1 2014-08-27 00:00:18 <sipa> hey
 2 2014-08-27 00:00:19 <Luke-Jr> too bad GitHub won't show you the edit history
 3 2014-08-27 00:00:30 <sipa> if travis likes headersfirst, can we just merge? :p
 4 2014-08-27 00:01:33 <gmaxwell> I'm okay with merging headersfirst but we cannot release without a system test on it.
 5 2014-08-27 00:02:09 <gmaxwell> Merging would be useful because we'll stop wasting your time rebasing, and we'll get more testing on it with it in master.
 6 2014-08-27 00:02:18 <sipa> i'm actually not okay with merging, as it means we're unable to do any tests on future pulls
 7 2014-08-27 00:02:59 <gmaxwell> oh crap, I didn't consider that. /me duncecap
 8 2014-08-27 00:03:21 <sipa> now, travis *SHOULD* fail on it
 9 2014-08-27 00:05:45 <gmaxwell> sipa: Can secp256k1 be changed to a coding style where {} are required instead of single statement if's only being seperated by indentation?  I don't like to style-wonk but unbraced ifs are a known source of bugs. (I personally prefer a style where unbraced is only allowed only if its all one line; but always brace is probably slightly less failure prone)
10 2014-08-27 00:06:07 <sipa> gmaxwell: ack
11 2014-08-27 00:08:58 <sipa> gmaxwell: i should also clang-format it
12 2014-08-27 00:10:23 <gmaxwell> sipa: also, it might be helpful to prefix the file names. For example, lto in MSVC has a fit if there are object files anywhere in the project with duplicate names. so generic source files like 'util' might cause problems for some parties embeding the librrary.
13 2014-08-27 00:10:49 <sipa> pfff :)
14 2014-08-27 00:11:04 <sipa> i find it annoying enough that all functions are prefixed
15 2014-08-27 00:11:25 <gmaxwell> The files are less annoying than functions. :P
16 2014-08-27 00:11:46 <sipa> but but! i use an 8.3 filename length fs!
17 2014-08-27 00:11:53 <sipa> actually, nvm
18 2014-08-27 00:12:04 <gmaxwell> sipa: also, can we take all the places with var-arrays and make them constants? (and add asserts to check that the constants are large enough?)
19 2014-08-27 00:12:29 <sipa> there should only be a few of those, but sure
20 2014-08-27 00:12:56 <sipa> or require an array of sufficient size to be passed in as argument, when we'd normally allocate one on the stack
21 2014-08-27 00:13:16 <gmaxwell> There are only a few.  Well, in particular, there is kind of a weird one in that the batch inversion used during setup is huge, and is never used with values anywhere near as big elsewhere.
22 2014-08-27 00:13:43 <gmaxwell> (I noticed this while trying to determine the amount of stack required)
23 2014-08-27 00:14:02 <sipa> which may be an argument in favor of passing in array
24 2014-08-27 00:14:09 <sipa> as that allows that huge array to be on the heap
25 2014-08-27 00:14:19 <sipa> avoiding the need for a large stack
26 2014-08-27 00:14:38 <gmaxwell> I believe that array turning setup is the peak stack user currently.
27 2014-08-27 00:15:37 <gmaxwell> (also getting rid of the var-arrays is needed both for C89 and C++ compatiblity, also: functions that use var arrays are slightly slower due to some additional management code that must be emitted)
28 2014-08-27 00:16:55 <sipa> go fix it :)
29 2014-08-27 00:17:45 <gmaxwell> yea yea, wanted to clear it with you before starting on it.
30 2014-08-27 00:18:37 <sipa> damn, at my parent's place in belgium
31 2014-08-27 00:18:51 <sipa> how do these people survive with a 1.5 Mbit connection (apparently)?
32 2014-08-27 00:19:17 <sipa> trying to download the blockchain takes several seconds per block, and kills pretty much all other surfing activity
33 2014-08-27 00:21:11 <gmaxwell> sipa: if its killing other stuff sounds like bufferbloat.
34 2014-08-27 02:28:17 <extor> Why does this say it's under active development but references mtgox as an exchange rate source and has no updates for a year? https://www.drupal.org/project/uc_bitcoin
35 2014-08-27 02:32:05 <justanotheruser> extor: probably because it isn't active and you have to be active to change the page to saying they're inactive
36 2014-08-27 03:26:53 <BlueMatt> gmaxwell: so its now a rule that we will only be required to be able to reorg 288 blocks?