<trungnt2910[m]>
__cdecl and __stdcall doesn't seem to make any sense outside of 32-bit x86.
HaikuUser has joined #haiku
HaikuUser has quit []
Begasus_32 has joined #haiku
<PulkoMandy>
We don't really use it, but some old BeOS sources have these things and it was easier to keep (stubbed) support in the compiler than to fix them all
<PulkoMandy>
Possibly people just ported the patches to newer gcc versions without really considering if it was still useful
<trungnt2910[m]>
And it's causing quite a lot of trouble for me right now.
<nielx[m]>
How are they causing trouble?
<trungnt2910[m]>
It causes trouble when the build system uses -Wall.
<trungnt2910[m]>
s/-Wall./`-Werror`./
<trungnt2910[m]>
Defining it to nothing (which many source files that aim to be portable do) causes a redefinition warning, while _not_ re-defining it causes an "attribute ignored" warning.
<PulkoMandy>
can't you #undef before you #define them?
<Begasus>
I usually patch "-Werror" out on 32bit if it's causing build errors
<trungnt2910[m]>
PulkoMandy: Yes but it's not so clean when I have to do it in various places in a source tree.
<trungnt2910[m]>
It's just more unnecessary code to fix a problem caused by unnecessary code.
<nielx[m]>
Why is it not an issue when building 32 bit? Does whatever you work on expect it to be defined on 32 bit already?
<trungnt2910[m]>
nielx[m]: __cdecl and __stdcall do have meanings on 32-bit.
<trungnt2910[m]>
They're calling conventions specific to 32-bit i386.
<trungnt2910[m]>
Invented by Microsoft and somehow followed by BeOS.
<trungnt2910[m]>
On 64-bit i386 and non x86 architectures these two have no meaning.
<nielx[m]>
I understand that. I am just trying to understand why this would be an issue on 64 bit for whatever you are doing, but not on 32 bit
<jessicah>
I think it's fine to get rid of them
<jessicah>
For the limited beos source archives that need it, better to patch them, and has no abi issues
<jessicah>
nielx[m]: probably on 32-bit, it's not an ignored attribute, so no warning?
<jessicah>
Easy enough to test
<nielx[m]>
fair enough
<trungnt2910[m]>
jessicah: Yeah that was what I was trying to say...
<botifico-c849d97b>
[haikuports/haikuports] kenmays 7632d34 - libreoffice: bumped to 7.3.7.2 (#8613)
<x512[m]>
FUSE was probably not considered when initial userlandfs code was written.
<trungnt2910[m]>
Is there anything (except lack of interest) that is preventing TIOCEXCL, TIOCINQ, and TIOCOUTQ from being implemented in Haiku?
cocobean has joined #haiku
<Begasus>
jessicah, you're a day ahead of us? ;) "Commits on May 9, 2023"
<trungnt2910[m]>
Yeah...
<trungnt2910[m]>
She's 2 hours ahead of me and it's already 10PM here.
<trungnt2910[m]>
s/already/nearly/, s/10PM/11PM/
<Begasus>
almost 3PM here :)
<cocobean>
begasus: Testing LO 7.3.7.2 / Haiku hrev56974 - All systems are go.... (PASSED)
<jessicah>
Yes, it's the ninth today
<jessicah>
NZ in the future :)
<Begasus>
let's wait for the buildmaster cocobean (could take a while with gcc11 being build twice) :)
<x512[m]>
trungnt2910: What is a purpose of TIOCEXCL, TIOCINQ, and TIOCOUTQ?
<x512[m]>
Is it a part of POSIX?
<trungnt2910[m]>
x512[m]: `TIOCEXL`: Places the current TTY into exclusive mode, so that no one except root (oops probably not so useful on Haiku) can re-`open` the TTY.
<trungnt2910[m]>
`TIOCOUTQ`: Gets the number of output bytes buffered.
<trungnt2910[m]>
`TIOCINQ`: Gets the number of input bytes available.
<trungnt2910[m]>
x512[m]: Not a part of POSIX (POSIX doesn't seem to specify `ioctl` values).
<trungnt2910[m]>
All of these are included in FreeBSD and not just GNU-only extensions.
<x512[m]>
trungnt2910: While you are at modifying GCC configuration: can you turn off -fPIE (make it do the same as -fPIC)?
<trungnt2910[m]>
Sorry, it's late now and I have uni tomorrow.
<trungnt2910[m]>
Unless it's an obvious fix.
Diver has joined #haiku
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
bbjimmy has joined #haiku
Dio9sys[m] has quit [Quit: Client limit exceeded: 20000]
zard has joined #haiku
Skipp_OSX has joined #haiku
Anarchos has joined #haiku
dqk has quit [Ping timeout: 480 seconds]
<Anarchos>
since this morning (french time) i get kdl at boot : usb ohci timeouts, acpi timeouts, "did not find boot partitions", i have to put some debug options to get haiku booting. Do anybody experienced that ?
<Skipp_OSX>
I'll check the latest nightly see if there are problems booting
<Anarchos>
thx
<cocobean>
Anarchos: hrev56974 ?? I've ran it for two hours with no issues yet.
<Anarchos>
yes
<Anarchos>
seems i got bfs corruption
<Skipp_OSX>
yeah hrev56974 boots fine here on my Lenovo X220
<Skipp_OSX>
BFS corruption is the worst I've had that happen to me before
return0e[m] has quit []
<Anarchos>
Skipp_OSX: yeah i too had my faire share of bfs corruption (i even got axel to modify checkfs...)
<Skipp_OSX>
hopefully the bugs get worked out before R1
<Anarchos>
Skipp_OSX: it is strange, i got no bfs trouble since 2017...
<Skipp_OSX>
sounds like you were due for some then
<Anarchos>
not sure how to look to solve them... i ran checkfs . after that , i got two instances of a partition ...
<Anarchos>
and error messages 'bad pheader', no idea what it is
<Skipp_OSX>
presumably some bfs data structure
<Skipp_OSX>
possibly related to pthreads
cocobean has quit [Quit: Page closed]
dqk has joined #haiku
linuxmaster has quit [Remote host closed the connection]
bbjimmy_64 has quit [Quit: Vision[]: i've been blurred!]
freddietilley has quit [Quit: WeeChat 3.8]
freddietilley has joined #haiku
randomuser86 has joined #haiku
* Anarchos
wonders if someone experienced bfs corruptions recently
bbjimmy_64 has joined #haiku
randomuser86 has quit []
Anarchos has quit [Remote host closed the connection]
gouchi has joined #haiku
frkzoid has quit [Ping timeout: 480 seconds]
<zdykstra>
I just upgraded to hrev56974, so far so good.
freddietilley has quit [Quit: WeeChat 3.8]
Diver has quit [Read error: Network is unreachable]
<PulkoMandy>
What would pthreads do in a filesystem? More likeliy it means "partition header" or something like that?
<Skipp_OSX>
ah yes that could be
dqk has quit [Read error: Connection reset by peer]
<waddlesplash>
trungnt2910[m]: that userlandfs KDL is a bug in userlandfs for sure
<waddlesplash>
nielx[m]: Qt Creator. but it sometimes has odd performance issues around code completion on Haiku that it doesn't have on other OSes
<waddlesplash>
PulkoMandy: pheader is elf
<waddlesplash>
sounds like Anarchos may have the MSI interrupts problem
<Halamix2>
If I get nightly image and update it, will it be the same as installing R1B4, adding master repo and running SoftwareUpdater? Ethernet connection for my PC was added in hrev56740 and that's one thing I need
<augiedoggie>
you can switch from beta to nightly, i think there is a guide on the website