2021-06-11 UTC
[fluffy], [tantek] and [jacky] joined the channel
# 01:33 [jacky] like granted this is only a problem b/c of how they store information
[schmarty], KartikPrabhu and gRegorLove_ joined the channel
capjamesg, reed, calebjasik, Abhas[m], batkin[m] and nekr0z joined the channel
# 07:40 petermolnar did they mention their blu-ray based cold storage? 'cause that's quite hard to delete.
oodani, hendursa1, KartikPrabhu, [grantcodes], [Murray], IWSlackGateway3, gRegor, [KevinMarks], hendursaga, [jacky], [tw2113_Slack_] and [pfefferle] joined the channel
# 15:29 sknebel or add "shredder" as a location to the blueray-moving bot :D
alex_, [tantek], BinyaminGreen[m], gRegor, [pfefferle]1, alex__, aaronpk_, michael, mikeputnam1 and HelpMeUnderstand joined the channel
[snarfed], rrix and [KevinMarks] joined the channel
reed joined the channel
KartikPrabhu and gRegor joined the channel
# 17:58 GWG Out of curiosity, what do people think a good default expiry is for an IndieAuth token?
# 18:01 GWG gRegor: I was going back and forth...
# 18:02 GWG The amount will be changeable in my plan on the authorize screen, so I suppose that 14 as a default is fine
# 18:03 GWG How does OAuth handle it? Can a client tell you something that indicates what token type it should give?
# 18:07 aaronpk no you wouldn't want the client to have any control over that. usually that's set by policies on the authorization server, which might depend on the client type or the user or the scopes issued
[fluffy] joined the channel
# 18:11 GWG aaronpk: I know you wouldn't want the client...but how do I know what the client type is?
# 18:11 gRegor My server defaults to 14 days and lets you check a box to make it not expire. There's also a UI to expire individual tokens or all at once.
# 18:11 aaronpk the only indication you have in indieauth is the client_id URL and what scopes it's requesting
# 18:12 GWG I'm more thinking of knowing whether a client is unattended or attended. So Quill vs OwnYourSwarm
# 18:12 aaronpk yeah that would be nice. we don't have anything in the flow that would indicate that
# 18:13 aaronpk you see this in some oauth servers where they request the scope "offline_access"
# 18:13 GWG aaronpk: Maybe that's a scope proposal
[manton] joined the channel
# 18:17 GWG In the meantime, will have to start somewhere
# 18:30 gRegor What are you thinking you would do different for offline access clients?
# 18:32 sknebel they cant as easily prompt the user for a new token
# 18:32 GWG gRegor: Might change the text at the least when authorizing.
# 18:32 GWG Those are the use case for refresh
# 18:38 aaronpk another example would be letting those apps get long lived tokens but limtiing the token lifetime on apps like quill that only work when the user is actively using them anyway
alex11 and barnaby joined the channel
# 18:50 barnaby I’m posting a video to my site rather than yt for the first time in forever, trying to figure out what settings will get me the best results in the least space
# 18:50 barnaby got a bunch of transcoding jobs queued up in handbrake
# 18:51 barnaby so far, nothing has been smaller than the original 1080p mov file which DaVinci Resolve gave me
# 18:52 barnaby the 1080p webm ended up being 190MB vs 133MB for the .mov
# 18:53 barnaby which is especially weird as I’m pretty sure the webm is 30fps and the (much smaller) original was 50fps
# 18:59 barnaby [snarfed]: do you know which video formats successfully get POSSEd to twitter via bridgy publish?
# 19:05 barnaby seems like webm codec via handbrake is consistently much larger than .h264/mp4
# 19:11 barnaby so, for my 01:44 long video, the smallest I managed to get it with handbrake was as a 720p H.264 mp4 file, at 85MB
# 19:12 barnaby webm was ~1.5x larger than H.264 at both 1080p and 720p
# 19:12 aaronpk does handbrake use ffmpeg under the hood or is it its own thing?
# 19:13 barnaby likely ffmpeg, as the UI definitely looks like a thin layer over a command line tool :P
# 19:14 barnaby do you know of a particularly good ffmpeg option set for encoding web videos?
# 19:18 aaronpk I don't remember, it's been a while since I touched any of that. I think actually most of the time i've posted videos on my site it's either from instagram or just whatever my phone records not transcoded
# 19:23 barnaby so much for the webm hype :/ unless I’m missing something obvious
# 19:26 barnaby those were with the VP8 encoder, I’ll try one with VP9 and see if it makes any difference
# 19:40 [snarfed] Bridgy Publish currently only supports mp4 videos w/Twitter right now. I think they required that at some point, but maybe they’ve relaxed it since?
# 19:41 barnaby I just tried posting one, seems to be taking a while, but it took 3:30 to upload to my site so I imagine it will take some time to upload to twitter
# 19:46 aaronpk i wonder what's a good threshold for deciding to publish HLS instead of a single video file
# 19:46 barnaby yikes, webm VP9 is 151MB vs VP8 123MB, at the same settings
# 19:46 barnaby aaronpk: whatever it is, I’m guessing my 1:45 long video of wild hamster clips isn’t quite there yet
# 19:47 aaronpk cause otherwise someone visits that page on a phone and it downloads the whole thing
# 19:47 sknebel browsers should be clever enough to not load all of it if not needed
# 19:47 aaronpk one trick i've been meaning to look into more is providing an HLS playlist that points to byte offsets in a single video file
# 19:48 sknebel fairly sure even on desktop they only load the start and then fetch more if it is played
# 19:48 barnaby so at least on ff I have to explicitly start playing it, and it takes a few seconds before it has buffered enough to start
# 19:48 barnaby so it’s definitely not loading anything before it’s asked to
# 19:48 [snarfed] barnaby I’m not seeing any recent bridgy publish requests for that page. did you do it interactively?
# 19:49 barnaby [snarfed]: nope, should have been triggered via webmention. I’ll try re-saving the post and see if that works
# 19:49 barnaby more likely my webmention endpoint tried something stupid like loading an 85MB file :/
# 19:50 barnaby oh yeah looks like it’s an issue on my side. saving just fails for some reason. I’ll look into it after dinner
# 19:57 petermolnar be memory limits - eg php-fpm -, but the python ones could go completely rogue.
# 19:57 petermolnar > more likely my webmention endpoint tried something stupid like loading an 85MB file :/ - I'm still extremely tempted to respond with a certain gzip to queries like obvious wordpress sniffing on my static site, but the malicious person in me wonders how many webmention receivers could deal with a gzip bomb without taking down the server behind them. Depending on the implementation, there should
# 19:57 petermolnar it's a good thing that unlike in xml and yaml, there is no option to make a recursive json.
[tantek] joined the channel
# 20:00 GWG petermolnar: I actually do a head on mine to check type and I limit size
# 20:01 barnaby GWG: yeah pretty sure that’s what I’ll be implementing after I’m done cooking :D
# 20:02 barnaby I might also just filter out URLs ending with common media extensions e.g. png mp3 mp4 etc from webmention processing altogether
# 20:04 barnaby I originally thought it might be interesting to send them so that people could get notified if I hotlink their content, but I don’t think anyone does that
# 20:07 barnaby I just downloaded a video from twitter, it’s a 720p H.264 mp4
# 20:08 barnaby 5.6MB for 26 seconds, that’s 0.22MB per second. My video is 100 seconds long, so it shouldn’t be much more than about 22MB
# 20:09 Loqi curl has -1 karma in this channel over the last year (2 in all channels)
[fluffy] joined the channel
# 20:09 Loqi curl has 0 karma in this channel over the last year (3 in all channels)
# 20:11 sknebel barnaby: in handbrake you can set a target bitrate - set that to match what you want and see if the result looks good?
# 20:12 GWG petermolnar: That's helpful. But it would have returned an error message due size.. just didn't surface it
# 20:22 GWG I know how to fix that. I'll make a note
# 20:22 GWG But I put a lot of safety measures into that code
# 20:23 GWG Including refusing to download media mime types
[jacky], [schmarty], gRegor, hendursa1, [Ana_Rodrigues] and [tw2113_Slack_] joined the channel
# 21:09 barnaby okay, the bit rate definitely explains the video size difference! my video is 6Mb/s, the twitter sample is more like 1, which explains the vast size difference
# 21:11 barnaby so now to queue up a bunch of renders at different bit rates and decide where to make the size/quality tradeoff
# 21:11 barnaby then to document it all on /video so I don’t have to think about it in the future
# 21:11 Loqi document has 1 karma over the last year
# 21:14 barnaby maybe I can even upload a handbrake preset file somewhere
# 21:15 aaronpk 6mb/s is a good target for high quality 1080 video
hendursaga joined the channel
# 21:20 barnaby two minutes of 720p 3Mb/s is a much more reasonable 41MB, starting to see a noticeable quality dip now
# 21:46 [tantek] is that like high resolution hamsterdance or something?
# 21:47 [tantek] oops thought this was #indieweb-chat for some reason 🐹
# 21:48 barnaby is happy to spread his wild hamster obsession everywhere
# 21:48 Loqi tantek has 18 karma in this channel over the last year (62 in all channels)
# 21:54 barnaby anyone know how the “Constant quality: RF [x]” slider works in handbrake?
# 21:54 barnaby what seem like small changes on it can result in vast file size and quality differences
# 21:58 sknebel it uses some video quality estimation thing to evaluate how individual pieces of the video look after compression, and adjusts the bitrate for the pieces until they all hit a specific quality value
# 22:01 sknebel not the clearest text, at least for my sleepy brain, but some details that match what I remembered
# 22:01 barnaby at least for this video, it looks like RF 20-35 is the range where the quality is still acceptable, but the file size starts dropping dramatically
# 22:05 Loqi sknebel has 7 karma in this channel over the last year (32 in all channels)
[aciccarello] joined the channel
# 22:16 barnaby looks like the (now 25MB) video file is still too big for twitter?
# 22:17 barnaby I edited it specifically to be twitter-compatible ;)
# 22:21 sknebel right, but can it do the polling if finalize doesn't succeed synchronously?
# 22:23 sknebel that'd be my guess given the error message and the response examples in that doc link
# 22:23 [snarfed] heh, lots of this info is news to me since i implemented twitter video upload many years ago
# 22:25 [snarfed] i really shouldn’t have to know this much about twitter’s internal video processing
# 22:37 [snarfed] OK barnaby feel free to try again. I give it a 50/50 chance this time 😎
# 22:55 [snarfed] yup, video upload seems ok, but when bridgy then tried to tweet with it, twitter replied “Not valid video”
# 22:55 [snarfed] gotta run but I can look later. apologies for the trouble!
# 22:56 barnaby I’ll resist the temptation to post it manually, so we can test it again later :)
# 22:58 barnaby hmm or maybe I’ll try posting it manually, to see if the twitter web UI complains about the video file
# 22:58 barnaby that way we know if it’s just twitter not liking my particular hamster video file, or a bridgy issue
# 23:00 barnaby huh well the twitter web video upload UI is somewhat lacking
# 23:00 barnaby I posted it, and the form just greys out, with no progress or activity indicator
# 23:02 barnaby okay, twitter web upload worked eventually, so it looks like it’s not an issue with that particular file
# 23:07 [tantek] at some point I gave up on posting video because it got so involved
# 23:09 barnaby yeah me too. not something I can motivate myself to do very often, but in the case of the hamsters, it made sense. It’s not something I’d put on youtube, but is something I want on my site
wagle joined the channel