03:32 <whitequark> sb0: no. it did not flash an older package.

03:32 <whitequark> observe: StartSun May 1 23:43:24 2016

03:32 <whitequark> Partname b'7k325tffg900\x00'

03:32 <whitequark> Date b'2016/05/01\x00'

03:33 <whitequark> Time b'23:55:18\x00'

03:33 <whitequark> that quite seems like the recent package to me.

05:08 <sb0> whitequark, <sb0> and ***last time*** it definitely flashed an older package.

05:09 <whitequark> which last time? this was in build #657, that failed

05:09 <whitequark> and #656 as well.

05:10 <sb0> friday at ~6pm it flashed the ~2pm package

05:10 <sb0> er, saturday.

05:11 <whitequark> I have no idea what build are you referring to, I don't have timestamps in HKST

05:11 <whitequark> but all builds before #656 did have a bug where the wrong gateware was flashed

05:12 <sb0> that was before your "partially clean conda caches" commit

05:12 <sb0> so ok, this is fixed then

07:30 <sb0> hahaha telegram desktop comes with a 12k lines Qt patch

07:40 <whitequark> completely unsurprising. the vk.com guy has an extreme case of NIH.

07:42 <sb0> they are also not using much of the standard widgets

07:42 <sb0> it looks more like a webpage or smarphone app than a qt app

07:43 <whitequark> see above

07:44 <whitequark> dunno why you bother with telegram though. snake oil is snake oil

07:46 <sb0> it can be used by the average smartphone person, and unlike whatsapp they don't ban me for using linux (they even maintain a GPL linux client and document their protocol)

07:46 <whitequark> yes. and they rolled their own crypto. which means they can't be trusted.

07:48 <sb0> https://telegram.org/blog/cryptocontest

07:48 <sb0> sounds more like NIH than evil intent

07:50 <whitequark> yes. NIH is enough.

07:51 <sb0> I'm not sending anything that cracking is worth anywhere near $300k, so good enough

07:52 <whitequark> I suppose that works

08:11 <whitequark> oh and btw, telegram doesn't use end-to-end encryption unless you opt in

08:11 <whitequark> (which is also buggy and rather trivially broken, the brokenness being a race condition in the protocol that's, surprise, out of scope for their contest)

08:12 <whitequark> hm, no, scratch the last statement, that was a different issue

08:13 <whitequark> still doesn't use end-to-end by default

16:27 <sb0> https://github.com/mheily/libkqueue

16:31 <whitequark> interesting

19:58 <cr1901_modern> sb0: What is the equivalent of the flash_extra.py script in MiSoC after the API change?

23:17 <Neuron1k> Hi guys! Could you tell me what library do you use to make fixed point arithmetic simulation in python? I am looking for something similar to fixed point toolbox in Matlab.