| *** tpb has joined #photonsdi | 00:00 | |
| *** futarisIRCcloud has joined #photonsdi | 02:41 | |
| *** futarisIRCcloud has quit IRC | 07:00 | |
| *** se6astian|away is now known as se6astian | 08:00 | |
| *** se6astian is now known as se6astian|away | 08:23 | |
| *** se6astian|away is now known as se6astian | 09:04 | |
| *** Bertl_zZ is now known as Bertl | 09:40 | |
| felix_ | things i learned today so far: you can't buy the inxpensive ticket to the airport the ticket vending machine at the station near the airbnb. but the oebb app is so much better than the wiener linien app; this app just worked as i expected :) | 10:59 |
|---|---|---|
| se6astian | great :D | 11:00 |
| *** Bertl is now known as Bertl_oO | 14:38 | |
| felix_ | what neither the spec nor the recommendation says (they only have a lookup table for that): the protection bits in the xyz-word are actually 4 different parity bits. somehow i get the impression that when the code is finished, it'll probably be a better reference than the spec /o\ | 16:53 |
| kierank | that's how we implement xyz: | 17:05 |
| kierank | [Field][VBI] */ | 17:05 |
| kierank | static const uint16_t sav_fvh_cword[2][2] = {{0x200, 0x2ac}, {0x31c, 0x3b0}}; | 17:05 |
| kierank | static const uint16_t eav_fvh_cword[2][2] = {{0x274, 0x2d8}, {0x368, 0x3c4}}; | 17:05 |
| kierank | the sdi crc isn't too bad, it's the audio BCH polynomial that's annoying | 17:08 |
| felix_ | this is my completely untested implementation https://gist.github.com/felixheld/8073c06b6c5d647b0e04a7b5f7e7f622 (not sure if i used the migen records there properly, but that code isn't finished yet) | 17:08 |
| tpb | Title: gist:8073c06b6c5d647b0e04a7b5f7e7f622 ยท GitHub (at gist.github.com) | 17:08 |
| felix_ | yeah, it's standard crc, but with the crc state bits counted in the wrong direction | 17:08 |
| kierank | 21:30:31 <Kjetil> And then you have secondary forums that will specify how to actually use the standards | 17:09 |
| kierank | I had a big argument with them about this a few months ago | 17:09 |
| felix_ | and the spec didn't say that; only the recommendation from the itu | 17:09 |
| felix_ | heh | 17:10 |
| felix_ | oh and with the protection bits and the information what they actually are, i can do a 1 error correction there :) | 17:12 |
| felix_ | do commercial sdi implementations check if the 10th bit of anc data is the inverse of the 9th bit or di they just discard the 10th bit? | 17:15 |
| *** se6astian is now known as se6astian|away | 17:15 | |
| felix_ | ah, and as far as i've seen the spec also doesn't say why the 10th bit is the inverted 9th bit: it's an easy way to make sure that the 10bit word can never be in the reserved ranges | 17:17 |
| * felix_ is afk for a while; changing trains soon | 17:17 | |
| Kjetil | There is probably also some 8-bit compatability in there | 17:19 |
| Kjetil | or maybe not | 17:21 |
| kierank | in our software implementation we don't check the parity | 17:28 |
| kierank | I've not seen commercial equipment do so | 17:28 |
| kierank | but they do check the crc and bch and xyz | 17:28 |
| kierank | and some check line number | 17:29 |
| kierank | and payload id | 17:29 |
| Kjetil | I guess most ANC implementation will pass the data along. As it will be worse to drop the packet then let a packet with a minor bit error through | 17:30 |
| *** se6astian|away is now known as se6astian | 19:11 | |
| *** se6astian is now known as se6astian|away | 21:57 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!