00:02 <isd> Hm, figured out where the test binary was an ran it directly. Segfault, will investiage further.

00:57 <isd> Unsurprisingly doesn't segfault on -O0. Interestingly, segfaults *sooner* on -O1 than -O2

00:58 <isd> I don't really have the energy for this right now. I'll come back to it later.

03:23 <ocdtrekkie> So Dropbox is replacing FOSS with not-FOSS, it seems: https://hackpad.com/

03:51 <DanC> re refactoring in js: I've used flow to help with that. I suppose typescript is more mature, but its type system bothers me. I guess it's getting better.

03:51 <DanC> meteor bugs me because of all the global state (ambient authority)

04:11 <ill_logic> zarvox: thanks.

06:58 <ill_logic> Is there any hope of having the sandstorm data dir symlink to another place?

06:59 <ill_logic> I have it installed on my system partition and I'm out of room.

06:59 <ill_logic> I symlink it and I get messages about /var/somethingorother not existing

07:01 <ill_logic> I guess this is the container stuff.

07:48 <ill_logic> I guess I'll resize my partitions. It's lvm2.

12:42 <grammar34> Hi folks. Quick question; is there support for bitmaps in capnproto, or do we have to model them still with a struct of Booleans?

13:13 <dwrensha> grammar34: nothing has changed on that front recently

13:14 <dwrensha> grammar34: can you elaborate on your use case?

13:42 <grammar34> I have a bitmap in C/C++ of possible values a = (1 << 0), b = (1 << 1) etc...

13:42 <grammar34> so I have to export it via RPC

13:43 <grammar34> I guess current options are a) struct with booleans b) List of enums, and check manually the uniqueness of the values in the list (sort of an std::set)

14:24 <dwrensha> A List(Bool) might work

14:24 <dwrensha> but struct-with-boolean-fields sounds better

14:25 <grammar34> would that be really compressed; i.e. a struct with say 5 bools would be 1 byte in the wire?

14:25 <dwrensha> grammar34: note that in capnproto-c++, it's possible to call AnyStruct.getDataSection() to recover the raw bitmap

14:26 <grammar34> Well yes, but if the bitmap extends 8 bits you have endianess issues

14:26 <dwrensha> yes, capnproto tighely packs the fields

14:26 <grammar34> So I've already discarded data

14:26 <dwrensha> getDataSection gives you an array of bytes

14:26 <dwrensha> there should be no endianness issues

14:27 <grammar34> shift operations are endian dependent (1 << 12) over a 4 byte type is different in LE and BE

14:29 <dwrensha> the array returned by getDataSection() is in little-endian order

14:30 <grammar34> I know; but I have LE and BE systems :)

14:30 <grammar34> I could craft a library that does the bitmaping thing so that I could transport the bitmap as an array of octets, but I much prefer to use any facility capnp provides

14:31 <grammar34> So I think I will go with a list of enums on one case and a struct with bools on the other

16:08 <simpson> grammar34: Note that Capn's padding means that you have an 8-byte minimum IIUC.

16:09 <simpson> grammar34: If you stick to the Capn accessors, then there shouldn't be endianness issues. I don't know whether that's too slow for you; I don't know what your speed targets are.

17:09 <grammar34> simpson; not sure I get your comment. Point is, directly copying the 4 byte storing the bitmap with data() won't work, because when I read back the value from a different endianesss machine, bitmaps won't match (if bitmaps > 8 bit)

17:09 <grammar34> even if I have enough room with 8 byte, problem is still there

17:09 <grammar34> one nice feature is endianess independence

17:09 <simpson> grammar34: I figured you'd use the list-of-bools approach.

17:10 <simpson> By moving the encoding into Capn, you completely remove the question of whether endianness matters.

17:10 <grammar34> sure, with list of bools work

18:32 <isd> DanC: yeah, I think there are a good number of PLs out there that could be decent re: ocap if you just fixed the stdlib

18:34 <isd> Go's stdlib is terrible in this regard, but it really doesn't need to be.

18:34 <isd> I think there's some work on porting python to cloudabi; I haven't looked but I'm interested to see what they're doing with the libs