rifkik has quit [Remote host closed the connection]
clktmr has quit [Ping timeout: 250 seconds]
rjtobin has joined #zig
benjamin-l has quit [Ping timeout: 250 seconds]
knebulae has quit [Read error: Connection reset by peer]
knebulae has joined #zig
dimenus has quit [Remote host closed the connection]
n_1-c_k has quit [Read error: Connection reset by peer]
n_1-c_k has joined #zig
Ichorio_ has joined #zig
Ichorio has quit [Ping timeout: 264 seconds]
benjamin-l has joined #zig
benjamin-l has quit [Remote host closed the connection]
benjamin-l has joined #zig
doublex_ has joined #zig
marler8997_ has joined #zig
<marler8997_>
Hey, does zig document somewhere that the windows binary require VCRUNTIME140_1.dll?
doublex has quit [Ping timeout: 250 seconds]
doublex_ has quit [Ping timeout: 250 seconds]
waleee-cl has quit [Quit: Connection closed for inactivity]
doublex has joined #zig
benjamin-l has quit [Ping timeout: 250 seconds]
benjamin-l has joined #zig
benjamin-l has quit [Remote host closed the connection]
benjamin-l has joined #zig
rjtobin has quit [Ping timeout: 250 seconds]
benjamin-l has quit [Ping timeout: 250 seconds]
Ichorio_ has quit [Ping timeout: 264 seconds]
benjamin-l has joined #zig
<tgschultz>
marler8997_ see #3391
<marler8997_>
thanks
chemist69 has quit [Ping timeout: 276 seconds]
chemist69 has joined #zig
mahmudov has quit [Ping timeout: 250 seconds]
benjamin-l has quit [Ping timeout: 245 seconds]
benjamin-l has joined #zig
benjamin-l has quit [Ping timeout: 276 seconds]
pixelherodev has joined #zig
n_1-c_k has quit [Read error: Connection reset by peer]
n_1-c_k has joined #zig
ltriant has quit [Quit: leaving]
FireFox317 has joined #zig
mikdusan1 has joined #zig
mikdusan has quit [Ping timeout: 265 seconds]
FireFox317 has quit [Ping timeout: 268 seconds]
FireFox317 has joined #zig
<FireFox317>
andrewrk: I solved the struct field names, but I have one more question about the function parameter names. Because to my understanding there is no AstNode available in the ZigTypeFn or in any of its members. How would you solved it then?
<FireFox317>
I can read the answer when I'm not online through irclog :)
FireFox317 has quit [Remote host closed the connection]
<daurnimator>
bgiannan: oooo, I saw this being discussed the other day
<daurnimator>
bgiannan: have a look at the extended attributes on the file with `xattr -l` I think it was
<daurnimator>
bgiannan: also IIRC someone said you only get that if you download with a browser; more specifically: safari. you can probably download with curl and have it work
<daurnimator>
(or just delete the extended attributes?)
<daurnimator>
bgiannan: otherwise you're the first person to actually hit the supposed upcoming "must be signed" problem that we thought wasn't hitting until next year
<bgiannan>
daurnimator, the required code-signing comes with catalina
<mikdusan1>
bgiannan: yes that's the quarantine sillyness. manually downloading from commandline via curl or wget or whatever circumvents the browser tagging downloads with quarntine flag
<bgiannan>
btw downloaded with chrome not safari
<mikdusan1>
chrome probably uses the same system libs to fetch urls
<mikdusan1>
and double-clicking the .xz file then uses a .app that is quarantine-aware and puts that bit on each extracted file
<bgiannan>
huh they removed Availability.h now i can't import sdl anymore
<mikdusan1>
did you upgrade xcode by chance?
<bgiannan>
yeah i think i'm missing a `xcode-select --install`
<mikdusan1>
first try nuking zig-cache dir
<mikdusan1>
it has a path setting to xcode in there.
<bgiannan>
ok
<bgiannan>
did both now it works
<bgiannan>
thx mikdusan1
<mikdusan1>
yw. this exact thing bit me a week ago
<bgiannan>
funny i had none of those issue on catalina beta
<mikdusan1>
it's pretty clear macOS is moving towards sanboxing model just like iOS, iPadOS, etc.
<bgiannan>
yes, quite sad
LargeEpsilon has joined #zig
<mikdusan1>
my only real gripe: where is 24-bit (true color) xterm color support for Terminal.app :(
<bgiannan>
you use the stock terminal app?
<mikdusan1>
yes i really like it
<bgiannan>
what do you like about it vs iTerm ?
<mikdusan1>
i don't use it
mikdusan1 is now known as mikdusan
n_1-c_k has quit [Read error: Connection reset by peer]
n_1-c_k has joined #zig
sepisoad has quit [Quit: Connection closed for inactivity]
_whitelogger has joined #zig
LargeEpsilon has quit [Ping timeout: 246 seconds]
rifkik has joined #zig
nrdmn has quit [Read error: Connection reset by peer]
LargeEpsilon has joined #zig
<rifkik>
FINALLY
<rifkik>
I FINISHED THE 65816 Emulator
<rifkik>
Now, onto PPU
<mq32>
neat!
rifkik has quit [Remote host closed the connection]
doublex has quit [Ping timeout: 245 seconds]
doublex has joined #zig
FireFox317 has joined #zig
<FireFox317>
andrewrk: #3420 is ready for review. I'm pretty sure that this is the solution you talked about yesterday
<FireFox317>
Maybe the naming of some variables can be a bit better.
doublex has quit [Ping timeout: 245 seconds]
kllr_sbstn has joined #zig
FireFox317 has quit [Remote host closed the connection]
doublex has joined #zig
<fengb>
rifkik: if SNES is anything like the Gameboy, the PPU is the hardest part to get right :P
<daurnimator>
fengb: I think emulating the snes was plagued with timing issues
<daurnimator>
like; there's 4 clock domains or something; and some games take advantage of slightly out of sync clocks
<fengb>
Oof really?
<fengb>
That sounds awful
n_1-c_k has quit [Read error: Connection reset by peer]
<daurnimator>
there used to be really good articles about it on byuu.org but the author deleted them.
<daurnimator>
you can probably find them in the wayback machine....
n_1-c_k has joined #zig
<fengb>
"But run a game like Air Strike Patrol, which writes to the display brightness register multiple times per scanline, and you must synchronize after every clock cycle if you want full accuracy."
<fengb>
Hardware limitations are quite amazing at pushing devs to do really crazy things
<pixelherodev>
True enough
dingenskirchen has quit [Remote host closed the connection]
dingenskirchen has joined #zig
<n_1-c_k>
'@cImport(@cInclude("sqlite3.h"))' fails with 'file not found'. '@cImport(@cInclude("/usr/include/sqlite3.h"))' succeeds, as does '#include <sqlite3.h>' in C. Is expected that the full path is required in zig?
<mq32>
n_1-c_k, no, but you have to add system include paths in the build.zig file
<daurnimator>
n_1-c_k: you need to pass the include dirs
<mq32>
at least that's what i needed
<mq32>
<fengb> Hardware limitations are quite amazing at pushing devs to do really crazy things
<daurnimator>
n_1-c_k: (to zig build*)
<mq32>
oh yeah, and it's really fun to do that!
<daurnimator>
n_1-c_k: if you use .addSystemLibrary or whatever it is from your build.zig, it should set up include paths for you
<n_1-c_k>
mq32, daurnimator, thanks.
<andrewrk>
mikdusan, I wonder how it would affect memory and perf if we started doing string interning
<andrewrk>
I could add some instrumentation to measure how it would affect memory
<daurnimator>
I haven't had a look at the new docs yet. Are they online anywhere?
<daurnimator>
andrewrk: I guess target will be selectable?
<andrewrk>
daurnimator, nope, I have a plan to merge multiple dumps into one, so the target will be a multibuild
<daurnimator>
k
<andrewrk>
having a selectable target is an ok hack to solve a difficult problem, but we can do better
<daurnimator>
I think this is already showing us lots of std fixes we need to make :P
<andrewrk>
yep!
<andrewrk>
this is really contributor friendly too, it's just 1 page of js rendering based on a json dump of semantic analysis
<daurnimator>
hmmm the http and valgrind structs seem to be empty right now?
<andrewrk>
daurnimator, yes there's still lots to do. in this case the thing to do is to display the text: "this struct is not referenced or tested"
<daurnimator>
also leaving the parameter name out seems to be a not-great default choice. e.g. https://ziglang.org/documentation/master/std/#std;crypto.gimli.Hash.final the description mentions writing the hash to `out`, but that's not available
<andrewrk>
I'm right about to merge FireFox's PR to fix that. I think it would be better to hold off feedback until a couple more days
<daurnimator>
k, no problem :)
<andrewrk>
unless you want to get involved in developing it :)
<daurnimator>
I'm still traveling for a bit and am rather limited in capabilities
<daurnimator>
I should be back to normal in 2 weeks
<andrewrk>
fair enough. try this use case: use the keyboard shortcuts to search for std.io.readFileAlloc and have a look at the error set description
<companion_cube>
this stdlib docs is already better than anything C has…
<presiden>
for a longer indepth local doc, info page + emacs is a good combo
<mq32>
presiden: on windows? :D
<mq32>
companion_cube: my problem was that i didn't understand that i could just do "map.insert("Foo", "bar");" and be happy with that
<presiden>
maybe, if emacs can detect where the infopages installed?
<companion_cube>
ah well, that's not the stdlib :D
<companion_cube>
mq32: right, HashMap has some cool magic
<mq32>
<companion_cube> ah well, that's not the stdlib :D
<companion_cube>
I mean, it's nice to be able to do `map.insert("a", "b")` and not have to do `map.isnert("a".to_string(), "b".to_string())`
<mq32>
yeah i know, but it's the example i can remember
<companion_cube>
:)
<companion_cube>
From/Into are pretty classic
<mq32>
yeah, if you know it, which i didn't
<mq32>
and Rust is kinda hard to grasp with all the implicit stuff going on
<andrewrk>
that might be a limitation of the complexity of the language rather than the organization of the docs
<presiden>
ah, and also for the old days, Delphi and it's docs is quite good too
<companion_cube>
yeah, although it's not exactly implicits
<companion_cube>
it makes `?` much more usable
<companion_cube>
(and also, indeed, conversion between borrowed and owned)
<presiden>
I love languages that provides local copy of it's documentation
<mq32>
presiden: please stop! i have to work with those on a daily basis and the docs are just good for "i want to hack something together", but not for "i want to know who owns the memory" ^^
<mq32>
there is TBytes, which is a "byte array thingy"
<mq32>
it has no documentation at all if bytes are guaranteed sequential in memory, so i am copying them manually one-by-one instead of using memcpy or similar
<presiden>
mq32, I think that's more of the quality of the docs, and unrelated to whether the docs available locally or not
<mq32>
yeah, that's true. i was commenting on "Delphi and it's docs is quite good too"
<mq32>
but i also love local readable docs
<mq32>
best case is readable in a terminal which doesn't fit zigs docs right now, but could be done by statically generating HTML
<companion_cube>
in a terminal you can also generate manpages
<mq32>
companion_cube, yeah, but i think the info format would be better suitable, as it is hyperlinked
<companion_cube>
but it's mostly an emacs thing -_-
<mq32>
so gnu info is actually an emacs module?
<companion_cube>
ah, right. I just never could figure out the key bindings -_-
<mq32>
now i'm a bit confused :D
<companion_cube>
man uses my pager (less) which has vim like keybindings
<companion_cube>
info is… weird :p
LargeEpsilon has quit [Ping timeout: 246 seconds]
<mq32>
yeah, info is weird, but actually quite logical
<mq32>
also, you can always press "h" for help :D
<companion_cube>
thanks, will try it again I guess :)
<companion_cube>
although most of the stuff I care about is available as manpages anyway
<mq32>
yeah, but gcc for example has the full documentation available
<companion_cube>
gnu stuff, heh
<mq32>
all flags, all attribute, ld has full linker script documentation and everything :)
waleee-cl has joined #zig
kllr_sbstn has quit [Ping timeout: 250 seconds]
<mq32>
my small game project is having some good progress in the last days. if i find time in the hackerspace today, i'll probably have some basic gameplay implemented later :)
<daurnimator>
presiden: indeed. we'll need a docs => man page thing too
wootehfoot has joined #zig
kllr_sbstn has joined #zig
nrdmn has joined #zig
kllr_sbstn has quit [Ping timeout: 265 seconds]
gustav_o has joined #zig
<gustav_o>
hi everyone
<gustav_o>
how do I go about debugging the following: broken LLVM module found: Call parameter type does not match function signature! [LLVM IR] ... Unable to dump stack trace: debug info stripped'
n_1-c_k has quit [Read error: Connection reset by peer]
n_1-c_k has joined #zig
<andrewrk>
hi gustav_o. the debug process for that looks like using cli flags to enable dumping LLVM IR of the module, finding out what is wrong, and then trying to produce a minimal zig test case. there is probably already an issue open for this which has already done this work
halbeno has joined #zig
clktmr has joined #zig
<gustav_o>
andrewrk ah ok, so reduce example then --verbose-llvm-ir and try to find out what's wrong? I see that there are a few issues with the same llvm error, I'll investigate!
<gustav_o>
keep up the good work by the way :)
<gustav_o>
I'm starting a small game engine in zig, going to be fun
<gustav_o>
So one strange thing I encountered when trying to debug this was that if I comment out a while loop in the middle of a function (the meat of the function), I get a type error ("use of undeclared identifier 'Self'") because I forgot to declare it. The strange thing is that zig seems to skip that error when the while expression is there and proceed
<gustav_o>
to the code generation step (which later causes the llvm error). Shouldn't type checking be deterministic and always pass before code gen, or am I missing something with how zig has structured its compiler?
<gustav_o>
Or is this related to comptime maybe?
<andrewrk>
gustav_o, you might be experiencing how zig does not even semantically analyze branches known to be dead at comptime
<gustav_o>
Ah yes, that was the reason.. so zig saw that the while never terminated so the code below did not get analyzed then. So in that case there should be an "unreachable code" error?
<gustav_o>
(I hope I'm not notifying "so" every time I write the word, the webchat does it automatically)
<andrewrk>
gustav_o, hmm yes there appears to be a regression in the "unreachable code" error for a loop that never terminates
<gustav_o>
Is there an issue for it that you know of?
<gustav_o>
oh it's newer.. the wording in it sounded like there was no concept of unreachable code
FireFox317 has joined #zig
<FireFox317>
andrewrk: Thanks for merging my PR, the docs are really coming together!
<andrewrk>
FireFox317, yeah! thanks for your contribution
<andrewrk>
I have an idea to do multi-target docs, gonna try to do that today
<andrewrk>
this is why we have done the ast nodes thing, because this is how you "join" multiple analysis dumps together - first you merge primitive types & ast nodes together, then you can start trying to merge the rest of the types together
<FireFox317>
Ah yeah indeed, by just refering to the already existing astNodes!
<andrewrk>
you end up with some things being additive, and some things having a *set* of possible values rather than a single one. the docs must represent this, because this is reality! on different targets this decl can be different values
<FireFox317>
And these docs can be generated by running them on the different CI's and with the help of the qemu thing right?
<gustav_o>
looking forward to the std docs, is there any eta?
<gustav_o>
hmm.. my problem relates to me trying to write to a slice that is passed as argument to a function as a pointer, ie. "parseIdent(self: Recognizer, allocator: *Allocator, result: *[]const u8)", and I feel I'm doing something wrong here. How do I pass an output-argument to a function that allocates memory and assigns it?
<andrewrk>
gustav_o, eta is today or tomorrow
<gustav_o>
wow :)
<n_1-c_k>
After much effort I achieved the triumph of printing "SQLITE_ABORT is 4." using this program, <https://paste.debian.net/plain/1105756>. But how do I declare a "database" variable of the C type 'sqlite3 *'? I've only managed a variety of "invalid token"s.
<daurnimator>
n_1-c_k: though I'd expect there is some `open` function.... so you'd do something like `const db = c.sqlite_open(someargs)`
<Sahnvour>
andrewrk: could you remind me where's the code finding libc ? the sys_include_path has a typo it seems
<n_1-c_k>
daurnimator, thanks, I now have a different error, which is progress.
<n_1-c_k>
daurnimator, ah it was a type, it works now, thanks!
<n_1-c_k>
s/type/typo/
<Sahnvour>
nevermind, found it
<andrewrk>
MSVC has really brought a lot of problems to us. this new dll dependency, code generation errors, making the include directory layout more complicated
<andrewrk>
this is a good reason to self host
FireFox317 has quit [Remote host closed the connection]
<tgschultz>
is there a reason we need to be on 2019?
<andrewrk>
no, I think it was a mistake to upgrade so quickly
<andrewrk>
but as mikdusan points out, it's unclear that downgrading will solve the new dll dependency issue
<andrewrk>
however it would solve the code generation bug
<Sahnvour>
andrewrk: note that the solution for forcing static CRT linkage on old cmake versions is to use the override file I talked about the other day
<gustav_o>
andrewrk: Ok I found the bug, it seems that zig doesn't enforce handling error sets when calling functions and at the same time deref:ing a pointer: https://godbolt.org/z/13mff7
<gustav_o>
not sure if it's only pointers to slices though, next thing to try..
<gustav_o>
(hm, the second case is not a pointer but a slice though)
kllr_sbstn has joined #zig
mahmudov has joined #zig
Ichorio has joined #zig
kllr_sbstn has quit [Ping timeout: 268 seconds]
DutchGh0st has joined #zig
<DutchGh0st>
Hi. I wonder how I *propperly* call `zig test <file>` within an azure pipeline CI?
<DutchGh0st>
atm I just download the zig compiler, and test within the build process
<daurnimator>
DutchGh0st: 'properly'?
<daurnimator>
DutchGh0st: you can have a test target in your build.zig
<DutchGh0st>
well yeah sure but then the pipeline doesnt have any tests
<daurnimator>
?
<DutchGh0st>
it then runs 0 tests
<andrewrk>
DutchGh0st, maybe you can look at zig's own build.zig for inspiration
<DutchGh0st>
going off now, but Im gonna read back the chat, so Ill see any suggestions :)
<andrewrk>
why bother downloading old versions of zig?
<DutchGh0st>
(want to enforce older compiler support)
<andrewrk>
I would recommend against that
DutchGh0st has quit [Ping timeout: 250 seconds]
<n_1-c_k>
Thanks to earlier help, I've progressed from a compile time to a link time error, <https://paste.debian.net/plain/1105764>. How do I persuade it to find fcntl64?
<andrewrk>
n_1-c_k, fcntl64 should be provided by libc, which in your case is glibc
<andrewrk>
what system are you on?
<n_1-c_k>
andrewrk, debian stable.
<n_1-c_k>
The C version does compile and run, so presumably does find it.
<andrewrk>
n_1-c_k, I think I might have figured it out, are you using a static zig from ziglang.org/download?
<andrewrk>
It looks like fcntl64 was added in GLIBC_2.28. When you build for the native target, zig tries to figure out which glibc version is native. If your zig is dynamically linked, it looks at its own libc dependency to find which glibc version is native. Otherwise there's a TODO to do something more sophisticated, and thenh it falls back to the default of GLIBC_2.27
<andrewrk>
so the proper fix to this is to improve native glibc version detection
<andrewrk>
I have a plan for this, which is to essentially do ldd on /usr/bin/env, and find which glibc it is
<gustav_o>
I save all bugs I encounter and will probably return to them to see if I can fix them, once I've gotten a bit further with my engine
<gustav_o>
*zig bugs
FireFox317 has joined #zig
samtebbs has joined #zig
FireFox317 has quit [Remote host closed the connection]
ltriant has joined #zig
doublex has quit [Ping timeout: 250 seconds]
mps has quit [Ping timeout: 265 seconds]
mps has joined #zig
<mikdusan>
andrewrk: re: string interning; for the results of ir_resolve_str?
kllr_sbstn has joined #zig
<andrewrk>
I mean in general, across the entire codebase for most strings
<andrewrk>
Buf is about 2x overhead
<scientes>
then fix buf?
<mikdusan>
ah. so on that note i was wondering if you could shed some light on `ir_eval_const_value`,
<mikdusan>
it creates an ir_executable, does potentially a ton of work,
doublex has joined #zig
<mikdusan>
but is there kind of a wrap here suitable for arena? arena.create(); eval(); take/copy value to main heap; arena.destroy()
<andrewrk>
yes I think there is an opportunity for this, however there are plenty of things allocated that must not be destroyed during this time
<andrewrk>
for example anything to do with ConstExprValue
kllr_sbstn has quit [Ping timeout: 240 seconds]
<andrewrk>
I think it would be good to make IrInstruction have a pointer to ConstExprValue rather than embedded, this way IrInstructions can be deleted
<samtebbs>
Is there a neat Zig way of getting the maximum value of an integer type?
<andrewrk>
however even these are referenced sometimes. it will require careful code review to see which things can be dropped from memory and which must remain
gustav_o has quit [Remote host closed the connection]
wootehfoot has quit [Read error: Connection reset by peer]
porky11 has quit [Quit: Leaving]
n_1-c_k has quit [Read error: Connection reset by peer]
n_1-c_k has joined #zig
<fengb>
Oh wow
<fengb>
I think my favorite part is builtins :P
<waleee-cl>
do the ? operator work? I use a vim-like extension to ff that isn't always working perfectly in "let all keys through"-mode
<waleee-cl>
or rather perhaps, does it do anything yet? :)
<andrewrk>
it works. i welcome suggestions to make it friendlier to extensions which mess with keyboard shortcuts
benjamin-l has joined #zig
benjamin-l has quit [Ping timeout: 268 seconds]
<andrewrk>
I'll do a stream in a bit to demo the doc generation and show how to contribute
rifkik has joined #zig
<mq32>
andrewrk, whyyyy? i wanted to get sleep this night! *laughing in zig addicted*
benjamin-l has joined #zig
<waleee-cl>
andrewrk: I'd guess that the fault lies more with tridactyl
benjamin-l has quit [Ping timeout: 240 seconds]
clktmr has quit [Ping timeout: 276 seconds]
<waleee-cl>
hm, but it doesn't seem to work with all extensions disabled either
<waleee-cl>
ah, nvm, firefox didn't start a new instance with --safe-mode for some reason and loaded the extensions anyway
Protty has joined #zig
<waleee-cl>
well, doesn't seem to work for with a extension-less pristine firefox 69.0.2 instance
<andrewrk>
waleee-cl, use the browser's debugger to figure out why. main.js, line 1197
rifkik has quit [Remote host closed the connection]
<waleee-cl>
heh, it seems to not match any of the cases
<andrewrk>
what value is it producing?
<waleee-cl>
ShiftRight, seems like this might be a keyboard layout issue?
Protty has quit [Remote host closed the connection]
squeek502 has joined #zig
kllr_sbstn has joined #zig
return0e has quit [Remote host closed the connection]
<waleee-cl>
(late night brain, after a setxkbmap us), of course that wasn't the fault, but the onWindowKeyDown-function is triggered with only the shift-key