00:33:01dxrt_ is now known as dxrt
00:33:02dxrt quits [Changing host]
00:33:02dxrt (dxrt) joins
00:52:41dxrt quits [Client Quit]
00:52:59dxrt joins
00:53:01dxrt quits [Changing host]
00:53:01dxrt (dxrt) joins
01:10:43prebala-publici joins
01:32:03themadpro joins
03:54:41<tech234a>I'm almost ready to upload the YouTube Community Contributions data to IA. It has 1,361,998 files across 406,394 folders, and is 9.46 GB in size (11.7 GB size on disk). I assume putting that number of files and folders uncompressed into one item isn't advisable. How should I package the data for upload?
03:59:39<OrIdow6^2>9.46 is plenty small enough for a single item
03:59:45<OrIdow6^2>Just put it in a tar or something
04:00:02<OrIdow6^2>*9.46 GB
04:00:29<tech234a>Alright, I was more worried about the folder count
04:00:34<OrIdow6^2>If you want people to be able to access the files individually, use an uncompressed tar, or something like zip that has an index at the beginning
04:01:17<@JAA>Is there some reasonable way to split it up into smaller groups? One tar or zip with that many files and directories is not going to work very well I think.
04:01:36<@JAA>I mean, it'll be fine for upload and download, but the viewer on IA won't like it most likely.
04:01:55<tech234a>It looks like I can divide it by first two characters of a the ID and it will be ~400 folders per grouping
04:02:49<tech234a>Probably it would be one item with ~4032 zip files in it then
04:03:14<@JAA>Hmm, that on the other hand sounds like a bit too many files in one item again.
04:03:15<OrIdow6^2>Oh, you suppose the viewer displays
04:03:18<@JAA>Maybe just one character?
04:03:24<OrIdow6^2>Whoops, did not mean to hit return
04:04:35<OrIdow6^2>*Oh, you're right, suppose the viewer displays them all at once
04:04:53<tech234a>One character per item would be ~64 zips per item
04:05:24<tech234a>And I guess the item IDs can be sequential and I would be able to request a collection be created?
04:06:55<@JAA>That would work I guess. I meant all in one item, but one tar/zip per character. That'd work out to ~6.5k directories per tar. Not sure how much the viewer can handle, but if the files are small enough, that might be fine.
04:08:27<@JAA>Could always try that first and then split it up further if it doesn't suffice.
04:09:36<OrIdow6^2>I think the viewer can handle enough that it's nearly crashed my browser in the past
04:09:59<tech234a>Yeah I think browser crashing is the problem I've had with large items previously
04:10:10qw3rty_ joins
04:10:36<tech234a>Can the viewer display the contents of zip files within zip files?
04:10:50<tech234a>I'd like to be able to link to specific files from the index search page
04:13:24qw3rty quits [Ping timeout: 240 seconds]
04:14:26<tech234a>Well, I'll try doing a zip within a zip within one item
04:15:07<@JAA>Just found an item with a zip in a tar, and that doesn't work.
04:15:13<@JAA>So I assume zip in zip doesn't either.
04:17:42<tech234a>Hmm...
04:17:54<@JAA>That tar has 17k files and opens fine in the viewer, although the browser doesn't like it all that much.
04:18:21<@JAA>For reference, it's this: https://archive.org/download/ftp-70.58.96.236
04:20:11<OrIdow6^2>https://archive.org/download/youtubeannotations_21/V5.tar looks to have about 300k files, and it lists, though it takes up about 1.2 GB of memory in browser
04:20:22<@JAA>That's somewhat comparable to when you'd split it up into one tar per character prefix, which would end up at 21k files or so (and 6.5k dirs).
04:21:19<tech234a>Perhaps I’ll just do one zip per first character then
04:22:05<tech234a>That should generally mean less than 2,000 files per zip
04:32:49<tech234a>Looked at a different character and found a much high file count... I think I'll try one folder for each first character, and one zip for each first two characters
05:39:35DogsRNice quits [Read error: Connection reset by peer]
08:49:16@arkiver quits [Quit: .]
08:49:30arkiver (arkiver) joins
08:49:30@ChanServ sets mode: +o arkiver
14:37:26Arcorann quits [Ping timeout: 244 seconds]
14:43:05dxrt quits [Client Quit]
14:43:23dxrt joins
14:43:26dxrt quits [Changing host]
14:43:26dxrt (dxrt) joins
15:13:51themadpro is now known as themadpro_
15:14:02themadpro_ is now known as themadpro
19:47:55prebala-publici quits [Client Quit]
19:48:15prebala-publici joins
19:52:26prebala-publici quits [Client Quit]
19:52:45prebala-publici joins
19:57:30prebala-publici quits [Client Quit]
19:57:50prebala-publici joins
20:06:56prebala-publici quits [Client Quit]
20:07:44prebala-publici_ joins
20:11:46prebala-publici_ quits [Read error: Connection reset by peer]
20:12:07prebala-publici joins
21:58:45prebala-publici quits [Read error: Connection reset by peer]
23:22:20<tech234a>Alright, files are now uploading after fixing some problems with case-insensitivity. I had to make the data directory case sensitive and use the `-ssc` option for 7-Zip. (Apparently Windows allows making individual directories case-sensitive using `fsutil file setCaseSensitiveInfo . enable`.)
23:23:03<tech234a>The upload is going rather slowly... is there any way to perhaps upload multiple files at once? Right now I
23:23:08<tech234a>Right now I
23:23:17<tech234a>Right now I'm using `ia upload youtube-community-contributions .`
23:23:48<tech234a>(whoops hit enter instead of the apostrophe)
23:30:17<@JAA>Imagine using Windows... ;-)
23:31:10<@JAA>You can run multiple `ia upload`, each uploading some distinct subset of files, but it's generally recommended against because it can cause problems. It's probably fine in this case though.
23:31:34<tech234a>Alright, thanks, will consider it