[Ffmpeg-devel-irc] ffmpeg-devel.log.20160115

[burek]

[00:30:21 CET] <cone-777> ffmpeg 03Ganesh Ajjanagadde 07master:2e4fd16f5b90: lavc/aacenc: use isfinite to simplify isnan/isinf logic
[00:34:05 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:f8bc0137bdf8: lavu: prevent overflow in av_clip_intp2_c
[01:00:57 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:763c572801a3: asfdec_o: only set asf_pkt->data_size after sanity checks
[01:00:58 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:aa180169961b: asfdec_o: reject size > INT64_MAX in asf_read_unknown
[01:00:59 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:0002d845e873: asfdec_o: check avio_skip in asf_read_simple_index
[01:01:00 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:74474750f1ac: asfdec_o: prevent overflow causing seekback
[01:01:01 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:3776a72962b0: asfdec_o: make sure packet_size is non-zero before seeking
[01:01:02 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:0e32153e9c29: asfdec_o: break if EOF is reached after asf_read_packet_header
[01:01:03 CET] <cone-777> ffmpeg 03Andreas Cadhalpun 07master:c29e87ad55a2: asfdec_o: check for too small size in asf_read_unknown
[01:01:55 CET] <J_Darnley> Is it michaelni I should ask about getting push access?
[01:24:15 CET] <jamrial> is the new asf demuxer good enough to replace the old one by now?
[01:27:56 CET] <Timothy_Gu> J_Darnley: yes
[01:28:11 CET] <Timothy_Gu> kierank: trac.ffmpeg.org is timing out for some reason. is that expected?
[01:28:29 CET] <kierank> works4me
[01:29:35 CET] <J_Darnley> Timothy_Gu: thanks
[01:29:43 CET] <kierank> can't remember what the sudo password is
[01:31:50 CET] <gajjanag> jamrial: spot anything else, or is patchv3 ready to be posted?
[01:35:18 CET] <kierank> we should sign this https://github.com/dear-github/dear-github
[01:36:27 CET] <jamrial> gajjanag: not as far as i could see, so post it
[01:40:22 CET] <gajjanag> kierank: why not, but do you think if github changes in response some day we will use it more?
[01:40:27 CET] <kierank> no
[01:40:32 CET] <kierank> we want the opposite
[01:40:38 CET] <kierank> pull requests disabled
[01:40:43 CET] <kierank> so that *everyone* uses the ml
[01:40:46 CET] <gajjanag> jamrial: posted
[01:42:06 CET] <gajjanag> kierank: ah, I think it will slowly take care of itself; but it won't hurt to sign
[01:45:50 CET] <gajjanag> kierank: for a hard measure, maybe https://nopullrequests.appspot.com/ is useful
[01:46:02 CET] <kierank> we have https://github.com/FFmpeg/FFmpeg/pull/153
[01:47:13 CET] <gajjanag> kierank: I meant if you want to automate some of the work, instead of comment/manual close of each one
[01:47:26 CET] <J_Darnley> kierank: that has a "it is ${CURRENT_YEAR}" :)
[01:54:31 CET] <kierank> J_Darnley: gonna submit 422 asm for review?
[01:55:15 CET] <J_Darnley> Yes, probably tonight
[01:55:27 CET] <J_Darnley> I was tinkering with it some more
[01:55:45 CET] <J_Darnley> (and fixing the 1 bug I saw)
[02:26:33 CET] <durandal_1707> kierank: I guess you need to implement update_thread_context  
[02:27:30 CET] <kierank> Lol question in one chan and reply in other
[02:29:27 CET] <durandal_1707> I don't want to wake them :P
[02:46:15 CET] <llogan> kierank: i see one FFmpeg dev signed it (but under another project)
[02:46:45 CET] <J_Darnley> Woo! Snow!
[02:47:39 CET] <llogan> Timothy_Gu: what's with PRs like this? https://github.com/FFmpeg/FFmpeg/pull/169
[02:51:33 CET] <Timothy_Gu> llogan: no idea. Somebody who has no idea how GitHub works I guess.
[02:51:49 CET] <llogan> IIRC, there are several like that
[02:52:58 CET] <ac_slater> hey all. When I do `-b:v 500kb`, for example, how does the value make it to an encoder?
[02:53:00 CET] <Timothy_Gu> Yeah seems like so. https://github.com/FFmpeg/FFmpeg/pull/157 is a different case: this person is trying to merge the upstream to their local tree
[02:53:14 CET] <J_Darnley> ac_slater: magnets
[02:53:27 CET] <ac_slater> J_Darnley: haha
[02:53:37 CET] <Timothy_Gu> ac_slater: umm what do you mean?
[02:54:13 CET] <ac_slater> Timothy_Gu: I'm writing an encoder and I want to use the `-b:v` parameter. Is t AVCodecContext->bit_rate? 
[02:54:15 CET] <J_Darnley> Seriously, the value will be pasred and set in the context structure
[02:54:18 CET] <ac_slater> Is it  *
[02:54:22 CET] <Timothy_Gu> Yes.
[02:54:40 CET] <ac_slater> ah ok. "b" != bit_rate. Thanks guys
[02:55:10 CET] <Timothy_Gu> ac_slater: see https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/options_table.h#L45
[02:55:18 CET] <Timothy_Gu> `OFFSET(bit_rate)`
[02:55:29 CET] <ac_slater> I was looking for that! Thanks guys
[02:55:31 CET] <Timothy_Gu> which expands to `offsetof(AVCodecContext,bit_rate)`
[02:55:40 CET] <Timothy_Gu> np
[02:57:47 CET] <llogan> i'll be back in 10 days.
[03:40:41 CET] <cone-777> ffmpeg 03Michael Niedermayer 07master:7ec9c5ce8a75: avcodec/pngenc: Replace memcpy by av_image_copy()
[03:40:42 CET] <cone-777> ffmpeg 03Michael Niedermayer 07master:8f4c3e4b9221: avcodec/pngenc: Fix mixed up linesizes
[03:40:43 CET] <cone-777> ffmpeg 03Mats Peterson 07master:b58cfa616c16: lavf/mov: Confine 0x00000000 to raw/twos fourcc mapping to version 0 sample descriptions
[03:44:52 CET] <ac_slater> is it appropriate to ask about existing encoders here?
[03:48:51 CET] <ac_slater> well, if not, ignore it. So, I have an encoder that only encodes YUV420P. Should I use swscale to convert the input source's pixel format, or is there a better way? Thanks!
[03:49:45 CET] <J_Darnley> If you're writing an encoder for libavcodec then only allow 420as input and ffmpeg will take care of the conversion for you.
[03:50:36 CET] <J_Darnley> Or is it some other encoder you are writing?
[03:50:55 CET] <ac_slater> in my AVCodec struct instance, I set `.pix_fmts = AV_PIX_FMT_YUV420P;`  
[03:50:59 CET] <ac_slater> is this what you maen
[03:51:01 CET] <ac_slater> mean * ? 
[03:51:06 CET] <J_Darnley> Yes
[03:51:13 CET] <ac_slater> J_Darnley: yes, sorry. I'm writing for libavcodec
[03:51:16 CET] <ac_slater> thanks mate
[03:53:45 CET] <J_Darnley> PS I think it is a list so you'll need to terminate it with 0 or whatever value is needed.
[03:54:18 CET] <ac_slater> J_Darnley: I was being brief, it is a list. AV_PIX_FMT_NONE is also in there
[03:54:30 CET] <J_Darnley> ah good
[03:55:46 CET] <J_Darnley> Oh my.  "Digibetes"?  What has the media invented now?
[03:58:16 CET] <ac_slater> J_Darnley: I'm having some newbie issues with the general flow of an avcodec encoder. when my encode() function is called, the proper pixel format conversions already happened?
[03:58:23 CET] <J_Darnley> Yes
[04:00:46 CET] <ac_slater> hmm. Then I guess I don't know how to *properly* copy yuv420p data to my encoder's buffer. Using memcpy(), everything is greenish. Should I be using av_image_copy* ? 
[04:00:58 CET] <J_Darnley> Maybe.
[04:01:08 CET] <ac_slater> yea, of course you wouldnt know ;P
[04:01:11 CET] <J_Darnley> You do know what the "p" means, right?
[04:01:12 CET] <ac_slater> very specific things here
[04:01:18 CET] <ac_slater> packed?
[04:01:22 CET] <J_Darnley> No
[04:01:24 CET] <ac_slater> damn
[04:01:27 CET] <J_Darnley> planar
[04:01:44 CET] <J_Darnley> The samples are split into the first 3 data planes
[04:01:55 CET] <ac_slater> right
[04:02:07 CET] <J_Darnley> The first for luma
[04:02:18 CET] <J_Darnley> and the 2nd and 3rd for chroma
[04:02:27 CET] <J_Darnley> (I don't remeber which order)
[04:02:30 CET] <ac_slater> I know that much at least
[04:02:35 CET] <J_Darnley> Good
[04:02:44 CET] <ac_slater> and the bit depth, etc
[04:02:53 CET] <J_Darnley> Sorry if I'm being too... basic
[04:03:18 CET] <J_Darnley> I would ask, can't yourencoder work directly from the input image?
[04:03:28 CET] Action: J_Darnley curses his spacebar
[04:03:29 CET] <ac_slater> it's necessary sometimes. And yes
[04:04:09 CET] <ac_slater> I don't need to do anything except ensure it's yuv420p (it says "packed planar")
[04:04:32 CET] <J_Darnley> Now that's odd phrasing
[04:04:38 CET] <J_Darnley> Where is that?
[04:05:09 CET] <ac_slater> have you ever heard of OMX ? 
[04:05:15 CET] <ac_slater> openMAX
[04:05:16 CET] <J_Darnley> No
[04:05:21 CET] <J_Darnley> Oh maybe that
[04:05:29 CET] <ac_slater> it's a generic interface for encoders/decoders
[04:05:37 CET] <ac_slater> my hardware encoder happens to use that
[04:06:16 CET] <ac_slater> so, when I configure the OMX layer, my encoder will only accept YUV420PackedPlanar
[04:07:34 CET] <ac_slater> the comment for that enum value is "packed for each payload in planar slices"
[04:10:29 CET] <ac_slater> J_Darnley: ^ 
[04:15:50 CET] <J_Darnley> Sorry I don't know what that means
[04:16:08 CET] <J_Darnley> and a little searching hasn't turned up something I understand
[04:16:18 CET] <J_Darnley> I'm sure others might know
[04:16:32 CET] <ac_slater> J_Darnley: thanks for the help mate
[04:16:48 CET] <J_Darnley> If you want you could send an email to ffmpeg-devel.
[04:16:51 CET] <ac_slater> If building didnt take > 5mins each time, I would more trial and error
[04:16:57 CET] <ac_slater> J_Darnley: I'm thinking about it
[04:17:03 CET] <J_Darnley> Just make sure you mention openmax
[04:17:31 CET] <J_Darnley> When you change 1 file it shouldn't take that long to build
[04:17:46 CET] <J_Darnley> Or is that on some low-power device?
[04:18:00 CET] <ac_slater> yea it's not ffmpeg, it's the other stuff
[04:18:05 CET] <J_Darnley> ah
[04:18:39 CET] <ac_slater> thanks for the advice and help mate
[06:27:28 CET] <cone-777> ffmpeg 03James Almer 07master:e3c188e72c02: avcodec/rmdec: add missing av_log argument
[12:09:24 CET] <durandal_1707> is it ok to add afftfilt filter?
[12:10:56 CET] <ubitux> durandal_1707: what would that be?
[12:11:18 CET] <ubitux> A->V?
[12:11:34 CET] <ubitux> A->"A"?
[12:17:16 CET] <durandal_1707> yes
[12:20:43 CET] <ubitux> "yes" is not a valid answer here :p
[12:41:49 CET] <wm4> lol
[12:41:59 CET] <wm4> what's "A" anyway?
[12:42:13 CET] <durandal_1707> Anarchy
[13:18:15 CET] <nevcairiel> durandal_1707: did that ever visit the ML? i didnt see it
[13:19:12 CET] <durandal_1707> nevcairiel: what are you talking about?
[13:19:32 CET] <nevcairiel> afftfilt
[13:20:50 CET] <durandal_1707> I haven't written it yet
[13:52:46 CET] <atomnuker> what the hell, I get a segfault in "if (!last && class && class->option && class->option[0].name)"
[13:53:36 CET] <atomnuker> this could only happen if the class doesn't exist, yet its there
[13:58:00 CET] <durandal_1707> atomnuker: code?
[13:58:36 CET] <atomnuker> the dirac encoder I posted in the ML
[14:02:56 CET] <durandal_1707> is there software that forbids RE in its EULA and uses ffmpeg? 
[14:17:08 CET] <cone-165> ffmpeg 03Michael Niedermayer 07master:6ba42b6482c7: avformat/hls: More strict url checks
[14:17:39 CET] <nevcairiel> atomnuker: your DiracEncContext lacks an AVClass as the first member, maybe that causes it?
[14:20:08 CET] <Daemon404> that would definitely cause it
[14:26:00 CET] <kierank> durandal_1707: pretty much most software
[14:35:48 CET] <atomnuker> nevcairiel: thanks, works now
[14:36:29 CET] <Daemon404> kierank, which trumps which in the EU though
[14:36:34 CET] <Daemon404> EULA vs RE for compat law
[14:36:40 CET] <kierank> RE
[14:36:44 CET] <Daemon404> figured
[14:36:45 CET] <kierank> compat law iirc
[14:36:52 CET] <nevcairiel> law > some random text some company writes
[14:37:00 CET] <Daemon404> thats debatable in the usa
[14:37:09 CET] <nevcairiel> thats because in the USA the companies make the law
[14:57:59 CET] <kierank> J_Darnley: is there any reason why you can't add code for the 10-bit v210 path as well?
[14:58:04 CET] <kierank> or is it sufficiently different
[15:03:31 CET] <J_Darnley> To be honest: I didn't look at it.
[15:04:02 CET] <J_Darnley> I guess it should be possible
[15:04:35 CET] <J_Darnley> Once I groked the split lanes I just unrolled the loop using the high dqword
[15:07:47 CET] <J_Darnley> Gramner: I need to ask you about the v210 checkasm test.  Are you around?
[15:29:25 CET] <gajjanag> all: some updates on fclose; the real reason that one checks fclose is that ferror does not set errno, and thus gives no info on cause of failure
[15:30:27 CET] <gajjanag> if all one cares about is simply returning a failure code, it will be enough in the read-only case to check ferror
[15:50:19 CET] <cone-165> ffmpeg 03Michael Niedermayer 07master:cfda1bea4c18: avformat/hls: Even stricter URL checks
[16:01:23 CET] <gajjanag> michaelni: do you view this as sufficient justification to check fclose even in the read-only case?
[16:11:35 CET] <michaelni> gajjanag, i need to make a few new relaeses due to that hls security bug, ill leave it to the other devels to review/comment on fclose
[16:14:30 CET] <gajjanag> michaelni: ok, thanks. btw, no hurry at all, and I will be gone for 2 weeks, so feel free to comment later if you wish
[16:23:15 CET] <durandal_1707> new A->V filter: abitscope
[16:23:57 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:2f65366c112d: avformat/mov: Update handbrake_version threshold for full mp3 parsing
[16:23:58 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:0a6324fe9315: swscale/utils: Fix intermediate format for cascaded alpha downscaling
[16:23:59 CET] <cone-165> ffmpeg 03Andreas Cadhalpun 07release/2.8:83c0df90b0a0: mjpegdec: extend check for incompatible values of s->rgb and s->ls
[16:24:00 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:709abc74f5ef: avcodec/put_bits: Always check buffer end before writing
[16:24:01 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:14c63dc9b0d8: swscale/utils: Use normal bilinear scaler if fast cannot be used due to tiny dimensions
[16:24:02 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:aee9aa92161f: avcodec/h264_slice: Fix integer overflow in implicit weight computation
[16:24:03 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:6457cd17be8a: avcodec/motion_est: Fix mv_penalty table size
[16:24:04 CET] <cone-165> ffmpeg 03Thomas Mundt 07release/2.8:3fb97a01ce74: avfilter/vf_scale: set proper out frame color range
[16:24:05 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:36861557a7a6: avcodec/mpegvideo_enc: Clip bits_per_raw_sample within valid range
[16:24:06 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:d79b6dd9f9f7: avformat/ivfenc: fix division by zero
[16:24:07 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:cb4ba7456a79: avformat: Add integer fps from 31 to 60 to get_std_framerate()
[16:24:08 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:df4ee166f1a3: avcodec/mss2: Check for repeat overflow
[16:24:09 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:34e06ce309c9: avcodec/mjpegdec: Fix negative shift
[16:24:10 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:9cad20822370: avcodec/g2meet: Check for ff_els_decode_bit() failure in epic_decode_run_length()
[16:24:11 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:b8382a1faeb2: avcodec/dvdec: Fix "left shift of negative value -254"
[16:24:12 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:e5567f2f6982: avcodec/aacdec_template: Check id_map
[16:24:13 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:f05e0e76d657: avcodec/wavpackenc: Headers are per channel
[16:24:14 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:d7f7586c3fcc: avcodec/wavpackenc: Check the number of channels
[16:24:15 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:d408261b6514: avcodec/mpeg4video: Check time_incr
[16:24:16 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:f459afdd7247: avformat/asfenc: Check pts
[16:24:17 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:90c2256ee52c: avformat/aviobuf: Fix end check in put_str16()
[16:24:18 CET] <cone-165> ffmpeg 03Maxim Andreev 07release/2.8:b7d54d6e0726: avformat/hls: forbid all protocols except http(s) & file
[16:24:19 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:c781a1e14b68: swscale/yuv2rgb: Factor YUVRGB_TABLE_LUMA_HEADROOM out
[16:24:20 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:1fd955606216: swscale/yuv2rgb: Increase YUV2RGB table headroom
[16:24:21 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:cdac68fbab69: swscale/utils: Detect and skip unneeded sws_setColorspaceDetails() calls
[16:24:22 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:be9b89314f32: swscale: Move VScalerContext into vscale.c
[16:24:23 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:07da25548f79: swscale/vscale: Check that 2 tap filters are bilinear before using bilinear code
[16:24:24 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:0c5a71fececb: avcodec/pngenc: Replace memcpy by av_image_copy()
[16:24:25 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:4676e20a9da6: avcodec/pngenc: Fix mixed up linesizes
[16:24:26 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:23b903aaf4ee: avformat/hls: More strict url checks
[16:24:27 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:28f89bc439be: avformat/hls: Even stricter URL checks
[16:28:12 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.8:e303ca3ac94f: Update for 2.8.5
[16:28:13 CET] <cone-165> ffmpeg 03James Almer 07release/2.8:063928910ff8: configure: bump copyright year to 2016
[16:36:57 CET] <gajjanag> durandal_1707: wip, or finished (can't see on ml)?
[16:38:01 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:d41b91ba33a5: swscale/utils: Fix intermediate format for cascaded alpha downscaling
[16:38:02 CET] <cone-165> ffmpeg 03Andreas Cadhalpun 07release/2.7:a7a7c6e44c40: mjpegdec: extend check for incompatible values of s->rgb and s->ls
[16:38:03 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:b65c06ee5efe: avcodec/put_bits: Always check buffer end before writing
[16:38:04 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:d7a13ff23b5e: swscale/utils: Use normal bilinear scaler if fast cannot be used due to tiny dimensions
[16:38:05 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:5b14f436c635: avcodec/h264_slice: Fix integer overflow in implicit weight computation
[16:38:06 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:2c58d734132f: avcodec/motion_est: Fix mv_penalty table size
[16:38:07 CET] <cone-165> ffmpeg 03Thomas Mundt 07release/2.7:2f6d3459d908: avfilter/vf_scale: set proper out frame color range
[16:38:08 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:3dd0d43d8cf7: avcodec/mpegvideo_enc: Clip bits_per_raw_sample within valid range
[16:38:09 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:2316c3586591: avformat: Add integer fps from 31 to 60 to get_std_framerate()
[16:38:10 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:8c08247990ca: avcodec/mss2: Check for repeat overflow
[16:38:11 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:7cfea0d41cb0: avcodec/mjpegdec: Fix negative shift
[16:38:12 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:3820c6a979ca: avcodec/dvdec: Fix "left shift of negative value -254"
[16:38:13 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:d755045e7801: avcodec/aacdec_template: Check id_map
[16:38:14 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:d0fd9bec1f8d: avcodec/wavpackenc: Headers are per channel
[16:38:15 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:858f43043f74: avcodec/wavpackenc: Check the number of channels
[16:38:16 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:0e3c36518d99: avcodec/mpeg4video: Check time_incr
[16:38:17 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:f1cdd93517c8: avformat/asfenc: Check pts
[16:38:18 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:a2b234b917bf: avformat/aviobuf: Fix end check in put_str16()
[16:38:19 CET] <cone-165> ffmpeg 03Maxim Andreev 07release/2.7:cde38373d134: avformat/hls: forbid all protocols except http(s) & file
[16:38:20 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:f5060bce3108: swscale/yuv2rgb: Factor YUVRGB_TABLE_LUMA_HEADROOM out
[16:38:21 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:f0cc6f74fcfe: swscale/yuv2rgb: Increase YUV2RGB table headroom
[16:38:22 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:119659b759a6: swscale/utils: Detect and skip unneeded sws_setColorspaceDetails() calls
[16:38:23 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:123d356829d6: avformat/hls: More strict url checks
[16:38:24 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:e681e92d2c8a: avformat/hls: Even stricter URL checks
[16:38:25 CET] <cone-165> ffmpeg 03James Almer 07release/2.7:d4a7e8979338: configure: bump copyright year to 2016
[16:38:26 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.7:99f3f5a8d595: Update for 2.7.5
[16:39:31 CET] <Gramner> J_Darnley: I'm here now
[16:40:00 CET] <J_Darnley> Ah
[16:40:03 CET] <durandal_1707> gajjanag: just an idea
[16:40:56 CET] <nevcairiel> durandal_1707: whats it do
[16:42:00 CET] <J_Darnley> Gramner: I think I need to change the v210 test to allow testing different widths (xmm vs ymm)
[16:42:26 CET] <J_Darnley> Is there another test you think I should look at copying?
[16:44:09 CET] <durandal_1707> nevcairiel: displays what bits are used in audio
[16:44:17 CET] <Gramner> I don't think there's anything that does that which you can copy
[16:45:05 CET] <J_Darnley> okay
[16:45:23 CET] <J_Darnley> I'll look in more detail later
[16:46:14 CET] <gajjanag> durandal_1707: another filter idea, fir/convolve (http://manpages.ubuntu.com/manpages/natty/man1/sox.1.html), I needed this once for a project 
[16:50:47 CET] <Daemon404> how is it that this hls thing is 'vuln', but a dref box in a mp4/mov file can happily open a relative path?
[16:51:03 CET] <michaelni> atomnuker, aac-pns-encode test fails on 32bit bsd http://fate.ffmpeg.org/report.cgi?time=20160115044841&slot=x86_32-debian-kfreebsd-gcc-4.3
[16:51:48 CET] <michaelni> seems also happening on some linux boxes
[16:52:33 CET] <kierank> Daemon404: nobody reported it yet
[16:52:48 CET] <Daemon404> harr harr
[16:53:14 CET] <JEEB> lol
[16:53:30 CET] <atomnuker> damn, the deviation is 98
[16:53:40 CET] <gajjanag> Daemon404: noob q, what is a dref box
[16:54:15 CET] <JEEB> reference to another sample basically
[16:54:19 CET] <JEEB> (in another file or otherwise)
[16:54:32 CET] <JEEB> it's what the segment linking in matroska got its ideas off of
[16:54:37 CET] <Daemon404> gajjanag, 'box' is a term from the isobmff (mp4) spec
[16:54:43 CET] <Daemon404> it's a tag/structure
[16:54:52 CET] <Daemon404> a dref box can reference external tracks
[16:55:00 CET] <Daemon404> it's common for NLE software.
[16:55:14 CET] <Daemon404> JEEB, matroska is much less powerful
[16:55:22 CET] <Daemon404> i dont think it can do it per-track
[16:55:24 CET] <Daemon404> only full file
[16:55:46 CET] <cone-165> ffmpeg 03Michael Niedermayer 07master:48f5244df7ea: avfilter/vaf_spectrumsynth: Fix mixed declaration and statment
[16:56:12 CET] <RiCON> (and the files need to have the same tracks)
[16:56:26 CET] <nevcairiel> yeah matroska is just time-based slicing
[16:56:37 CET] <Daemon404> RiCON, and even then it is iffy
[16:56:43 CET] <Daemon404> i rememebr problems with flac
[16:56:54 CET] <gajjanag> Daemon404: so the problem is that it can reference an arbitrary external track?
[16:57:15 CET] <Daemon404> we have checks to disallow 1) absolute paths 2) differing origins
[16:57:21 CET] <Daemon404> but we happily allow relative local paths
[16:57:43 CET] <Daemon404> (i think this appleus to http too when opening the master mp4 over http)
[16:58:04 CET] <gajjanag> Daemon404: ok, so in your opinion what is the set of paths we should allow?
[16:58:12 CET] <Daemon404> by default? none.
[16:58:14 CET] <JEEB> Daemon404: yeah, matroska just has you setting segment UUIDs per timeline
[16:58:44 CET] <Daemon404> gajjanag, my point was more that the "security researcher" field can be kind of a cricle jerk.
[16:58:53 CET] <JEEB> RiCON: in theory it's not limited ('cause it's not specified, d'uh). but in practice DShow stuff was quite picky :D
[16:58:57 CET] <Daemon404> like this hls 'vuln'
[16:59:21 CET] <gajjanag> Daemon404: I do agree, I was just trying to understand this a bit better
[16:59:23 CET] <JEEB> <@Daemon404> i rememebr problems with flac <- audio frames not matching on the exact timestamps of the segments
[16:59:44 CET] <JEEB> so audio would leak from "other side" of the link
[17:00:03 CET] <JEEB> <pre-OP><linked OP><loud bang right in the beginning>
[17:00:13 CET] <JEEB> the bang could be heard if the audio frames didn't match
[17:00:19 CET] <Daemon404> JEEB, sounds like current problems when segmenting for dash :P
[17:00:24 CET] <JEEB> :D
[17:00:25 CET] <Daemon404> except browsers will just stop playing
[17:00:29 CET] <Daemon404> depending on the vedor
[17:00:38 CET] <JEEB> DASH is a fucking piece of shit implementation wise
[17:00:48 CET] <Daemon404> only on client side
[17:00:52 CET] <JEEB> yes
[17:00:55 CET] <JEEB> muxing is OK
[17:01:00 CET] <Daemon404> and only because everyone just implements youtubes subset
[17:01:04 CET] <Daemon404> and bugs
[17:01:23 CET] <JEEB> yes, and then people just say "well I didn't try to create files that would break things"
[17:01:26 CET] <JEEB> and I go :V
[17:01:35 CET] <JEEB> as if I'm *trying* to create files that break things
[17:02:27 CET] <Daemon404> i see you arr familiar with the broadcast vendor industry
[17:02:29 CET] <Daemon404> are*
[17:03:53 CET] <JEEB> also I'm an asshole that doesn't accept a multimedia workaround as a "fix", it's not "fixed" I just made something a bit more retarded to make it work
[17:08:15 CET] <gajjanag> JEEB,Daemon404: I have seen this many times, and would like an insider opinion as to why multimedia is so broken
[17:08:48 CET] <Daemon404> it's an incredibly vast thing to say "multimedia is broken"
[17:08:53 CET] <Daemon404> you are going to have to be more specific.
[17:09:12 CET] <Daemon404> (also i dont agree with the sweeping generalization)
[17:18:57 CET] <gajjanag> Daemon404: I guess it is really how one views things, at some level, yes, one can play stuff, but there are many things that don't make sense
[17:18:59 CET] <gajjanag> https://www.youtube.com/watch?v=NoWV2JdoNnc
[17:19:40 CET] <Daemon404> im not going to watch a 44 minute video to discene somethign specific you mean
[17:20:55 CET] <gajjanag> Daemon404: not referring to anything in particular, but see minute 13 or so
[17:21:28 CET] <cone-165> ffmpeg 03Michael Niedermayer 07master:da6f34516b60: avcodec/x86/fmtconvert: Add emms to int32_to_float_fmul_array8_sse()
[17:21:28 CET] <Daemon404> heh
[17:21:45 CET] <Daemon404> the bits at 13 minutes can be appleid to almost any industry
[17:21:50 CET] <Daemon404> it's not unique to multimedia.
[17:26:36 CET] <gajjanag> Daemon404: ok, you see the problem for me is that this my first semi-serious software work, I have been kept insulated from most of the industry ;)
[17:27:29 CET] <gajjanag> Daemon404: maybe 2 years from now I will see similar things in academia, who knows
[17:27:38 CET] <Daemon404> academia has a different set of problems
[17:27:50 CET] <Daemon404> namely being 90% politics, and living in a bubble
[17:28:24 CET] <Daemon404> (and also no journals requrie cod to be checked/published)
[17:30:29 CET] <gajjanag> Daemon404: http://reproducibility.cs.arizona.edu/v2/index.html, something I saw a while back
[17:31:15 CET] <Daemon404> yeah ive seen it
[17:31:46 CET] <Daemon404> i'd also note i tend to see a lot more interesting papers (IM) come out of R&D (ms research etc) thna most universities (with a few exceptions)
[17:31:51 CET] <Daemon404> s/IM/IMO/
[17:32:29 CET] Action: Daemon404 should go see if anything good has been published lately
[17:34:29 CET] <Daemon404> anyone opposed to a patch to disable all dref opening
[17:34:39 CET] <Daemon404> teh default state can be up for debate, of course.
[17:37:19 CET] <gajjanag> Daemon404: can you outline any possibly legit use cases of dref opening
[17:37:51 CET] <Daemon404> yes
[17:37:53 CET] <Daemon404> NLEs use it
[17:37:57 CET] <Daemon404> to avoid remuxing 400gb files
[17:38:04 CET] <Daemon404> same reason NLEs use edit lists
[17:38:11 CET] <Daemon404> (which we dont properly support)
[17:39:24 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:bb70069e6a5c: swscale/utils: Fix intermediate format for cascaded alpha downscaling
[17:39:25 CET] <cone-165> ffmpeg 03Andreas Cadhalpun 07release/2.6:a5fb5a18e3e2: mjpegdec: extend check for incompatible values of s->rgb and s->ls
[17:39:26 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:9c4d8343bb30: avcodec/put_bits: Always check buffer end before writing
[17:39:27 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:dbe194002a79: swscale/utils: Use normal bilinear scaler if fast cannot be used due to tiny dimensions
[17:39:28 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:85732714ce94: avcodec/h264_slice: Fix integer overflow in implicit weight computation
[17:39:29 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:f3cf94fa47d1: avcodec/motion_est: Fix mv_penalty table size
[17:39:30 CET] <cone-165> ffmpeg 03Thomas Mundt 07release/2.6:482ca8241d77: avfilter/vf_scale: set proper out frame color range
[17:39:31 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:81bac54ab3cd: avcodec/mpegvideo_enc: Clip bits_per_raw_sample within valid range
[17:39:32 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:c0315b74e256: avformat: Add integer fps from 31 to 60 to get_std_framerate()
[17:39:33 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:aa5ae8b2136f: avcodec/mss2: Check for repeat overflow
[17:39:34 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:bb1263185b60: avcodec/mjpegdec: Fix negative shift
[17:39:35 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:526b5facc59e: avcodec/dvdec: Fix "left shift of negative value -254"
[17:39:36 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:60956a9dc862: avcodec/wavpackenc: Headers are per channel
[17:39:37 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:6d483543217f: avcodec/wavpackenc: Check the number of channels
[17:39:38 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:9b8d1a183d46: avcodec/mpeg4video: Check time_incr
[17:39:39 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:fbfef00e6883: avformat/asfenc: Check pts
[17:39:40 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:6d5fca34a629: avformat/aviobuf: Fix end check in put_str16()
[17:39:41 CET] <cone-165> ffmpeg 03Maxim Andreev 07release/2.6:8b93f6676b89: avformat/hls: forbid all protocols except http(s) & file
[17:39:42 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:4bafd677953f: swscale/yuv2rgb: Factor YUVRGB_TABLE_LUMA_HEADROOM out
[17:39:43 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:1ba90ce49e27: swscale/yuv2rgb: Increase YUV2RGB table headroom
[17:39:44 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:f0b6705e1a76: swscale/utils: Detect and skip unneeded sws_setColorspaceDetails() calls
[17:39:45 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:232c2ed4a4a5: avformat/hls: More strict url checks
[17:39:46 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:82db8bcffae0: avformat/hls: Even stricter URL checks
[17:39:47 CET] <cone-165> ffmpeg 03James Almer 07release/2.6:80fc1f3399f1: configure: bump copyright year to 2016
[17:39:48 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.6:3fd30f53a99e: Update for 2.6.7
[17:41:13 CET] <gajjanag> Daemon404: ok, so in such cases is there typically any relation typically between the paths that may be used to think of a good default?
[17:41:31 CET] <Daemon404> im not sure what you mean
[17:41:48 CET] <gajjanag> like, e.g same directory, or something like that
[17:42:14 CET] <Daemon404> it's already restriected to relative paths and same origin by default
[17:42:17 CET] <J_Darnley> kierank: FYI I have the 10-bit v210 asm working. I'll get it into a patch after dinner.
[17:42:18 CET] <Daemon404> absolute paths are opt-in
[17:43:28 CET] <wm4> relative paths including .. ?
[17:43:44 CET] <Daemon404> ?
[17:44:38 CET] <wm4> <Daemon404> it's already restriected to relative paths and same origin by default
[17:44:57 CET] <gajjanag> Daemon404: how about something like concat, i.e have a safemode flag
[17:45:00 CET] <wm4> "relative paths" doesn't sound very safe
[17:45:24 CET] <Daemon404> master.mov may open subdir/slave.mov
[17:51:18 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:5d865cb375c4: avutil/mathematics: Fix division by 0
[17:51:19 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:65e7ad746fab: avformat/matroskaenc: Check codecdelay before use
[17:51:20 CET] <cone-165> ffmpeg 03Rainer Hochecker 07release/2.5:50aacb334a34: avformat/utils: estimate_timings_from_pts - increase retry counter, fixes invalid duration for ts files with hevc codec
[17:51:21 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:07b96325311b: avformat/mxfenc: Do not crash if there is no packet in the first stream
[17:51:22 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:75e4949ec128: avfilter/vf_mpdecimate: Add missing emms_c()
[17:51:23 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:a5b4b807757f: avcodec/h264_refs: Fix long_idx check
[17:51:24 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:c1521dd418db: swscale/utils: Fix intermediate format for cascaded alpha downscaling
[17:51:25 CET] <cone-165> ffmpeg 03Andreas Cadhalpun 07release/2.5:f48392197d36: mjpegdec: extend check for incompatible values of s->rgb and s->ls
[17:51:26 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:1363d0759df3: avcodec/put_bits: Always check buffer end before writing
[17:51:28 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:c3abdd0f4461: swscale/utils: Use normal bilinear scaler if fast cannot be used due to tiny dimensions
[17:51:28 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:d472b83abc76: avcodec/h264_slice: Fix integer overflow in implicit weight computation
[17:51:29 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:1b3dffc0c7ff: avcodec/motion_est: Fix mv_penalty table size
[17:51:30 CET] <cone-165> ffmpeg 03Thomas Mundt 07release/2.5:53025e8681c1: avfilter/vf_scale: set proper out frame color range
[17:51:31 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:2111d5161926: avcodec/mpegvideo_enc: Clip bits_per_raw_sample within valid range
[17:51:32 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:7974642cfd60: avformat: Add integer fps from 31 to 60 to get_std_framerate()
[17:51:33 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:4c6b48a08e5a: avcodec/mss2: Check for repeat overflow
[17:51:34 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:e2285e7cc83f: avcodec/mjpegdec: Fix negative shift
[17:51:35 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:1d02458c0672: avcodec/dvdec: Fix "left shift of negative value -254"
[17:51:36 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:71b640230ec8: avcodec/wavpackenc: Headers are per channel
[17:51:37 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:4865d66b76f7: avcodec/wavpackenc: Check the number of channels
[17:51:39 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:1c99807cb704: avcodec/mpeg4video: Check time_incr
[17:51:39 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:8419971a42a8: avformat/asfenc: Check pts
[17:51:40 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:a7dea6f2cdce: avformat/aviobuf: Fix end check in put_str16()
[17:51:41 CET] <cone-165> ffmpeg 03Maxim Andreev 07release/2.5:e183075dc11b: avformat/hls: forbid all protocols except http(s) & file
[17:51:42 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:9c67c52acc94: swscale/yuv2rgb: Factor YUVRGB_TABLE_LUMA_HEADROOM out
[17:51:43 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:dc6a9162a2d2: swscale/yuv2rgb: Increase YUV2RGB table headroom
[17:51:45 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:4089af67ab93: swscale/utils: Detect and skip unneeded sws_setColorspaceDetails() calls
[17:51:45 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:8d5060192cf2: avformat/hls: More strict url checks
[17:51:46 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:83913c6e7e1e: avformat/hls: Even stricter URL checks
[17:51:47 CET] <cone-165> ffmpeg 03James Almer 07release/2.5:6456043ff76e: configure: bump copyright year to 2016
[17:51:48 CET] <cone-165> ffmpeg 03Michael Niedermayer 07release/2.5:fba2fdaea32e: Update for 2.5.10
[17:58:45 CET] <cone-165> ffmpeg 03James Almer 07release/2.8:b8fa3ff95d88: avcodec/wavpackenc: print channel count in av_log call
[18:06:40 CET] <cone-165> ffmpeg 03Michael Niedermayer 07n2.8.5:HEAD: avcodec/x86/fmtconvert: Add emms to int32_to_float_fmul_array8_sse()
[18:11:54 CET] <cone-165> ffmpeg 03James Almer 07master:36778627e249: x86/intmath: add missing early clobber to output operands
[18:23:50 CET] <cone-165> ffmpeg 03James Almer 07release/2.5:7e297b4bc6af: avcodec/wavpackenc: print channel count in av_log call
[18:23:51 CET] <cone-165> ffmpeg 03James Almer 07release/2.6:b901d1bbc9a0: avcodec/wavpackenc: print channel count in av_log call
[18:23:52 CET] <cone-165> ffmpeg 03James Almer 07release/2.7:a697577a641d: avcodec/wavpackenc: print channel count in av_log call
[18:33:11 CET] <J_Darnley> Dammit Telenet.  Are you going to disconnect me again?
[18:35:24 CET] <cone-165> ffmpeg 03James Almer 07n2.7.5:HEAD: x86/intmath: add missing early clobber to output operands
[18:35:25 CET] <cone-165> ffmpeg 03James Almer 07origin/release/2.7:HEAD: x86/intmath: add missing early clobber to output operands
[18:53:32 CET] <jamrial> http://pastebin.com/cYRAseHb
[18:54:15 CET] <jamrial> happens with mingw-w64 x86_32. can someone confirm it's not just me, and/or also test on linux x86_32?
[18:55:34 CET] <nevcairiel> that happened to me randomly/rarely, well not sure if that specifically, but sometimes float dsp test would fail and re-run would fix it
[18:55:55 CET] <nevcairiel> maybe you should checkasm it to make sure its all behaving ;)
[18:57:15 CET] <jamrial> it's that seed value specifically
[18:57:35 CET] <Daemon404> could it be some derpy driver changing teh fp state?
[18:57:45 CET] <Daemon404> oh, ok
[19:02:18 CET] <jamrial> happens only if i use the sse2 or avx optimized versions of vector_dmul_scalar. It goes away if i force cpuflags to 0
[19:02:27 CET] <jamrial> mingw-w64 x86_64 is fine
[19:04:04 CET] <kierank> av_packet_from_data is insane
[19:04:37 CET] <kierank> forces you to use av_malloc and av_free
[19:11:45 CET] <Daemon404> kierank, it's expect though
[19:11:49 CET] <cone-165> ffmpeg 03Paul B Mahol 07master:7d76294ce006: avfilter: add ahistogram multimedia filter
[19:11:54 CET] <Daemon404> it is to be consumed nternally
[19:11:55 CET] <kierank> but what if my memory is from elsewhere
[19:12:05 CET] <kierank> I want to output to hardware or whatever
[19:12:18 CET] <kierank> bit hypothetical but it exists these days with kernel bypass
[19:12:30 CET] <Daemon404> sounds niche?
[19:12:46 CET] <Daemon404> presumably the internal code needs the properties av_malloc can guarantee
[19:12:57 CET] <kierank> if so then ok
[19:13:27 CET] <Daemon404> that said, retty sure you can override av_mallocs implemention in configure
[19:13:31 CET] <Daemon404> pretty*
[19:13:37 CET] <Daemon404> gl;hf though
[19:13:42 CET] <kierank> well the point is to output to my own memory
[19:13:44 CET] <kierank> not to change malloc
[19:14:11 CET] <Daemon404> output what? encoded packets?
[19:15:03 CET] <kierank> yes
[19:15:17 CET] <Daemon404> inside libav*?
[19:15:21 CET] <kierank> a lot of our encoded packets have a higher bitrate than some people's raw frames
[19:15:23 CET] <Daemon404> or to feed to libav*
[19:15:32 CET] <kierank> frame -> dirac -> avpkt
[19:15:56 CET] <Daemon404> im struggling to figure out if you want to call the func internally or from user code
[19:16:01 CET] <Daemon404> and why it needs to be an avpacket at all
[19:16:14 CET] <kierank> I need to encode to dirac so i get an avpacket out
[19:16:27 CET] <kierank> I want it in my own memory though so I use av_packet_from_data
[19:16:55 CET] <Daemon404> does get_buffer2 not work for encoders?
[19:16:58 CET] Action: Daemon404 checks
[19:17:11 CET] <kierank> no ff_alloc_packet(2)
[19:18:00 CET] <Daemon404> hmm yea
[19:18:16 CET] <Daemon404> maybe it can be extended to use such a callback
[19:18:22 CET] <Daemon404> i dont know how nutty it is or not
[19:18:26 CET] <Daemon404> and what SIMD will break
[19:24:43 CET] <wm4> don't packets need to be allocated with padding? not sure if that applies to output
[19:25:46 CET] <cone-165> ffmpeg 03Piotr Bandurski 07master:23ac355487e7: avformat/riff: add YUYV FourCC (Drastic YUYV)
[19:25:47 CET] <cone-165> ffmpeg 03Piotr Bandurski 07master:eded2e4fac98: avformat/riff: add C210 FourCC (Canopus C210)
[19:25:52 CET] <Daemon404> wm4, sure but you have to do that anywya
[19:25:55 CET] <Daemon404> its documented as such
[19:26:02 CET] <Daemon404> av_malloc doesnt automagically do it
[19:29:19 CET] <wm4> yeah
[19:29:51 CET] <wm4> (which I find strange... why doesn't ffmpeg just use standard malloc, and then some sort of aligned-and-padded malloc alternative)
[20:06:17 CET] <jamrial> can somebody confirm if that seed fails for you as well? the issue appears to be in compare_doubles(), where the check apparently can't handle the values generated by lfg with that seed
[20:09:51 CET] <J_Darnley> What system did you want checked?
[20:11:35 CET] <jamrial> it fails for me using mingw-w64 x86_32. it would be nice to check that and also linux x86_32
[20:12:09 CET] <J_Darnley> curses I need to reconfig
[20:13:29 CET] <J_Darnley> FYI it works fine on linux64
[20:17:12 CET] <J_Darnley> odd
[20:17:59 CET] <J_Darnley> Fails for me too
[20:18:48 CET] <J_Darnley> Same error as yours
[20:18:56 CET] <jamrial> what system?
[20:19:55 CET] <J_Darnley> Windows 7 64-bit, i686-w64-mingw32, run through cygwin64
[20:20:05 CET] <jamrial> ok
[20:20:37 CET] <jamrial> does this patch fixes it without breaking every other float_dsp test for you? http://pastebin.com/cDEQREzN
[20:21:17 CET] <J_Darnley> What!?  I don't have d2u installed?
[20:23:40 CET] <J_Darnley> That did fix that test.  How do I run "other tests"?
[20:24:38 CET] <J_Darnley> Other runs without the seed work
[20:25:26 CET] <jamrial> float_dsp-test runs a lot of tests. if you didn't get any new error then they are all fine
[20:25:41 CET] <J_Darnley> I did not get any errors
[20:26:33 CET] <J_Darnley> kierank: I sent 10-bit v210 asm to the list
[20:28:23 CET] <kierank> Can you send to libav if possible
[20:28:48 CET] <J_Darnley> Now, or would you like me to wait for comments?
[20:29:22 CET] <kierank> Now please
[20:33:45 CET] <jamrial> anyway, i'm not sure if that new check is truly correct. for all i know it may fail with other seed values
[20:35:43 CET] <J_Darnley> kierank: done, you may need to tell them to check their moderation log
[20:36:07 CET] <kierank> Thanks
[20:36:31 CET] <J_Darnley> (yes, there are the "awaiting moderation" replies)
[20:45:35 CET] <jamrial> or maybe http://pastebin.com/n8p18db5
[20:46:51 CET] <J_Darnley> That also works
[20:48:19 CET] <jamrial> if someone that actually knows that they are doing wants to try to fix this themselves, or confirm that's a good fix, then that'd be great
[20:48:33 CET] <jamrial> in any case it's not really important, i guess. it's just a test that fails with some specific seed values
[20:51:19 CET] <BtbN> I wonder how long it'll take the ML-Server to notice that my servers IP changed. And then it will unload a bunch of mails in my direction...
[21:19:01 CET] <Compn> do you need me to fix any queued mails ? what ml if so
[21:19:23 CET] <Compn> oh thats libav ml, nevermind then
[21:47:32 CET] <J_Darnley> Compn: yes, sorry about the confusion
[21:51:18 CET] <cone-165> ffmpeg 03Paul B Mahol 07master:d4ce63a1bf25: avfilter/af_sidechaincompress & af_agate: use audio fifo from lavu
[21:52:21 CET] <BtbN> Compn, I just got myself a new server, so all my domains got a new IP. The ffmpeg-devel and ffmpeg-user inboxes are suspiciously quiet now.
[21:56:04 CET] <kylophone> I'm finishing up an audio filter that needs to upsample to 192kHz and then back down to the original sample rate.
[21:56:13 CET] <kylophone> Wondering if I can set that in query_formats()
[21:57:08 CET] <durandal_1707> no, it would upsample but no downsample
[21:57:16 CET] <kylophone> or is it better to handle the upsampling > downsampling internally with libswresample?
[21:57:59 CET] <wm4> why does it need to downsample back to original rate?
[21:58:13 CET] <durandal_1707> depends how you want filter is used
[21:58:50 CET] <durandal_1707> 192000 is for audiophiles? :P
[21:59:51 CET] <kylophone> The broadcast spec for a "true-peak" limiter says to upsample to 192
[22:01:23 CET] <kylophone> I tried libswresample, but ended up with a 16 sample delay in my buffer
[22:01:44 CET] <kylophone> which is annoying to deal with
[22:03:29 CET] <durandal_1707> delay is going to be there anyway
[22:19:18 CET] <klue> What versions does the recent ffmpeg vulnerabiliy affect? (The one that provides SSRF, local file reads, etc)
[22:20:00 CET] <BtbN> ok, the ml found my new IP now
[22:22:15 CET] <gajjanag> jamrial: is v3 of fma ok?
[22:24:52 CET] <Compn> klue : got a link to the cve ?
[22:25:46 CET] <J_Darnley> I think he means that overblown hls+concat issue from yesterday
[22:26:08 CET] <jamrial> gajjanag: the new assembly looks correct, so assuming that difference in the output is ok, then yes
[22:27:11 CET] <gajjanag> jamrial: difference seemed fine, it was just some +/- 0 and nan shifting thing
[22:27:14 CET] <klue> Compn: J_Darnley: I do not know the CVE. And yes, that one.
[22:27:22 CET] <kylophone> I'm assuming this is normal for libswresample: http://i.imgur.com/Yy89i7E.png
[22:27:37 CET] <kylophone> the front of the buffer gets filled with zeros
[22:28:05 CET] <kylophone> and the rest of the buffer is delayed by a certain amount
[22:28:16 CET] <mark4o> CVE-2016-1897 and CVE-2016-1898 http://seclists.org/oss-sec/2016/q1/91
[22:28:46 CET] <kylophone> Second frame starts at 0.1 in that image
[22:28:58 CET] <klue> mark4o: Thanks!
[22:31:53 CET] <Compn> lol
[22:32:01 CET] <Compn> yes, thats how concat works
[22:32:13 CET] <Compn> you specify local file and encode it to wherever, in this case, http server.
[22:32:35 CET] <Compn> its like issuing a cve "i emailed my passwords.txt file to bill gates. its all thunderbirds' fault"
[22:33:13 CET] <wm4> no it's not
[22:33:35 CET] <JEEB> oh right
[22:33:40 CET] <JEEB> playlists to playlists
[22:33:40 CET] <Compn> put ffmpeg in applications group and dont give it permission to /etc/
[22:33:43 CET] <cone-165> ffmpeg 03Paul B Mahol 07n2.6.7:HEAD: avfilter/af_sidechaincompress & af_agate: use audio fifo from lavu
[22:33:57 CET] <JEEB> but still, lol
[22:34:29 CET] <JEEB> sure tho, it's a bug
[22:34:30 CET] <Compn>  i like the other claim too. "rename playlist as avi and ffmpeg still knows its a playlist" ... should this be a bug or feature ?
[22:34:59 CET] <wm4> surely you know ffmpeg probes files by content and not by file extensions
[22:35:10 CET] <Compn> wm4 : http://seclists.org/oss-sec/2016/q1/85
[22:35:15 CET] <Compn> did you read it?
[22:35:24 CET] <J_Darnley> Not on one thread I read.  People were debating that.
[22:36:03 CET] <gajjanag> wm4: so the fundamental reason why fclose needs to be checked is _only_ for diagnostics; else ferror suffices for the read only case
[22:36:20 CET] <wm4> not that again
[22:36:22 CET] <Compn> i'm surprised the guy didnt include the option to convert /etc/passwords to video using the text decoder and then upload to http://...
[22:36:56 CET] <Compn> erm /etc/passwd
[22:37:00 CET] Action: Compn lols
[22:37:55 CET] <mark4o> Compn: hls doesn't have a way to specify filters for the urls in the playlist
[22:37:57 CET] <Compn> (note to anyone reading this irc in archives, i do not speak for ffmpeg security or the project. i'm just a guy)
[22:39:48 CET] <Compn> yeah, probably hls shouldnt be using concat...
[22:39:51 CET] <Compn> ehe
[22:39:56 CET] <Compn> although it is interesting feature.
[22:42:43 CET] <gajjanag> all: what is the policy in situations like this, where wm4 seems to refuse to discuss the issue
[22:47:06 CET] <cone-165> ffmpeg 03Ganesh Ajjanagadde 07master:5989add4ab4e: lavu/x86/lls: add fma3 optimizations for update_lls
[23:11:14 CET] <gajjanag> all: random, minor stuff has been pushed to https://github.com/gajjanag. clear wip stuff has been marked at the tip of my master, rest are things I am ok with 
[00:00:00 CET] --- Sat Jan 16 2016