#qi-hardware IRC log for Sunday, 2016-02-28

--- Sun Feb 28 201600:00
solrize_poor channel, nobody talks here any more.  everyone lurks waiting for someone else to say something.04:16
kyakoh, finally someone said something!07:04
eintopfparty!08:22
DocScrutinizer05\o/  \o\  /o/  /o\  \o/  /o\  \o/  \o>  <o>  \o/ /o\  <o> _o_08:48
DocScrutinizer05~s°/o\° **CLAPP** °g08:50
Action: kyak celebrates furiously09:04
wpwrakhmm. given (independent) secret keys a and b and C(x,y) being the Curve25519 scalar multiplication, we know that C(a, C(b, 9)) == C(b, C(a, 9)). now, does C(x, y) have all the properties of "normal" multiplication, especially is it cummutative and associative ?16:18
wpwrakif C(x, y) is like "normal" multiplication, then the above is a*(b*9) =[assoc] (a*b)*9 =[commut] (b*a)*9 =[assoc] b*(a*9), where =[foo] means "equals, given <foo>"16:19
wpwrakso this would explain the equality in understandable terms. alas, the definition of C(x, y) is somewhat cryptic. (well, it's cryptography ...)16:21
wpwrakof course, when i try it, C(a, B) != C(b, A) :-(   (where A = C(a, 9), etc.). galling.16:36
Action: wpwrak grumbles about papers using "A" for alice's public key and the reference implementation using "A" for a parameter defining the curve16:53
wpwrakalas, C(a, b) != C(b, a). so much about commutative, or keeping things at the level of high-school math :(16:55
wpwrakbut at least i got the rest right. that's already something16:58
rjeffriesI wish anelok project had gone forward. It was not exactly what I need/want (to SECURELY store passwords etc), but close enough.18:24
rjeffriesI get by by using 1Password on Android. I ONLY use it as a secure store where I enter and then later look up credentials.18:25
wpwrakyeah, anelok is currently in starvation mode. a bit of a vicious circle: no funds -> very limited progress -> long time until marketable product (*) -> no funds18:31
wpwrak(*) not necessarily the final thing, just something that can be used to get funds18:32
rjeffriesunderstood. what sucks is you have created what seems to me (I am no expert) a robust architecture for a flexible hardware based password safe.18:34
wpwrakwell, i've created part of it. there are still a number of things missing.18:38
wpwrakmajor things, like "on-disk" data structures, wireless communication, host software, etc.18:39
wpwrakand of course actual hardware for the latest design :)18:39
rjeffrieshere's something to consider: there might be a niche market for a device that has no (functioning) wireless at all18:43
rjeffriesI guess I need to go look at what is out there. I use an android mobile phone, then also browsers on e.g. Chrome OS and some Windows.18:44
wpwrakdropping wireless would certainly simplify things a lot. but then, a lot of people have gone mobile in the last years, so that would be a pretty severe limit18:55
wpwrakalso, a good password is long and cryptic. so using human-assisted data transfer wouldn't be much fun18:55
wpwrak(of course, anelok does support that. but that should be the fallback / legacy-compatibility choice, not your only option)18:56
wpwrakso i'm afraid "no wireless" is not much of an option. wrong decade :)18:57
wpwrakat least the choice to use primary cells is paying off, with lithium becoming substancia non grata on planes: http://avherald.com/h?article=49460e5c18:59
wpwrak(only as far as cargo is concerned, but still)19:00
rjeffriesI am mainly mobile, and naturally would much prefer a wirless option. and I am sure my passwords are not optimal. but I type them in my hand. grr.19:04
rjeffriesI guess on anelog v. 3 (smile) I'd hope for a somewhat more robust display.19:05
rjeffriesI think primary cells are a smart design tradeoff. I don't remember, do you have ability to monitor how much battery capacity remains? there are some cute chips that do that job.19:06
rjeffriesbut they add cost and board space and consume a small amount of power19:08
wpwrakhmm yes, should connect the battery to some ADC input. not sure how good monitoring would be but can't hurt to have it as an option.19:13
rjeffrieseven a +/- 15 percent would help. People will always be nervous about remaininf capacity.19:17
rjeffriesand you could save power by only measuring, oh, maybe once a month?19:18
wpwrakjust connected it to a comparator / ADC input of the cMCU :) so the cMCU can measure the battery voltage. turning this into something actually meaningful is left as an implementation detail ;-)19:28
wpwrak(accuracy) the problem is that battery chemistry is not very linear. so the battery can recover when not under load, its voltage varies with temperature, and voltage swings (caused by load) increase as the battery discharges19:30
wpwrakso it may be fairly tricky to get a useful battery status indication. but at least we can now measure the voltage.19:39
wpwrakthanks ! might be a useful feature19:39
wpwrakand fixed reverse-charging of the battery as well. one more item off the to do list19:46
wpwrakhmm, tweetnacl doesn't export poly1305 (hash function). that's inconvenient ...20:02
--- Mon Feb 29 201600:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!