Was just skimming through but seems like you gotta really make the code ugly for an alien compiler
#uACPI - a portable and easy-to-integrate ACPI implementation
real jetty
slim panther
not yet i went to bed after that
rustic compass
Maybe wait until they added c23? To the fork?
slim panther
Maybe wait until they added c23? To the fork?
they dont even have c95
jaunty fox
they do have c23 binary constants anyway https://github.com/open-watcom/open-watcom-v2/commit/ef4cac7bb2d4cc5c0d8d72c90e80f0aff7cb8940
hollow elm
they do have c23 binary constants anyway https://github.com/open-watcom/open-wat...
yeah but that requires using a recent build of ow
and not one released in the 90s
jaunty fox
and not one released in the 90s
surely to god rdos isn't using the pre open watcom?
hollow elm
surely to god rdos isn't using the pre open watcom?
probably not since they did mention open watcom specifically, but if the build they're using doesn't have stuff it's supposed to have then it must be some old one
slim panther
still don't understand them using openwatcom to begin with
frank canopy
i actually have encountered some scenarios where openwatcom is the recommended solution though idk if rdos is doing any of them
but the general thing ive heard is that openwatcom is way less bad at real mode compilation than gcc or clang
so it might be natural to then use one compiler for everything and if you have real mode parts that need compiled for some reason that lends itself to wanting openwatcom
id just. not compile to real mode. but thats just me ig lmao
no clue why rdos in particular is using openwatcom though
gentle peak
i believe rdos doesn't use a flat memory model?
thats the impression ive gotten from random messages ive seen on the osdev forums at least
frank canopy
oh yeah that might do it
the reason ive seen given to use openwatcom for real mode compilation is it handles non-flat segmented memory models better
slim panther
idk but the forum post mentions 64-bit support
fiery turtle
Took a few hours, but I can now compile kernel C files on the host machine, even on Windows. It re-creates the kernel source tree with some files copied verbatim and some containing fake test stubs, thanks to cmake magic it even automatically re-copies them if the source was changed
i made it so i can specify the kernel files I want to copy to the synthetic kernel tree
Now i can start writing tests finally
very nice
tbh the worst part was stubbing shit for MSVC, but its very convenient to debug stuff with the msvc debugger in visual studio
fiery turtle
also fixed a small uacpi bug that monkuous found a while back
slim panther
we should try running uacpi on ps4 
fiery turtle
id like to see a dump from it
slim panther
i dont have one sadly
fiery turtle
i have one but its not hacked
calm latch
i dont have one sadly
I have a hackable one
slim panther
zamn
calm latch
Does it have ACPI?
calm latch
I thought it wasn't a PC
slim panther
well
it isn't an ibm compatible pc
but it has hpet
and tsc
it's a real oddball of a system though
hollow elm
it has ahci as well i think
iirc only for the bluray drive tho, the hdd is over an usb-sata adapter 
hollow elm
it has an arm co processor which it exposes via pci??
afaicu it's an arm soc that also has a pcie endpoint on one side that has access to it's entire address space (and vice versa, mapping the host memory into the soc address space)
hollow elm
wouldn't be surprised if it was almost just a pc
slim panther
i cant find anything about linux on it
hollow elm
maybe without legacy ibm pc peripherals
but without insane stuff like the arm soc as a southbridge
slim panther
"""""southbridge"""""
hollow elm
i cant find anything about linux on it
hm can't find anything either, apart from articles that did s/xbox/xbox one/g for some reason
slim panther
no way there's no homebrew on the xbox
i guess there isn't any incentive because of dev mode
hollow elm
🤯 most people won't hack your system if you let them run their own code on it
hollow elm
🤯 most people won't hack your system if you let them run their own code on it
it's like what happened with the ps3 :^)
slim panther
are there any consoles out there which could be feasible for hobby osdev
hollow elm
anything with webkit is just a ticking time bomb so theoretically yeah
slim panther
wii u
hollow elm
ps4/ps5
calm latch
are there any consoles out there which _could_ be feasible for hobby osdev
Dreamcast
hollow elm
although the switch is just a nvidia tegra x1 more or less so you could just buy a dev board instead :^)
slim panther
most of the 1st gen ones have broken down
calm latch
I have Wii U which I need to hack anyway
hallow marten
Wii is doable
calm latch
Since its drive broke
slim panther
i have a wii u
hallow marten
Has been around for a while so more documentation most likely
slim panther
which is still fully functional
hallow marten
Kinda funny the Wii U was a flop but I know lots of people who have one me included
slim panther
or 5
jaunty fox
are there any consoles out there which _could_ be feasible for hobby osdev
it depends on whether you like powerpc or not
i think it's high time for a powerpc hobby os here
slim panther
i gotta do device tree support
calm latch
is there devicetree on wii/wii u?
fiery turtle
wtf is this PR
it disables logging on openwatcom entirely, adds random tabs, tons of ifdefs and whatever this is https://github.com/uACPI/uACPI/blob/4cc7489483ecb68e44ca8cbee07aecd6cd33eae0/rdos/acpi.wpj
what have i brought upon myself
fiery turtle
0
i would be fine accepting openwatcom support if it wasnt that invasive, but holy shit at least clean up your pr
gentle peak
can watcom really not handle that? that's been standard c since C fucking 90
gentle peak
also i'm guessing (hoping) the logging disable thing is because C90 doesn't have vararg macros and watcom doesn't support it?
fiery ice
can watcom really not handle that? that's been standard c since C fucking 90
watcom can't even handle C99 stuff
lol
gentle peak
yea but this is older than C99
fiery ice
it should at least support it
slim panther
can openwatcom do scanf 
gentle peak
like C90 literally defines assert-without-ndebug in terms of that construct
fiery turtle
also i'm guessing (hoping) the logging disable thing is because C90 doesn't have...
Like what the actual fuck
fiery ice
we are in 2025 why use old compilers that don't even work properly?
slim panther
exactly
fiery turtle
https://github.com/uACPI/uACPI/issues/122#issuecomment-2872369004
@calm latch can you teach them how to make a proper microkernel pls
slim panther
can you ask him that
gentle peak
What The Fuck
slim panther
why no gcc
gentle peak
Just use the fucking _Pragma
fiery ice
Just use the fucking _Pragma
doesn't support it
slim panther
it doesn't have that
gentle peak
it does
fiery turtle
What The Fuck
Did he just apply that to literally everything
slim panther
slim panther
bruh
gentle peak
Did he just apply that to literally everything
yep
slim panther
again, just don't support it
tell him to use a c99 compiler
I'd bet even fucking tcc can compile uacpi
kindred beacon
also lol @ the random rdos directory
fiery turtle
Like the thing is it can be supported cleanly, but wtf is this slop
real jetty
Lmao at always inline
fiery ice
also lol @ the random `rdos` directory
yeah, wtf is that file
gentle peak
this does not have the same semantics!!! (unless those are the only fields in that struct but i don't think so)
fiery turtle
this does not have the same semantics!!! (unless those are the only fields in th...
They aren't
Like wtf
gentle peak
yeah, wtf is that file
it looks like some bespoke serialization format
fiery ice
oooof
real jetty
Yeah i was about to say lol
kindred beacon
I'm sorry but this PR is just low-quality and should stand no chance unless fundamental changes are made IMHO
fiery turtle
He didn't even bother abstracting it behind platform.h
Just slap everything under ifdefs
kindred beacon
going around and bending backwards to a compiler that doesn't support stuff like FUCKING VARIABLE DECLARATIONS in the middle of blocks is insane
slim panther
^
fiery turtle
It doesn't even support them at the begining
slim panther
just say min requirement is C99
fiery turtle
Only at function entry
real jetty
Yeah.. and youd need to keep maintaining it with this burden of a compiler always in mind
slim panther
true
fiery turtle
Exactly
fiery ice
and also, uACPI in userspace?
fiery turtle
Works fine if your kernel is sane
slim panther
and also, uACPI in userspace?
that's fine
fiery turtle
Like @calm latch
fiery ice
that's fine
can it call io ports directly or it needs syscalls?
fiery turtle
There are multiple ways on x86
gentle peak
can it call io ports directly or it needs syscalls?
depends on the kernel, but x86 has two mechanisms for allowing userspace to access io ports
iopb (which allows you to specify which io ports can be accessed) and iopl (which gives access to all io stuff, including disabling/enabling interrupts)
calm latch
can it call io ports directly or it needs syscalls?
You can also make it a syscall, it's not a big deal
fiery ice
k
calm latch
Like io doesn't have to be fast
gentle peak
that too, io port access is incredibly slow anyways
fiery ice
i'll keep my uACPI on kernel space
fiery turtle
Depends on your design
calm latch
https://github.com/uACPI/uACPI/issues/122#issuecomment-2872369004
Why are spinlocks not necessary?
gentle peak
because it's single-threaded, presumably
kindred beacon
thank me later
shit like this gets me real mad
even if I obv don't call any shots there
kindred beacon
HAHAHAHAHA tests fail
fiery turtle
he managed to break CI as well
lmfao
/Users/runner/work/uACPI/uACPI/source/tables.c:981:5: error: call to undeclared library function 'memcpy' with type 'void *(void *, const void *, unsigned long)'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
memcpy(id.signature.text, signature_string, 4);
^
/Users/runner/work/uACPI/uACPI/source/tables.c:981:5: note: include the header <string.h> or explicitly provide a declaration for 'memcpy'
to really top it all off, it breaks compile for sane compilers
gentle peak
or gcc
gentle peak
because this is not a clang-specific error, the only reason macos is the one that failed is because it was fastest
i wonder if msvc even allows that when compiling with reasonable options
also, shouldn't that be using uacpi_memcpy instead of memcpy?
fiery turtle
lets see
calm latch
If they're doing it in userspace anyway, why not use a sane compiler?
fiery turtle
also, shouldn't that be using uacpi_memcpy instead of memcpy?
true actually
kindred beacon
I mean realistically memcpy is always going to be there anyways
for gcc/clang at least
slim panther
nobody uses those
kindred beacon
ah sorry forgor
fiery turtle
i wonder if msvc even allows that when compiling with reasonable options
D:\a\uACPI\uACPI\source\tables.c(981,5): warning C4013: 'memcpy' undefined; assuming extern returning int [D:\a\uACPI\uACPI\tests\runner\reduced-hw-build\test-runner.vcxproj]
kind mantle
nobody uses those
hm yes nobody uses GCC or clang
Lily-CC is clearly the better option
fiery turtle
he added that memcpy himself btw
its not in the original uacpi source
so he didnt even check his code
gentle peak
well watcom probably allows it
fiery turtle
msvc is still insane
luckily i have warnings as errors
kind mantle
luckily i have warnings as errors <:trl:829316641191886888>
For real though adding -Werror=extra to my library projects is the best thing I've ever done
fiery turtle
well watcom probably allows it
he said
The op code tables fail to compile, and I have no idea why but I suspect it is related to the incredibly complex macro structure used. I will need to redo this to decrease complexity, and in fact, also to increase readability (and debuggability) of the code.
so i guess he never actually compiled it
real jetty
Ah how nice it is from him to increase readability
gentle peak
(also, the opcode table issue is probably due to the [] = ... syntax, since i believe that's C99)
gentle peak
love that he (implicitly) complains about readability and then prs this
fiery turtle
lmao
gentle peak
oh wait .wpj probably stands for watcom project doesn't it
this is just an ide config file
kind mantle
Why exactly are people seriously using the watcom compiler if it doesn't support basic shit like compound initializers?
gentle peak
Why exactly are people seriously using the watcom compiler if it doesn't support...
it supports several things mainstream compilers don't, like segmentation
and real mode (proper real mode, unlike 80386-compatible-running-in-real-mode)
fiery turtle
also he hardcoded pointer_size as 4, does it not support 64-bit x86?
jaunty fox
Why exactly are people seriously using the watcom compiler if it doesn't support...
it does but they didn't complete it
kind mantle
it supports several things mainstream compilers don't, like segmentation
Inb4 Lily-CC competes with watcom by supporting segmentation but also c17
vast kestrel
i guess there isn't any incentive because of dev mode
That and because it's one of the most locked down PCs in existence, the bios is fully signed, both ram and pcie traffic is encrypted, it has full iommu, it uses all the fancy hyperv based VBS, it uses different certificates than the rest of the windows ecosystem.
fiery turtle
how do they not support such basic things if its an actively maintained compiler
jaunty fox
it seems rdos guy is not enabling C99 compile mode, so he doesn't want things like e.g. VA_ARGS
gentle peak
most of the bs could've been avoided by simply enabling a compile mode??
jaunty fox
nor stdint.h nor many other things
kind mantle
it seems rdos guy is not enabling C99 compile mode, so he doesn't want things li...
That seems extremely hard to work around with any modern C project
fiery turtle
it seems rdos guy is not enabling C99 compile mode, so he doesn't want things li...
wtf?
jaunty fox
wtf?
that's my inference as to why he'd strip out the log macros
also why he hoisted declarations to top of blocks
fiery turtle
Why not enable it?
jaunty fox
presumably he builds in c89 mode only
fiery turtle
Maybe he doesnt know about it
kind mantle
It is beyond me why any project developed today supports any compiler older than C99
jaunty fox
he doesn't like C in any case: https://www.rdos.net/rdos/cversus.htm
kind mantle
he doesn't like C in any case: https://www.rdos.net/rdos/cversus.htm
I wish him good luck then
fiery turtle
I can tell by the quality of his PR
gentle peak
this is not true unless you use watcom or something similar
gentle peak
this is not true if you use a mainstream compiler like gcc/clang (because i don't count msvc as mainstream, and thus the two mainstream compilers support eachothers extensions for the most part)
gentle peak
which watcom does btw
that's surprising
kind mantle
He's thinking of reaaaaaally old compilers, so old that none of them actually exist today
jaunty fox
that's surprising
it's an ancient optimisation
kind mantle
I seriously doubt he can write assembly as fast as from GCC/clang
jaunty fox
but watcom
i don't stand for watcom slander, it's a fantastic compiler which unfortunately needs significant work on its modern c/c++ support
gentle peak
wait so why did he get rid of all of the binary literals
is he using an ancient version of it or something?
fiery turtle
i guess so
kind mantle
You know, with all the modern langs it's cool to hate like Rust and Zig, I wasn't expecting someone that seriously thought C was slow.
gentle peak
kind mantle
lmao
fiery turtle
this guy lmfao
kind mantle
There's good reason why modern OSes use paging
jaunty fox
the greatest hit is still the neanderthal aspergers theory
kind mantle
bruh what
gentle peak
i really hope this is just a ux thing
kind mantle
as well as discovering mind-to-mind communication
MF what scientology shit is this
fiery turtle
@jaunty fox do u still have a link to the commit that added 0b support?
jaunty fox
fiery turtle
Thanks, ill add this to my reply
it literally supports this in C99 + extensions mode
calm latch
If they're doing it in userspace anyway, why not use a sane compiler?
Especially since there is no ACPI on i386
kind mantle
WTF is this guy's problem with autism
Why does he want to paint us as some form of troglodytes
dense steppe
https://www.rdos.net/rdos/segment.htm what
Send a mail
Someone should refute his argument and send him a mail.
calm latch
how many people use uACPI with clang vs GCC?
fiery turtle
probably most use gcc
fiery ice
i use clang
north hull
clang!
calm latch
I mean me too
Oh, it's also him who's spreading misinformation about interrupts in userspace https://forum.osdev.org/viewtopic.php?p=347483#p347483
lofty dragon
i use both
kindred beacon
good that PR was just insane
calm latch
Does watcom not have atomic intrinsics as well?
calm latch
Wait
calm latch
only for global lock i think
Does this not use mutex?
fiery turtle
mutex between uacpi and firmware?
calm latch
No, the kernel API
fiery turtle
the global lock is a firmware thing
gentle peak
he replied
I agree that the rdos directory should not be there. Cannot find out how to remove these.
???
fiery turtle
yeah basic git is very hard
fiery turtle
Does this not use mutex?
calm latch
the global lock is a firmware thing
Oh, ok
calm latch
what the actual fuck
"i386 is the only architecture"
mortal yoke
also wtf #define uintptr_t unsigned int *
fiery ice
cursed stuff
fiery turtle
also wtf `#define uintptr_t unsigned int *`
what the hell
mortal yoke
that's also going to break any kind of arithmetic done on it, though you probably don't do any in uacpi?
jaunty fox
prob for watcom
they have at least these
fiery ice
wth are those libs
jaunty fox
targets
fiery turtle
that's also going to break any kind of arithmetic done on it, though you probabl...
probably not but still
fiery ice
oof
fiery turtle
they have at least these
so it does support 64-bits
fuck uacpi abstractions, include libc directly
fiery ice
string.h is memcpy?
calm latch
I think so
mortal yoke
yes memcpy is in string.h
jaunty fox
at this point frankly it would be better if someone else made a reasonable port to watcom
mortal yoke
or just don't bother and tell them to fuck off lol
calm latch
But string.h is not even related to watcom
fiery ice
preferably someone that knows what they are doing
calm latch
or just don't bother and tell them to fuck off lol
Again, why can't they use a different compiler, considering they're not even linking it with the kernel
fiery turtle
at this point frankly it would be better if someone else made a reasonable port ...
yeah this is just astonishing incompetence, from the lack of basic git knowledge to not knowing how watcom works (e.g. that it has _Pragma etc)
jaunty fox
Again, why can't they use a different compiler, considering they're not even lin...
just passing -zastd=c23 would probably eliminate 90% of the problems
or even zastd=whatever sets c99 + extensions like binary literals
fiery turtle
how do u write software for so many years and still lack basic skills like that
frank canopy
dear lord
calm latch
But that's not their case?
jaunty fox
mr rdos seems to be somewhat adapted to the demands of mostly assembly with a little c programming around 30 years ago
mortal yoke
lol
fiery turtle
also its supposed to be a POS os?
calm latch
who uses it?
fiery ice
i hope it has been tested VERY troughly then
dense steppe
so the difference in semantics between struct mystruct foo = { .x=1, .z=3 }; and struct mystruct foo; foo.x=1; foo.z=3; is that the former zero-initializes foo.y, while the latter leaves it uninitialized and therefore UB to use?
gentle peak
yes
dense steppe
this is even more cursed if you know Swedish
slim panther
how does this guy not know that difference
mortal yoke
i hope it has been tested VERY troughly then
I mean idk who in their right mind would buy it to begin with so it doesn't really matter how garbage it is
calm latch
Though I guess he has a bunch of drivers and network support
jaunty fox
this is even more cursed if you know Swedish
what means it in Swedish?
fiery ice
I mean idk who in their right mind would buy it to begin with so it doesn't real...
some comepany that wants cheap stuff
prob
dense steppe
what means it in Swedish?
"wälcome to reälity"
fiery ice
just use linux
well, we are talking about corporations, they are very stupid sometimes
dense steppe
"wälcome to reälity"
apart from that it's pretty reasonable language-wise, I guess
fiery turtle
they arent dumb enough to pay for a gpl os lol
jaunty fox
it's $10 a licence so it's no great cost
fiery ice
i saw POS with literal windows on them...
dense steppe
apart from that it's pretty reasonable language-wise, I guess
although it seems translated through e.g. google translate
fiery turtle
it's $10 a licence so it's no great cost
lol
pine leaf
i cant find anything about linux on it
3 hours too late, but yeah, nobody has fully pwned any Xbox since the 360, the original Xbox One (2013) even is still not hacked yet
even the 360 is debatable, you need hardware mods to run Linux
fiery turtle
they really dont want you to freely use the hardware do they
pine leaf
indeed, they actually are
they basically achieved perfection with the XBOne, and then just did that but again™️ for One S/X and Series S/X
fiery ice
they really dont want you to freely use the hardware do they
bc they are consoles, they want you to pay for multiplayer
jaunty fox
meanwhile PS3 was sold as an ideal machine to run linux on
fiery turtle
meanwhile PS3 was sold as an ideal machine to run linux on
like it was so cool u could dual boot linux and the ps3 os for gaming and work at the same time
but no fun allowed anymore
mortal yoke
bc they are consoles, they want you to pay for multiplayer
also they want to prevent piracy at all costs because it loses them money
which is probably the biggest reason why its all so locked down tbh
dense steppe
indeed, they actually are
they basically achieved perfection with the XBOne, and...
say that eg. you achieve arbitrary code execution in the context of a game (e.g. with some heap vuln and you are a really good pwner, so you can manage to get anything at all inthere)
is this something which has been done? how is process isolation etc. on the xbox (is it just NT)?
fiery ice
they are just greedy
fiery ice
also they want to prevent piracy at all costs because it loses them money
yeah, and also make you pay 90€ for a shitty game
calm latch
say that eg. you achieve arbitrary code execution in the context of a game (e.g....
They run stuff in a hypervisor
pine leaf
say that eg. you achieve arbitrary code execution in the context of a game (e.g....
then you own the Game VM at that point
you still don't get full hardware access
dense steppe
ahh
fiery ice
you literally can pay double at some points for console games vs pc one
dense steppe
so we gotta do hypervisor pwn
pine leaf
yes, except that you can't see it
calm latch
you literally can pay double at some points for console games vs pc one
Because consoles are sold for cheap, but you get revenue on the games
pine leaf
even at the bus level
fiery ice
Because consoles are sold for cheap, but you get revenue on the games
cheap?...
pine leaf
all data going in or out of the APU is encrypted
only time plaintext data exists is inside the APU silicon
fiery turtle
then you own the Game VM at that point
you still don't get full hardware access
so u need a huge chain of vulnerabilities basically?
fiery ice
1k$ ps5 pro that is literally the base 500$ version?
calm latch
Like a gaming console should cost much less than an equally powerful PC
fiery ice
Like a gaming console *should* cost much less than an equally powerful PC
they cost way more tho
calm latch
1k$ ps5 pro that is literally the base 500$ version?
New GPUs alone cost the same as consoles, or more
fiery ice
New GPUs alone cost the same as consoles, or more
gpus are literally cheap to get for them
dense steppe
so u need a huge chain of vulnerabilities basically?
an unreasonably long chain of vulns, and you need to somehow develop a working hypervisor exploit with no eyes
calm latch
they cost way more tho
The PS3 for example was sold for a loss
fiery ice
- they also lock them to 30 fps on most games
calm latch
(a big loss)
fiery ice
you get 60fps only for some
pine leaf
so u need a huge chain of vulnerabilities basically?
if you could find some entryway into the hypervisor from inside gameos or systemos, then yes
but nobody has exploited gameos yet, and there has only been 1 exploit in systemos, and it was patched immediately
fiery turtle
if you could find some entryway into the hypervisor from inside gameos or system...
can u explain what those are?
gameos etc
pine leaf
the chances of them ever being hacked is low tbh
there's just no drive to get into them, they're just mid PC hardware
nothing funny like PPC or ARM
calm latch
Wait, the pull request has disappeared again?
frank canopy
my guess is gameos is the os that runs under hypervisor and runs the game software
so the game software doesnt have to include an entire os
pine leaf
gameos etc
GameOS is where games run
SystemOS is where the "systen" UI and services run (chat features, dashboard, etc)
iirc there's one more VM running in the background too but I don't remember what it's called / does
all of that is running inside a tiny hypervisor that gatekeeps everything
each one is an independant copy of Windows
dense steppe
so you always have hw<->hypervisor<->(one of gameos and systemos)
fiery turtle
so its nested virtualization?
dense steppe
each one is an independant copy of Windows
that's why the games take so long to download??
calm latch
does openwatcom not support (void)x; for unused variables?
dense steppe
so its nested virtualization?
probably not nested
gameos and systemos probably communicate with each other via the hypervisor
pine leaf
so its nested virtualization?
not actually nested, no
it's multiple VMs running under one HV
not VMs inside of VMs
slim panther
does openwatcom not support (void)x; for unused variables?
lol
fiery turtle
using whpx?
fiery turtle
so its a new kernel?
pine leaf
iirc the one on the 360 was like, 64KB or something
dense steppe
it'd be funny if it was xen or something
fiery ice
windows runs under a HV too
pine leaf
so its a new kernel?
kinda
hard to really call it a kernel
dense steppe
windows runs under a HV too
(optionally)
fiery ice
when you use wsl
pine leaf
it's sole job is to run the VMs and make sure they don't try anything funny
pine leaf
iirc the one on the 360 was like, 64KB or something
I don't think we know how big the one on the Xbox One / Series line is
since it hasn't been exploited yet
nobody's dumped it
fiery ice
maybe it is in some leaks?
pine leaf
has M$ even had a leak recently, they haven't that I've heard of
last one I know of was the OG Xbox + NT4 + NT5 source leak
pine leaf
title suggests nothing Xbox related
I'm not gonna go check it myself, but I'd doubt it
anything Xbox-related would probably also be modern Windows-related, and that'd make a huge splash
fiery ice
anything Xbox-related would probably also be modern Windows-related, and that'd ...
i think the sdk got leaked
fiery ice
wow, so useful, the same public Windows SDK but with 5 different DirectX functio...
and i think the code for the xbox os
fiery turtle
BROOOOOOOOOOOOOO
lmfao
fiery ice
but only the original one
pine leaf
it's so similar that people have made compatibility layers w/ DLL injection, that make it just work™️ on regular Windows
fiery turtle
git is really hard what can u do
kindred beacon
you're welcome
fiery ice
git is really hard what can u do
using a normal compiler is too
calm latch
but 64-bit locked operations are not supported in 32-bit mode, and the atomic inc and dec cannot return the real value as an atomic operation.
Huh?
gentle peak
i386 doesn't have xadd, is he talking about that?
calm latch
i386 doesn't have ACPI either
gentle peak
well yeah but try telling that guy his acpi module will have to be i486+ at the very least
i586 if uacpi needs cmpxchg8b
calm latch
i586 if uacpi needs cmpxchg8b
I don't think it does?
fiery turtle
he should compile his acpi driver in 16-bit mode
gentle peak
pretty sure some of the cpus he's targeting don't have FP at all (yes these systems won't have acpi but above comment applies)
calm latch
And you can emulate the rest?
gentle peak
also 8-byte fp ops are only atomic on >=pentium iirc
fiery turtle
only 32 is used atm
gentle peak
xadd
slim panther
xdd
fiery turtle
xd
calm latch
x
fiery ice
⠀
fiery turtle
@kindred beacon lol thanks for the comment again, i dont have the patience to explain things like that
id rather just ignore it until it goes away
lmao
gentle peak
idk at this point because he made a bunch of changes to things that watcom just supports
gentle peak
The reason for these fixes is that the structures are only partly initialized. I cannot see the difference between declaring a variable and then initializing the same (partial) fields.
bro
fiery turtle
because it does support a lot of these things from what i can see
gentle peak
he's using some stupid version of watcom
i looked at his website earlier and it suggests he's using a version from jan 2025
fiery turtle
> The reason for these fixes is that the structures are only partly initialized....
btw buy my kernel
gentle peak
> The reason for these fixes is that the structures are only partly initialized....
this guy just does not know C
fiery turtle
i looked at his website earlier and it suggests he's using a version from jan 20...
yeah idk at this point
kind mantle
Casually includes stddef.h and still defines u?int[0-9]+_t
gentle peak
uintX_t is from stdint.h
kind mantle
i know
gentle peak
which doesn't exist in C90
slim panther
size_t
fiery turtle
apparently slapping an include <string.h> is fine but including stdint is not
kind mantle
This just gets stupider by the second
slim panther
priorities
it's a freestanding header, you can just copy paste it smh
i double dare yall to just tell him to use clang
kind mantle
??????
This does nothing other than break things
slim panther
???
nah trust me, this makes sure the compiler doesn't OOM while allocating the var name
mortal yoke
i double dare yall to just tell him to use clang
would be funny but also just ignore them and hope they will stop when nobody is responding to them
calm latch
were there 486s with ACPI?
kind mantle
tbh seeing how awful this person prolly is with views on autism, just ban from uACPI GitHub if he opens PRs again
fiery ice
why even target that ancient hardware anyway
fiery ice
tbh seeing how awful this person prolly is with views on autism, just ban from u...
wait what, where his views of autism?
kind mantle
Just go to his site's homepage
slim panther
every single time
kind mantle
I do not appreciate that and in my projects someone would be blanketly banned just for that.
Probably among other things, he tries to link autism to neanderthals, a species we combined with a very long time ago
kind mantle
He calls us "aspies" 💀
jaunty fox
i don't know if watcom has it
calm latch
Even if it doesn't, your C library should provide it
kind mantle
why even put it there
Bigots cannot afford to keep their mouths shut when they also have the option of hating on people
calm latch
otherwise what are you doing
fiery turtle
i left a comment on the PR, hopefully it makes it a bit more clear
but its basically what leo said but nicer
jaunty fox
Bigots cannot afford to keep their mouths shut when they also have the option of...
i don't think he's bigoted, i think he is just practicing pseudoscience
fiery ice
like ppl on X
fiery turtle
but its basically what leo said but nicer
i didnt want to offend him and stuff lol
kind mantle
i don't think he's bigoted, i think he is just practicing pseudoscience
Doesn't seem like a person I'd want to be around either way.
RDOS source code is distributed as GPL, except for commercial usage.
Not how GPL
frank canopy
i don't think he's bigoted, i think he is just practicing pseudoscience
considering how that stuff was created and by who idk theres much of a meaningful difference you can draw
fiery turtle
ah yes, the GPL-capitalism clause
regal snow
what happend
calm latch
Delucional
jaunty fox
this for instance is just wild speculation
fiery ice
oof
calm latch
what happend
A PR
kind mantle
A PR <:KEKW:913708372329644042>
A disastrous PR
fiery ice
a very very bad PR
mortal yoke
Delusional
also you are allowed to use it commercially for free as long as what you are trying to run it on isn't an embedded system
fiery turtle
basically a very incompetent person trying to port uacpi to openwatcom
kind mantle
basically a very incompetent person trying to port uacpi to openwatcom
A weird subset of openwatcom*
vast kestrel
At this point he is trying to port uacpi to whatever compilation it does
fiery turtle
yeah thats the thing
frank canopy
whats worse is this is almost tempting me to try and do it properly. and i barely know C and dont use openwatcom
fiery turtle
its not even an openwatcom port
fiery ice
basically a very incompetent person trying to port uacpi to openwatcom
wait, from when GTNH has a tag??
fiery turtle
its a port to his crappy os buildsystem
vast kestrel
It's like passing to GCC --be-stupid and porting to that
fiery turtle
whats worse is this is almost tempting me to try and do it properly. and i barel...
lmfao
kind mantle
It's like passing to GCC --be-stupid and porting to that
gcc -Dint=float
vast kestrel
Y e s
fiery turtle
wait, from when GTNH has a tag??
idk lol
frank canopy
i literally know enough c to make zig bindings for uacpi and thats it
fiery ice
lol
frank canopy
and this is tempting
vast kestrel
-Duintptr_t="unsigned int*"
gentle peak
The problem is that nobody wants to maintain openwatcom support
kind mantle
I know enough C to know that this rdos person doesn't know C.
gentle peak
Just adding support for it now seems simple enough
jaunty fox
whats worse is this is almost tempting me to try and do it properly. and i barel...
i have felt some temptation myself though i don't know a great deal about openwatcom (i have used it once in the past and i have a high opinion of it as a compiler despite its lag in modernising)
gentle peak
But that's gonna bitrot insanely quickly
fiery turtle
Just adding support for it now seems simple enough
if he adds a CI target ill maintain whatever breaks, but hes obviously not going to get that far lol
calm latch
Btw, can't uACPI just use atomic.h through some flag or something?
fiery turtle
hm?
calm latch
So people with weird compilers can just be directed to implementing that
kind mantle
Btw, can't uACPI just use atomic.h through some flag or something?
Do you mean stdatomic.h ?
calm latch
Yeah
gentle peak
Yknow I was gonna try myself but $ pacman -Ss watcom $
jaunty fox
what is shocking is he isn't even familiar with watcom's capabilities around e.g. struct packing, which i found by ctrl-f the manual
fiery turtle
atomic.h in uacpi is in platform/, its meant to be overriden by your platform unless you're lazy and use whatever is shipped by default
so yeah
gentle peak
So people with weird compilers can just be directed to implementing that
The problem is when the compiler doesn't support aspects of regular C syntax
Fixing that requires editing the main source
gentle peak
Yknow I was gonna try myself but ```
$ pacman -Ss watcom
$ ```
It's in AUR but that's yucky
kind mantle
Yeah but we found that lots of uACPI source code is supported by owc
calm latch
But uACPI is relying on compiler-specidic functions for atomics
fiery turtle
what is shocking is he isn't even familiar with watcom's capabilities around e.g...
ikr, this is scary incompetence for an open watcom shill
calm latch
But if your compiler doesn't support stdatomic, you can implement it yourself
fiery ice
idk
fiery turtle
But if your compiler doesn't support stdatomic, you can implement it yourself
i mean
calm latch
But if your compiler doesn't support stdatomic, you can implement it yourself
And is not one of the typically used ones
fiery turtle
the docs say it works for msvc gcc and clang, for anything else u gotta do it yourself
gentle peak
Fuck it I'm gonna see how much of the BS comes from watcom and how much comes from this guy
calm latch
What if it turns out to be a decent compiler? 
leaden fox
this for instance is just wild speculation
Bro you don't get it us Aspies are close relatives to Neanderthals man
leaden fox
I channel the neanderthal spirit through my sutism
gentle peak
Why is there an uninstal.exe in the linux package
$ file /opt/watcom/uninstal.exe
/opt/watcom/uninstal.exe: PE32 executable for MS Windows 4.00 (GUI), Intel i386, 5 sections``` ???run it
kind mantle
lmfaooo
leaden fox
Why is there an `uninstal.exe` in the linux package
lazy
fiery turtle
Why is there an `uninstal.exe` in the linux package
maybe it relies on wine
fiery ice
LOL WHAT
calm latch
Why not
gentle peak
does this support rdos in-tree??? or is this something else
fiery turtle
probably upstreamed support lol
gentle peak
/opt/watcom/rdos/dwarf.dll: PE32 executable for Unknown subsystem 0xffffad05, Intel i386, 4 sections``` lovelyprobably one of their hyperfixations, wouldn't be surprised if they're autistic themselves
fiery turtle
from what ive seen its likely
fiery turtle
an nt clone using open watcom
fiery ice
lol
fiery turtle
at least use goddamn msvc lol
fiery ice
does it even run windows .exe files?
kind mantle
probably one of their hyperfixations, wouldn't be surprised if they're autistic ...
I don't think so
kind mantle
does it even run windows .exe files?
prolly not
frank canopy
at least use goddamn msvc lol <:trl:829316641191886888>
does msvc handle segmentation properly?
because thats the single point in favor of openwatcom
kind mantle
Isn't segmentation extra painful if you use dynamic libs?
frank canopy
openwatcom can do non-flat segmentation and true real mode compilation thats its only benefit
fiery turtle
people might think open watcom is a pile of trash if they look at that pr lol
frank canopy
yeah lol
vast kestrel
I wonder how the prs he made to ow look like
fiery turtle
poor ow devs 
vast kestrel
He had an issue saying ow doesn't support switch with long long 
jaunty fox
I wonder how the prs he made to ow look like
he had some irritation with watcom using bool
fiery ice
i don't even wanna see his source
vast kestrel
Wait that's from fucking 2016
mortal yoke
i don't even wanna see his source
large parts of it are in asm
or more like 90% of it lol
hollow scaffold
Why would anyone compile using OW?
gentle peak
at least it supports #pragma once
fiery ice
do you hate yourself so much that you have to write bad code?
fiery turtle
at least it supports `#pragma once`
i was surprised he didnt remove that tbh lol
jaunty fox
i don't even wanna see his source
hollow scaffold
svn 😭
fiery turtle
ohh thats why hes so terrible at git
hollow scaffold
oh my
fiery ice
...
kind mantle
what
hollow scaffold
sorry
kind mantle
Does he store segments in eax or smth
calm latch
Wait
fiery ice
he fucking uses pack
gentle peak
right well i can't find a way to detect the pointer size without just checking a bunch of raw target macros (like __386__ and such)
fiery ice
why not use it at least in uACPI
gentle peak
guess that's why the guy hardcoded 4
kind mantle
right well i can't find a way to detect the pointer size without just checking a...
__SIZE_MAX__ or __UINTPTR_MAX__?
fiery turtle
guess that's why the guy hardcoded 4
Fair, it probably only supports i386 or x86_64 so that's ok
gentle peak
`__SIZE_MAX__` or `__UINTPTR_MAX__`?
doesn't have it :)
fiery turtle
U can check either or
kind mantle
doesn't have it :)
owc doesn't have those? bruv
calm latch
doesn't have it :)
Wouldn't C library provide it?
fiery turtle
Not in freestanding
gentle peak
i'm not particularly a fan of using c library headers in uacpi
calm latch
I'm defining those myself in pmOS
gentle peak
esp non-freestanding ones
fiery turtle
^
calm latch
Not in freestanding
Huh
fiery turtle
Well gcc does
calm latch
How is this not freestanding?
mortal yoke
UINTPTR_MAX is in stdint.h
gentle peak
I'm defining those myself in pmOS
you're defining SIZE_MAX/UINTPTR_MAX, the underscore-prefixed/suffixed versions are provided by the compiler
fiery ice
I'm defining those myself in pmOS
same, i just don't like the naming, lol
gentle peak
UINTPTR_MAX is in stdint.h
oh right yeah these ones are
calm latch
Yeah
hollow scaffold
hm
kind mantle
centralized
gentle peak
i guess this'll do
fiery turtle
i guess this'll do
But it supports 64 bit too no?
fiery turtle
Hmm
gentle peak
but it's not like i'm hardcoding anything without checking first, so shouldn't be too much effort to add if i figure out how to get a 64 bit build of it
fiery turtle
True
hollow scaffold
centralized
is there decentralized one, maybe with nfts and ai?
fiery turtle
Lol
fiery ice
is there decentralized one, maybe with nfts and ai?
nft code
kind mantle
Hm yes lettuce add NFTs generated by AI to git
fiery ice
every line of code is paid
gentle peak
types.h only has 65 lines???
hollow scaffold
nft code
every hobby OS kernel crash should be an NFT, they are all (mostly) unique
gentle peak
does this thing use the line number post-macro expansion or something
kind mantle
does this thing use the line number post-macro expansion or something
This reeks of "hehe moi first compoiler"
gentle peak
well it's just a warning so i won't bother
hollow crescent
types.h only has 65 lines???
maybe it's a different types.h
try adding an #error into the one you are looking at
gentle peak
it's a nonsensical warning (W135: Shift amount too large) but still
hollow scaffold
is there uacpi port to linux maybe
gentle peak
maybe it's a different types.h
oh right i was looking at the platform one, not the public one
hollow scaffold
This reeks of "hehe moi first compoiler"
My friend is making their own compiler with custom ABI and seemingly C syntax
gentle peak
they're on this c UACPI_OBJECT_DEBUG_BIT = (1 << UACPI_OBJECT_DEBUG), UACPI_OBJECT_REFERENCE_BIT = (1 << UACPI_OBJECT_REFERENCE), UACPI_OBJECT_BUFFER_INDEX_BIT = (1 << UACPI_OBJECT_BUFFER_INDEX),
kind mantle
Noobs taking on huge projects and making nothing standard is such a phenomenon
kind mantle
Hm yes 16-bit acpi
fiery turtle
Sure
gentle peak
right well this is annoying, openwatcom does support designated initializers, but only as actual initializers, e.g. &(struct uacpi_table) { .index = idx } is invalid, you have to add struct uacpi_table tmp_table = {} at the top of the function and set .index manually later
so it's only really useful if you already have all the values at the top of the function
gentle peak
Error! E1073: Invalid option '-zastd=c23' how fun, the version i'm using doesn't include support for either binary constants or the c23 option
fiery turtle
Maybe clone the github repo?
gentle peak
yeah maybe
gentle peak
although if i'm adding openwatcom support i want it to work on a released version of the compiler and not just trunk
fiery turtle
Its from January I think
fiery turtle
although if i'm adding openwatcom support i want it to work on a released versio...
Ye
gentle peak
Its from January I think
the build i'm using is Version 2.0 beta Mar 1 2025 02:29:14 (64-bit)
mortal yoke
it has a garbage build.sh that takes ages to run though lol
actually they do provide ci builds
gentle peak
oh this is also fun: struct x = {} is invalid syntax on owc! you have to either leave it uninitialized and use memcpy or something, or provide all fields explicitly
or maybe {0} works?
oh thank god
oh and designated initializers only support constant values
mortal yoke
calm latch
constexpr?
fiery turtle
oh and designated initializers only support constant values
Bruh
mortal yoke
now I know to avoid openwatcom at all costs at least (though I didn't even know about it to begin with before this garbage rdos stuff)
fiery turtle
There go some uacpi points
calm latch
You can probably rewrite it without using them 
Is this that difficult to implement in compiler?
mortal yoke
no
fiery turtle
`Error! E1073: Invalid option '-zastd=c23'` how fun, the version i'm using doesn...
Maybe we could get rid of binary constants idk, not that many of them are used
gentle peak
yeah i'm doing that
at least you do seem to be able to declare variables at the top of blocks, not just at the top of functions
which is a relief
gentle peak
seems so yeah
also seems like __VA_ARGS__ is broken? this triggers source/event.c(437): Error! E1030: Cannot take the address of an rvalue c uacpi_trace( "executing GPE(%02X) handler %.4s\n", event->idx, uacpi_namespace_node_name(event->aml_handler).text );
gentle peak
yea the manually expanded version doesn't work so it's not va args thankfully
fixed when replaced by c uacpi_object_name name = uacpi_namespace_node_name(event->aml_handler); uacpi_trace( "executing GPE(%02X) handler %.4s\n", event->idx, name.text );
fiery turtle
Oh well
fiery turtle
seems so yeah
He said it didn't I think(?)
gentle peak
well files where i used it compiled fine so
maybe it only works with -zastd=c99?
oh great the syntax (struct foo){} just doesn't work at all
so in some places i have to replace it with a memset
calm latch
0?
gentle peak
no, it's the (struct foo){} part
aka initializers as random expressions don't work
things like struct foo x = {0} work fine
but things like random.thing.foo = (struct foo){0} don't
fiery turtle
so in some places i have to replace it with a memset
There's a uacpi_memzero I think btw
gentle peak
ah yeah ty
i don't think watcom supports atomics at all
i'll probably have to use inline asm for those but for now i'll just make them assume single threaded no interrupts
fiery turtle
Yeah just do that
gentle peak
oh ffs apparently owc doesn't support offsetof(type, foo.bar)
you have to do offsetof(type, foo) + offsetof(<type of foo>, bar)
fiery turtle
What the hell
gentle peak
and i don't see anything like typeof
fiery turtle
I don't think I use that in uacpi
Since msvc didn't have that as well until recently
gentle peak
resources.c:882
NATIVE_F(gpio_connection, interrupt.triggering)
and a bunch more interrupt. stuff right below that
jaunty fox
there is __typeof seemingly
gentle peak
oh? i couldn't find it with a simple ctrl+f in the user manual
dire owl
Yeah, it's more the segmented memory model stuff.
jaunty fox
i sometimes just search the repo
dire owl
If you want compact, medium, large, or huge, what are your options?
fiery turtle
resources.c:882
That's type of?
gentle peak
`NATIVE_F(gpio_connection, interrupt.triggering)`
ohh wait no this isn't a offsetof foo.bar thing
jaunty fox
they appear to have the nested offsetof too
dire owl
If you want compact, medium, large, or huge, what are your options?
GCC-IA16 and bcc only support tiny and small. GCC-IA16 has experimental support for medium, but large and compact and huge are still an issue.
gentle peak
apparently interrupt is a fucking keyword in owc
or a predefined macro for a keyword or something
fiery turtle
I don't think I use type of at all otherwise it wouldn't compile under msvc
gentle peak
yeah it's not typeof
dire owl
GCC-IA16 and bcc only support tiny and small. GCC-IA16 has experimental support ...
That leaves the old Open Watcom, Microsoft C, and Borland Turbo C compilers.
And Open Watcom is the only one that is actively maintained afaik.
fiery turtle
apparently interrupt is a fucking keyword in owc
What the hell
dire owl
Oh, there's also Digital Mars.
gentle peak
guess i know the reason for this change now
fiery turtle
Lmao
dire owl
apparently interrupt is a fucking keyword in owc
yeah, for interrupt handlers
fiery turtle
Imagine reserving arbitrary words as keywords
dire owl
nah, they didn't give a fuck about the standard
some of those conventions are pre-standard afaik
calm latch
wait, will the watcom support be official?
tired temple
You guys are such a gossip. 😁 have written whole novel about rdos today here.
dire owl
he seems to have the mindset of a ~1992 DOS programmer
calm latch
maybe this server is too advanced 
jaunty fox
he has lofty ambitions and sometimes makes snide comments about uACPI not being real "portable c" but in fact you cannot implement uACPI in "portable c" - it needs must be in "really existing C" because true "portable c" is underpowered
calm latch
You guys are such a gossip. 😁 have written whole novel about rdos today here.
that's the hidden price of using uACPI
dire owl
Probably among other things, he tries to link autism to neanderthals, a species ...
it was actually very recent in terms of geologic time :^)
jaunty fox
and while it's a bit easier now with c11 atomics and such it's truly impossible in c90
dire owl
why does he like autism so much
idk, he did bizarrely make a autism screening questionnaire which was validated in a study
and is seemingly taken seriously by psychiatrists
gentle peak
wait it actually fucking compiled?
dire owl
and is seemingly taken seriously by psychiatrists
although idk if he loves or hates autistic people
his theory is definitely out there tho
gentle peak
i didn't even have to modify the opcode tables the guy was having trouble with
dire owl
also idr if he was neurodivergent himself
gentle peak
well i still have to link it
how does linking work with openwatcom
why is the help bnf
fiery turtle
wait it actually fucking compiled?
Damn that was quick
fiery turtle
why is the help bnf
Lmfao
dire owl
oh, he does seem to have a MSc
fiery turtle
wait it actually fucking compiled?
Any sad changes you had to make?
fiery turtle
he has lofty ambitions and sometimes makes snide comments about uACPI not being ...
Portable C without the interrupt keyword used mind you
gentle peak
Any sad changes you had to make?
none that i haven't mentioned here i think
i'm gonna have dinner then figure out linking
fiery turtle
Linking is gonna be fun
Btw any way to add watcom builds to CI?
So it doesn't bitrot
frank canopy
wait you actually got it to compile? damn
gentle peak
wlink is horrible but i got it to link (or, well, i got it to the point where all the errors are ones i expect: https://hst.sh/luwusozexu.coffeescript)
fiery turtle
Not bad
gentle peak
need to figure out where i'm supposed to get mem* from
oh there's a compiler driver i missed
that's useful
much better than using the compiler directly
still doesn't link anything providing memcpy and friends by default
got it! everything except uacpi kernel api resolves
test_runner uses new c++ features that watcom doesn't support unfortunately
it complains about <filesystem>, <string_view>, <variant>, <span>
and some syntax errors but i hope those are symptoms of those missing headers
fiery turtle
Yeah test runner uses cutting edge c++ unfortunately lol
Just for fun for the most part
He left a new comment complaining that it doesn't build still
While monkuous gets it to build in an hour
gentle peak
Yeah test runner uses cutting edge c++ unfortunately lol
I'm trying to port it anyways because it'd be nice to have as CI
fiery turtle
Yeah definitely
gentle peak
Wait does openwatcom not support fucking enum classes
fiery turtle
Bro
gentle peak
right and it doesn't have attributes either
of course
and no unordered_map either
great!
gentle peak
the in-source docs say since c++11
gentle peak
it also doesn't support auto vars
fiery turtle
💀
gentle peak
Error! E029: col(38) symbol 'nullptr' has not been declared wtf
fiery turtle
not even nullptr lol
slim panther
????? that's a keyword
gentle peak
at least cmake supports owc out of the box
fiery turtle
damn
gentle peak
apparently owc's std::map doesn't have .at(key) or [key]??? cppreference doesn't list any "since ..." for these
fiery turtle
how the fuck are u supposed to use it then lol
gentle peak
Error! E715: col(30) member 'at' has not been declared in 'std::map<std::basic_string<char,std::char_traits<char>,std::allocator<char>>,std::vector<std::basic_string<char,std::char_traits<char>,std::allocator<char>>,std::allocator<std::basic_string<char,std::char_traits<char>,std::allocator<char>>>>,std::less<std::basic_string<char,std::char_traits<char>,std::allocator<char>>>,std::allocator<std::pair<std::basic_string<char,std::char_traits<char>,std:
fiery turtle
note that it has slightly different semantics
.at throws
[] silently makes a new element
gentle peak
Fuck
fiery turtle
lol
frank canopy
lmao
fiery turtle
u mean == .end()
gentle peak
oh right yeah
fiery turtle
yeah looks correct
gentle peak
actually how the fuck am i supposed to replace auto there
jaunty fox
`Error! E029: col(38) symbol 'nullptr' has not been declared` wtf
perhaps you need a parameter to the c++ compiler, they appear to be having nullptr https://github.com/open-watcom/open-watcom-v2/pull/343
gentle peak
clangd doesn't even accept that
gentle peak
perhaps you need a parameter to the c++ compiler, they appear to be having nullp...
uhhh
my c++ args are managed by cmake
the cmakelists.txt has set(CMAKE_CXX_STANDARD 20) so it should have it
gentle peak
or std::map<...>::const_iterator
that worked, thanks
fiery turtle
std::decltype(m_parsed_args)::const_iterator?
without std:: sorry lol
gentle peak
it does need the ugly std::map<..., std::vector<...> > space between the closing brackets
fiery turtle
loool
jaunty fox
perhaps cmake is not doing it right, or i suppose you can convert uacpi to c++03
gentle peak
and yeah decltype isn't supported by owc
fiery turtle
c++11 too
jaunty fox
https://github.com/open-watcom/open-watcom-v2/pull/325 seemingly also there, is cmake broken?
gentle peak
i'll check out the generated makefile ig?
fiery turtle
there's VERBOSE thing u can use
gentle peak
cmake doesn't generate that for watcom wmake apparently
fiery turtle
what the hell is wmake lol
gentle peak
the make included with owc
fiery turtle
💀
gentle peak
and yeah cmake --build . -v makes no difference
ok yeah i manually added -za0x and now nullptr is recognized
decltype doesn't work still
fiery turtle
interesting
gentle peak
Error! E938: col(23) 'auto' is no longer a storage specifier in C++11 mode what do you mean "no longer" it didn't work in the first place
fiery turtle
`Error! E938: col(23) 'auto' is no longer a storage specifier in C++11 mode` wha...
why would that be an error
thats literally how c++ works
fiery turtle
is that different from type?
yeah like static vs auto
auto means automatically cleaned up
its implicit
gentle peak
tests/runner/argparser.h(257): Error! E157: col(38) left expression must be integral
tests/runner/argparser.h(257): Note! N393: col(38) included from tests/runner/test_runner.cpp(8)
tests/runner/argparser.h(257): Note! N717: col(38) left operand type is 'std::ostream (lvalue)'
tests/runner/argparser.h(257): Note! N718: col(38) right operand type is 'std::basic_string<char,std::char_traits<char>,std::allocator<char>> const (lvalue)'```
what the fuckon the stack variables for example
frank canopy
yeah so its disallowing auto as a type because auto is a storage specifier ig
gentle peak
you can't do stream << std::string("whatever");??
gentle peak
oh storage specifier i bet thats an issue
doesn't work either way
fiery turtle
lol
fiery turtle
yeah so its disallowing auto as a type because auto is a storage specifier ig
but in c++11 onwards its not a storage specifier
it means automatically deduce the type
frank canopy
```
tests/runner/argparser.h(257): Error! E157: col(38) left expression must be ...
lmao its using the c# rules for it
well sorta the c# rules for overloading say the right operand must be numeric iirc
not the left
gentle peak
oh for fucks sake does not support foreach
fiery turtle
thats also c++11 probably
gentle peak
doesn't support this init syntax either
calm latch
`Error! E938: col(23) 'auto' is no longer a storage specifier in C++11 mode` wha...
Did C++ have C auto?
frank canopy
lmao
fiery turtle
yeah pre c++11
calm latch
oh for fucks sake does not support foreach
that's just syntactic sugar?
like it can't be hard to implement
gentle peak
that's just syntactic sugar?
yeah its just annoying
calm latch
Did C++ have C auto?
Though C has cool auto now as well
gentle peak
Did C++ have C auto?
what does C auto even do
vast kestrel
Saying "C++" and "can't be hard to implement" is a recipe for disaster
fiery turtle
it just exists as the opposite of static
calm latch
also just auto is int
calm latch
also just auto is int <:trl:829316641191886888>
because everything is int in C
fiery turtle
this was auto
maybe it was const?
vast kestrel
i don't know c++ what am i supposed to do here
Rewrite the runner in C :^)