samth
2017-1-31 10:18:12

Can you explain why we only get 30 bits?



pnwamk
2017-1-31 13:16:58

in our discussions we always threw around “31 bits” and “63 bits” but we were forgetting about the sign bit


pnwamk
2017-1-31 13:17:38

I realized this and fixed it for our base-type bitfields, but I forgot about type masks until just now :frowning:


samth
2017-1-31 14:37:30

you should still be able to get 31 bits out of a fixnum, though


samth
2017-1-31 14:37:44

but it would probably be inconvenient


pnwamk
2017-1-31 14:53:04

if we figure out a sensible way to squeeze out that extra bit then cool


pnwamk
2017-1-31 14:53:17

but at the moment we’re not desperate for it, and the bitwise logic is nice and simple


pnwamk
2017-1-31 14:53:58

this bug is in the general 6.8 release =(


pnwamk
2017-1-31 14:54:12

it surprises me that it went this long undiscovered


pnwamk
2017-1-31 14:54:20

I thought the type mask changes were a while ago


samth
2017-1-31 14:54:38

no one uses 32 bit systems?


pnwamk
2017-1-31 14:55:30

:smiley:


pnwamk
2017-1-31 14:55:40

are non of the automatic builds on 32-bit systems?


pnwamk
2017-1-31 14:55:49

(or in a 32 bit vm?)


samth
2017-1-31 14:56:11

I don’t know


samth
2017-1-31 14:56:23

but we will make sure to fix that before 6.9


asumu
2017-1-31 17:30:18

@pnwamk I think some of the pre-release builds are 32-bit, like this build log http://pre-release.racket-lang.org/log/%7B1%7D%20Racket%20%7C%20%7B3%7D%20Linux%20%7C%20%7B1%7D%20x86%20(32-bit);%20built%20on%20Debian%207%20(Wheezy) but somehow didn’t trigger the issue.


samth
2017-1-31 17:30:44

@pnwamk said that it was non-deterministic


samth
2017-1-31 17:30:50

so maybe that’s why we didn’t see it


samth
2017-1-31 17:31:02

but I don’t know why it wouldn’t have happened occasionally


pnwamk
2017-1-31 17:31:12

Yah, I had to wiggle it a few times to make it happen


samth
2017-1-31 17:31:49

but the various nightly and pre-release builds happen a lot


samth
2017-1-31 17:31:55

so I’m surprised


pnwamk
2017-1-31 17:33:05

perhaps the issue was introduced earlier on, but the more recent changes caused it to happen more frequently


pnwamk
2017-1-31 17:33:12

I dunno


jerryj
2017-1-31 19:00:13

Is there any syntax in typed/racket that is analogous to java generics? I’m toying around with a java source generator and I’m trying to keep things consistent with what’s already available


samth
2017-1-31 19:02:16

@jerryj yes, see the All type constructor


samth
2017-1-31 19:02:46

jerryj
2017-1-31 19:03:15

@samth thank you!


lexi.lambda
2017-2-1 03:43:36

I’m suddenly getting a whole bunch of “CFNetwork SSLHandshake failed” errors when trying to install packages on macOS from http://pkgs.racket-lang.org\|pkgs.racket-lang.org. Can anyone reproduce that, or is it just me?


georges-duperon
2017-2-1 05:02:54

@georges-duperon has joined the channel