[Ffmpeg-devel-irc] ffmpeg-devel.log.20170218

[burek]

[00:06:54 CET] <kierank> J_Darnley: you should commit that imo
[00:14:19 CET] <llogan> i guess i should pay more attention to these things
[02:13:35 CET] <cone-241> ffmpeg 03Rostislav Pehlivanov 07master:b71f62a9c849: opusenc: initialize the emphasis coefficients on init
[02:13:56 CET] <atomnuker> RiCON: does it still crash for you with that sample?
[02:36:46 CET] <RiCON> atomnuker: yeah
[02:41:50 CET] <atomnuker> the backtrace should be cleaner now, could you give me a new one?
[02:43:24 CET] <cone-241> ffmpeg 03Steven Liu 07master:965f35b111a0: avformat/hlsenc: set default http method to PUT when method is null
[02:49:42 CET] <RiCON> atomnuker: https://i.fsbn.eu/5fzM.txt
[03:05:49 CET] <kierank> michaelni: mpeg2 mixed frame field is broken
[03:05:58 CET] <kierank> coming up with a testcase soon
[03:23:34 CET] <atomnuker> RiCON: can you apply this diff https://0x0.st/4tv.diff and see if it fixes the crash?
[03:28:23 CET] <RiCON> thinking it's because of windows' L32LL64?
[03:31:13 CET] <atomnuker> actually not sure, the Y vector shouldn't have those values ever since the sum should be K
[03:31:36 CET] <atomnuker> and K isn't more than like 64 and N isn't more than like 32
[03:33:17 CET] <atomnuker> I think I better fuzz the PVQ search
[03:36:41 CET] <RiCON> backtrace (still crashes): https://i.fsbn.eu/cB1F.txt
[04:16:06 CET] <atomnuker> RiCON: this should fix it: https://0x0.st/4tU.diff
[04:19:56 CET] <cone-241> ffmpeg 03Michael Niedermayer 07master:6baee21eb72d: avformat/m4vdec: Detect raw mpeg4video with unidentified non mpeg headers with a very low score
[04:27:09 CET] <RiCON> atomnuker: going sleep now, but over the other diff or just that one?
[04:42:21 CET] <atomnuker> RiCON: just this one
[12:18:33 CET] <BtbN> hm, why does my curl dislike the certificate on patchwork.ffmpeg.org?
[13:20:43 CET] <JEEB> wbs: finally got to testing your negative CTS offset patch, and in my limited testing it seems to work nicely enough. ( https://github.com/jeeb/ffmpeg/commits/isml_improvements )
[13:21:35 CET] <JEEB> the timestamp "sanity check" sure gets hit a lot by it so I just disabled it for negative CTS offsets :P
[13:26:43 CET] <cone-194> ffmpeg 03Miroslav SlugeH 07master:4cb8872eb79a: avcodec/cuvid: don't overwrite deinterlace at progressive input
[13:26:44 CET] <cone-194> ffmpeg 03Miroslav SlugeH 07master:2a2f6b28873e: avcodec/cuvid: add drop_second_field as input option
[13:40:03 CET] <nevcairiel> BtbN: its a startcom certificate, which has fallen from grace recently, so maybe your local cert store already purged that
[13:52:48 CET] <BtbN> seems like it, yeah
[13:53:18 CET] <JEEB> yeah, startcom certs should be replaced with a correctly configured Let's Encrypt as startcom really is kind of ded now
[13:53:19 CET] <BtbN> For some reason Chrome got "improved", so there is no way for it to show me any information about the certificate, not even the issuer or expiry date -_-
[13:53:26 CET] <JEEB> really? :o
[13:53:33 CET] <JEEB> that's... really bad design
[13:53:35 CET] <BtbN> Yep, the Cert Info dialog is completely gone
[13:53:42 CET] <BtbN> to enhance the user experience, to not confuse the user
[13:53:57 CET] <wm4> lol
[13:53:58 CET] <JEEB> right
[13:54:15 CET] <wm4> google software...
[13:55:31 CET] <BtbN> "Chrome version 56 seems to have removed this Details link from the popup. You can still press F12 or Ctrl+Shift+I to bring up the Developer Tools and go to Security tab instead. Then click the View certificate button on the Developer Tools panel that opens up, either at the bottom of the website or on the right side."
[13:57:10 CET] <BtbN> It's also aparently using sha1
[14:22:14 CET] <RiCON> atomnuker: https://i.fsbn.eu/b54r.txt
[14:35:54 CET] <nevcairiel> as you quoted, the link just moved, it confused me at first as well, but w hat do i care where i find the info
[16:25:54 CET] <KGB> [13FFV1] 15dericed opened pull request #52: move Michael's copyright statement from spec to pdf backmatter (06master...06move-copyright) 02https://git.io/vD7wy
[16:30:35 CET] <KGB> [13FFV1] 15dericed opened pull request #53: add self as co-author (06master...06add-self-as-co-author) 02https://git.io/vD7rW
[16:33:31 CET] <J_Darnley> I have finally made the change jamrial requested (twice).
[16:33:39 CET] <J_Darnley> Should I send the patches for review again or can I just push them?
[16:52:25 CET] <kierank> Just push
[17:05:35 CET] <durandal_1707> push push
[17:11:18 CET] <J_Darnley> :q
[17:27:23 CET] <cone-583> ffmpeg 03Michael Niedermayer 07master:56803218db9a: tests/fate/hevc: remove vsync drop from where it is not needed anymore
[18:44:05 CET] <jamrial> J_Darnley: just push, assuming it works and fate passes
[20:24:31 CET] <kierank> michaelni: just so I understand correctly, mpeg-2 "mbaff" and "paff" mixed should work right?
[20:24:43 CET] <kierank> so I can give the decoder frames or fields?
[20:24:50 CET] <kierank> or both
[20:28:06 CET] <cone-583> ffmpeg 03James Darnley 07master:9d815b7424b5: avcodec/x86: deduplicate PASS8ROWS macro
[20:28:07 CET] <cone-583> ffmpeg 03James Darnley 07master:e18bc2114f3d: avcodec/h264: add named parameters to x86 function
[20:28:08 CET] <cone-583> ffmpeg 03James Darnley 07master:7627df15d411: x86util: import MOVHL macro
[20:28:09 CET] <cone-583> ffmpeg 03James Darnley 07master:533688786799: avcodec/h264: sse2, avx h luma mbaff deblock/loop filter
[22:12:06 CET] <michaelni> kierank, yes but the mpeg2 doesnt allow some things like frame,field,frame, but switching between frames and fields should be fine
[22:12:21 CET] <kierank> ok I have a stream which fails
[22:12:25 CET] <kierank> when I separate the pictures myself
[22:12:40 CET] <kierank> I just need to write a testase
[23:16:16 CET] <jamrial> J_Darnley: think you could write a checkasm test for h264 loop filter functions?
[23:16:40 CET] <J_Darnley> I can try.
[23:16:53 CET] <J_Darnley> Are any h264 functions already present?
[23:35:06 CET] <jamrial> J_Darnley: yes, idct and a couple others
[00:00:00 CET] --- Sun Feb 19 2017