00:04 <andrewrk> there is no address. no machine code is generated for 0 bit types

05:57 <wink_> My wife is having some outpatient surgery tomorrow, I'll probably sign in at somepoint. See https://github.com/winksaville/zig-u0-tests for the tests I created. There are a few which are yielding "unexpected" results but overall seems pretty good.

05:57 <kristate> wink_: I hope everything goes well

05:58 <kristate> about to write a reply comment, but I really think u0 is in search of a problem where there is none

06:01 <wink_> Could be, but if u0 can be "properly" defined and works "as expected" I think its fine.

06:01 <kristate> wink_: I think it's like that pony language defining 1/0=0

06:02 <kristate> it's going to hide bugs, especially allowing @atomicRmw to be a noop

06:02 <wink_> yeah, I thought that was funny too.

06:02 <kristate> if you need u0, something is seriously wrong

06:03 <scientes> kristate, https://yarchive.net/comp/linux/malloc_0.html

06:03 <wink_> At the moment I have no immediate need but it just seems it could work.

06:03 <kristate> wink_: yeah, that's what makes it dangerous. NULL was the same way apparently

06:04 <kristate> it just _seemed_ like a good idea at the time

06:04 <scientes> NULL was a good idea, zig just does it better

06:04 <wink_> This is pre 1.0 and I think experiments are fine.

06:05 <kristate> wink_: I think that I am of that way, but I did not like andrewrk's attitude that he could not defend u0 but tried to force me to say what was wrong with it.

06:07 <wink_> In the end someone has to make a decision to make progress, mistakes will be made but that's ok. Fail fast is a good thing :)

06:07 <wink_> gtg, getting up early. Cheers.

06:07 <kristate> wink_: okay, be well

06:07 <kristate> thanks again

06:41 <wilsonk> Just stumbled onto this u0 debate (late I realize :) ). If zig is going to be really close to bare metal then there might be one use case for such a type (maybe). Ethernet technically transmits states 0, [ nothing ] and 1...so NOT just 0 and 1. This is because there needs to be a positive or negative voltage difference to produce the 0 or 1...nothing is the neutral state and is a state that needs to be there to avoid bias (i

06:41 <wilsonk> e. too many 0's or 1's in a row produce bias). Perhaps a language that can represent this state would be useful, even if only as a no-op...basically? Or perhaps only as a theoretical type that code-gens to nothing?

06:42 <wilsonk> Anyways, just my two bits. ( pun intended :) )

06:44 <wilsonk> I might also be talking out my butt, because I am trying to recall old CompSci information stuffed deep down in the old medula oblongata!!! Take with a grain of salt please ;)

10:25 <j`ey> how/where does @import happen in the compiler?

10:26 <j`ey> oh, github doesnt show the full ir.cpp file

12:31 <Hejsil> andrewrk, Seems bytesToSlice broke: https://github.com/ziglang/zig/issues/1551

13:32 <andrewrk> Hejsil, I'll fix it, thanks

13:32 <andrewrk> scientes, zig's allocator interface when asked for 0 bytes returns undefined sliced from 0..0

13:33 <andrewrk> so we don't even have to deal with this malloc issue, because we have slices

16:17 <andrewrk> kristate: It is not acceptable to characterize #1530 as "andrewrk could not defend u0". I would like you to take a break from Zig for a week, and then come back when you're ready to interact in an honest, positive, and open-minded way.

22:02 <andrewrk> whew... achieved the bug fix quota to be on schedule for 0.3.0

22:40 <unique_id> nice

22:41 <andrewrk> current schedule: https://clbin.com/fl3zL