam doesn't stop on the error...
<waddlesplash>
or you have an old libroot in non-packaged
Begasus has joined #haiku
<Begasus[m]>
installing Fedora in the VM hasn't been a great succes so far :P
<Yoke>
that's just using virtual machines in general tho, doesn't matter what program you use, something allways seems to go wrong
<Begasus[m]>
Haiku is running pretty fine on it
<Yoke>
I'll give you that, Haiku is pretty much flawless on a VM (other than trying to get anything other than a 4;3 ratio in VirtualBox)
<Yoke>
(but that is more of a VirtualBox problem)
<Skipp_OSX>
VMs don't do a very good job of passing through the GPU from the host to guest. Haiku runs on VM pretty well since it doesn't use GPU.
_-Caleb-_ has left #haiku [#haiku]
<Skipp_OSX>
But if you're using Fedora you're going to need a GPU to run anything very well. Of course some VMs do better with this than others especially the hypervisor type ones.
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<Yoke>
blessing and a curse that though ain't it, it'l run on pretty much anything so long as the cpu has the beans but that comes at the cost of havng little to no hardware acceleration :/
<nekobot>
[haiku/haiku] 2f7bf3b339cb - virtio_net: Don't print a message on SIOCGIFSTATS.
<Begasus>
k, finetuning the recipe and I guess Tokodon will be ready for a release tomorrow :D
<Yoke>
Guessing from the name that's something from KDE?
<MonniTheCat>
<waddlesplash> "or you have an old libroot in..." <- Can't find any extra copies of libroot.so in non-packaged
<Skipp_OSX>
KDE Mastodon client, yeah
<waddlesplash>
MonniTheCat: well, that's the only way I can think of that you'd get garbage in argument 3.
<waddlesplash>
err, in the new argument I mean
<waddlesplash>
check your listimage
<waddlesplash>
"listimage | grep libroot"
<waddlesplash>
and see if there's one coming from somewhere besides a package
<MonniTheCat>
waddlesplash: Only shows /boot/system/lib/libroot.so
<waddlesplash>
odd
<waddlesplash>
I don't know how you're getting garbage in there then
<waddlesplash>
But if you are that might explain the problems and why you're getting crashes
<waddlesplash>
because it means the userland syscall hooks are out of sync someho
<waddlesplash>
so no wonder runtime_loader crashes, it's calling the wrong syscalls, etc.
<waddlesplash>
try a completely clean build?
<MonniTheCat>
waddlesplash: rm -rf generated?
<waddlesplash>
yeah
<Begasus>
Skipp_OSX, as long as there isn't a native one this will do well enough, no one is pushed into installing it :)
Yoke has quit [Ping timeout: 480 seconds]
<Skipp_OSX>
We can run on everything and have GPU acceleration, we just haven't gotten the later yet. Yeah it's fine more software the better.
<Skipp_OSX>
Once Haiku gets 3d hardware acceleration the creep will set in inevitably as it has on all those other OSs.
<Begasus[m]>
I don't really have a need for the acceleration, fine as it is (PS running the Fedore from USB on another laptop was ... not so well either) :)
<Begasus[m]>
only could use some speed in building :D
<Skipp_OSX>
we need it for web stuff mostly
<MonniTheCat>
Recompiling Haiku and the buildtools from scratch will take a while ;)
<Yoke>
So i just had to completely reinstall my system, that was.......fun :/
<Anarchos>
Yoke why ?
<Anarchos>
Begasus[m] is it a game ?
<Yoke>
I was using iceweasel then the all of a sudden half of my screen is filled with what looks like a debug menu? but it was completely frozen, i then restarted the pc and it couldn't even boot it would just throw that same debug thing up on the screen
Anarchos has quit [Quit: Vision[]: i've been blurred!]
andreaa71 has left #haiku [#haiku]
Anarchos has joined #haiku
<Anarchos>
Begasus i tried knetwalk24. Too easy for me.
<Yoke>
so it's like with that one Mario64 speedrun? just a cosmic ray deciding it's gonna be annoying for it's fhemptosecond of existence?
gnaran[m] has joined #haiku
janking has quit [Quit: Vision[]: i've been blurred!]
<parabyte>
bit off topic, but kinda to do with haiku as i been playing with arm port
OscarL has joined #haiku
<parabyte>
so i purchased a new android box off ebay
<parabyte>
promised a certain cpu and 4 gigs of ram and 64gigs of storage
<parabyte>
after i removed the android, i was getting 1gig ram 8gig storage, had argument via ebay with the seller
<parabyte>
opened up the device, 1gig ram 8gig storage!
<parabyte>
be wary is what im trying to say :)
<OscarL>
Yoke: not that hard to get decent video on VirtualBox (avoiding the 4:3 resolution at start up, even). Lil' how-to: https://bpa.st/QTOQ
x44a has joined #haiku
<avanspector[m]>
waddlesplash: I've seen commits on btrfs in the kernel, there is some support for it?
<waddlesplash>
avanspector[m]: some RO support, yes
<waddlesplash>
MonniTheCat: it's possible bash is calling pipe2?
<MonniTheCat>
waddlesplash: I checked and it's not calling... lower 32-bits match the garbage in "flags", but higher 32-bits are not zeroed...
<waddlesplash>
lower 32 bits of what?
<waddlesplash>
how are we getting garbage then?
<MonniTheCat>
waddlesplash: RSI register
<waddlesplash>
ok
<waddlesplash>
but that doesn't explain how the syscall got called with bogus data instead of 0
<MonniTheCat>
waddlesplash: High bits of RDX, RSI and R14 are identical... That might mean those are some kind of pointers in user space...
angrystar170 has joined #haiku
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<scantysnax>
if this is inear a function call, r14 doesn't need to be saved, and rsi and rdx will contain function paramater 2 and 4 iirc
<scantysnax>
near*
Anarchos has joined #haiku
<scantysnax>
shouldn't be any garabe, though.
<scantysnax>
garbage*
<MonniTheCat>
scantysnax: That function has only two parameters...
<Begasus>
Anarchos, tried the "harder level"? :)
<scantysnax>
so they will be in rdi and rsi
<Begasus>
but yeah, it's not that hard mostly
<scantysnax>
respectively
<Anarchos>
Begasus not yet
<Begasus[m]>
ah! :)
mr_lou has joined #haiku
<Anarchos>
Begasus[m] i tried the difficult one. The most difficult seems not funny enough to me :)
MisthaLu has quit [Ping timeout: 480 seconds]
<Begasus[m]>
heh
<Begasus[m]>
Anarchos at least fun for a few minutes :)
<Anarchos>
Begasus[m] i have more fun to solve my bios drive detection :)
* Anarchos
will reboot to test...
<OscarL>
neeerd! :-P
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
btw, hello Begasus[m].
<MonniTheCat>
r14 reads out "basename $path", so clearly variable in memory...
Anarchos has joined #haiku
<Begasus>
Hi OscarL:)
<OscarL>
(I think MonniTheCat managed to get weirder bugs than me)
<Anarchos>
Begasus ok, if i remove all the traces i put, it works. I think too much traces make the bootloader crashes in some sense.
<Begasus>
don't mess too much with it Anarchos :) still need it in the VM :)
<Anarchos>
Begasus but i can confirm my patch works :)
<MonniTheCat>
OscarL: Going past function parameters is always fun bug...
<Begasus>
I'll let others test it Anarchos :D
<Anarchos>
waddlesplash i intend to write a lot of explanations and questions for my patch. What is the preferred format for that in the pull request ? Markdown ?
<waddlesplash>
if your patch really needs that much explanation it should probably be in some combination of inline comments, the commit message, and then docs/develop if really necessary
<Anarchos>
waddlesplash ok. Bootloader code is messy, and i made it more messy....:)
<Anarchos>
waddlesplash but it's more a matter that i have questions to get advices on how to do things right at some place..
mmu_man has quit [Ping timeout: 480 seconds]
Begasus has quit [Quit: Vision[]: i've been blurred!]
<MonniTheCat>
waddlesplash: I'm starting to think the garbage is lower bits of the userFDs array pointer... Still don't know how that could happen...
JakeSays1 has joined #haiku
mmu_man has joined #haiku
JakeSays has quit [Ping timeout: 480 seconds]
SLema has quit [Ping timeout: 480 seconds]
HaikuUser has joined #haiku
HaikuUser has quit []
oevl has joined #haiku
<MonniTheCat>
* array pointer or pointer to where "flags" variable is stored in memory... Still
<OscarL>
dovsienko: sorry, I should have been more specific (given that the error meessage doesn't gives enough context)... issue is with how Haiku defines the dirent struct.
<OscarL>
for example, on other recipes that used "buffer = std::malloc((sizeof(dirent) - sizeof(de.d_name))" got patched as "buffer = std::malloc(sizeof(dirent) path_size + 1);"
<OscarL>
* "+ path_size + 1);"
<dovsienko>
NAME_MAX + 1 is usually a safe substitute
<OscarL>
great. thank you!
<dovsienko>
I could not find at the time whether NAME_MAX includes the terminating NUL, so made it + 1
<dovsienko>
I mean, instead of the "sizeof(dirent.d_name)" part, not the entire structure
<dovsienko>
and in this case, since it subtracts the size, the simplest move could be to make it just sizeof(dirent)
<dovsienko>
(unless it makes a million copies of the buffer and you need to keep the sizing byte-perfect)
<dovsienko>
OscarL: that said, flagging the failure to compile to the developers may be the best way to resolve this long-term
<OscarL>
(but I keep forgetting the details :-D)
<dovsienko>
yes, that's the same issue
<OscarL>
From an earlier comment on that PR: "Ultimately the problem here is that GCC does not treat char[0] and char[] equivalently, or else I would have used the former, which returns sizeof()=0. (GCC throws errors when you try to access any data in a char[0], it doesn't on char[])."
<OscarL>
(I'm just a lowly organ-bank from sector 7... I think I'll leave rising such issues upstream to the people that actually understands the issue :-D)
<dovsienko>
char[] is the same as char *, char[0] reminds me of something that I cannot immediately recall
<waddlesplash>
it's a VLA at the end of a dirent
<dovsienko>
yes, that one
<waddlesplash>
the problem is that GCC 11 was optimizing stuf so that strcmp() on a char[1] was optimized out
<waddlesplash>
because GCC assumed "ot
<waddlesplash>
it's [1], it can't be longer!"
<waddlesplash>
so we changed to 0, but this had a similar problem IIRC
<waddlesplash>
so we finally just went with []
<waddlesplash>
which does work properly
<phschafft>
there has been a lot of discussion in the last like 30 years on char[0].
<phschafft>
or how tailing data in structs should be declared.
<phschafft>
and I must personally say that giving a definitive size... should mean that the object in question has that size.
<waddlesplash>
c89 did not allow just [] though is the issue
<waddlesplash>
so we have ifdefs for gcc2
<waddlesplash>
it works in newer GCC c89 mode though I guess
<phschafft>
that is correct. and the old style way was to just have a char c and then use &c.
<waddlesplash>
OscarL: not in an existing patch?
<OscarL>
only 1.69 has a patch for dirent
<dovsienko>
ah, right, it's the idiom of having a something[0] member in a struct to have the typing done at compile time and the sizing at run time. I used to see plenty of these in network protocol implementation code for protocols like OSPF and RIP
<dovsienko>
protocol that use TLVs and other variable-length encoding don't benefit from that pattern
<OscarL>
(dirent patch for boost1.69 is for a diffent lib/filesystem/ file, though)
tsukasa has joined #haiku
<phschafft>
please keep also in mind that the char[] syntax is part of C99. so this discussion is only pre-1999 compilers ;)
x10z has quit [Quit: My Unrecognized Mac has gone to sleep. ZZZzzz…]
x10z has joined #haiku
<dovsienko>
I suppose there's a corresponding version of C++, since Boost is a C++ library
<dovsienko>
anyway, you probably already know what to do
<phschafft>
dovsienko: I'm not a C++ guy. but my guess would be that as C++ compilers generally need to understand C, those who can work with C99 will have it in C++ as well, even if as an extention.
janking has joined #haiku
<dovsienko>
OscarL: please report the problem upstream, and if everything goes well, things will just work (eventually)
KapiX has joined #haiku
<OscarL>
just my luck now... "[errno -2147483648] vfork (Out of memory)
<OscarL>
" :-D
<scantysnax>
nice
<scantysnax>
also, hi OscarL
<OscarL>
dovsienko: will try (guess I can at least refer them to similar issues/discussions, even if I'm not up for proper discussions on the matter)
<OscarL>
hello there scantysnax!!! :-)
<scantysnax>
i was looking at the bt848 datasheet.....
<dovsienko>
OscarL: giving exact steps to reproduce would be a good starting point, such as "download this file, install a VM..."
<scantysnax>
it's a 160 pin beast
<scantysnax>
and i do not have the skills to write a driver for that as far as i know.
<dovsienko>
scantysnax: swap it for something that's already supported to some degree?
<OscarL>
scantysnax: 128 pins on the SAA713x family :-)
<scantysnax>
dovsienko, well, the other day i considered the possibility of porting freebsd bktr driver