01:08:13<@kiska>brad: Take a gander towards https://tracker.archiveteam.org/googleplus/ :D
02:00:01Dallas quits [Client Quit]
02:03:41Dallas (Dallas) joins
02:35:40<brad>@kiska: I definitely plan on running my own warrior. Maybe even a team of warriors. We’ll have to see how many I can afford to run. I know a bit about running containers in ECS on AWS, but I gather that would probably be rather expensive for running warriors. So, I might learn about the process there, and then try to figure out how to run containers somewhere else that is less expensive.
02:36:35<brad>But regardless of where they are running, I would also like to contribute to the process of automating their updates and coordination of tasks, whether through a traditional CI/CD process, or something else.
02:42:50<brad>As I understand it, the archive bot is for smaller projects, and the warriors are for larger ones? What is the break over, and how do you determine how large a crawl will be?
02:45:50<brad>I’ve also been involved in other projects like Mailman and NTP, since 2003. Would it be appropriate for me to request that they also get explicitly mirrored?
03:44:20etnguyen03 (etnguyen03) joins
04:40:46qw3rty__ joins
04:44:23qw3rty_ quits [Ping timeout: 252 seconds]
04:51:53<OrIdow6>brad: So I will first mention that the Warrior proper (the Virtualbox image) has been broken for ~2/3 of a year(?), basically due to its copy Ubuntu being too old for the zstd dependency of newer versions of ArchiveTeam's wget fork (wget-lua/wget-at), and there are several different parts, with several people responsible for them, that would have to be fit together to get it working again
04:52:13<OrIdow6>*and due to there being
04:52:52<OrIdow6>There is an intended new version that uses Docker that will hopefully come out soon - but I don't know the details
04:53:41<OrIdow6>In the meantime, most people run the scripts (which I will call "warrior projects", despite this being somewhat of a misnomer at present) in Docker anyway
04:55:37<OrIdow6>A warrior project essentially divides the site into pieces (items), sends each item out to a worker (warriors/Docker containers/raw scripts), and then sends the resultant warc files back to be concatenated together and sent to the Internet Archive
04:56:48<OrIdow6>These require writing a script for the workers to run (which controls that wget fork) and a lot of other preparation (such as making a list of items to be sent)
04:59:03<OrIdow6>ArchiveBot is more or less a regular recursive web crawl that only recurses through pages under a single URL prefix (which must be a host, at minimum) and to first-degree outlinks (as well as getting page requisites etc.)
05:00:35<OrIdow6>Which requires much less preparation, and has a much more limited degree of control (adding blacklist regexes and adjusting the inter-request delay and concurrency) that is exercised after the job is started
05:01:47etnguyen03 quits [Remote host closed the connection]
05:02:43<OrIdow6>Which means it takes much less labor, but is not as scalable, to site size (I think I've heard the figure of a few million URLs thrown around for when it starts getting slow) or to speed (there is a maximum of 9? threads, and those all have to be on the same machine)
05:03:26<OrIdow6>And of course there's no opportunity to write a script for the site, which means there's a lot less flexibility
05:05:40<OrIdow6>Occasionally rarer tools are used, and as with most things there are caveats to both these descriptions
05:06:27Ryz quits [Remote host closed the connection]
05:07:14Ryz (Ryz) joins
05:07:59<OrIdow6>I think there is a fair bit of automation with regard to building the Docker containers etc.? Not closely involved in that area
05:10:06<OrIdow6>And you can at least ask for those things to be saved in #archivebot , but the bigger it is, the less likely it is to get saved if it's simply proactive
05:14:39<OrIdow6>---------------------------
05:15:45<OrIdow6>Something that shows just how much SmackJeeves has been abandoned: the Forum link that appears fairly prominently on every non-reader page has been broken since July
05:19:49<OrIdow6>" I can't think of a single corporate acquisition like this that went well for the site's users ;_;" - you said it "eishiya" (https://web.archive.org/web/20191205204431/https://forum.smackjeeves.com/viewtopic.php?f=4&t=19772)
06:25:42godane (godane) joins
06:57:24tzt quits [Ping timeout: 240 seconds]
07:16:25Arcorann (Arcorann) joins
07:27:22Doranwen joins
07:46:20Mateon1 quits [Remote host closed the connection]
08:19:39Mateon1 joins
09:28:08Ryz quits [Remote host closed the connection]
09:28:57Ryz (Ryz) joins
11:29:56VADemon joins
12:38:15AlsoHP_Archivist joins
12:41:04HP_Archivist quits [Ping timeout: 240 seconds]
12:45:44AlsoHP_Archivist quits [Ping timeout: 240 seconds]
12:46:00AlsoHP_Archivist joins
13:25:46VADemon quits [Client Quit]
13:45:28AlsoHP_Archivist quits [Client Quit]
13:45:44HP_Archivist (HP_Archivist) joins
13:56:44Arcorann quits [Ping timeout: 240 seconds]
14:28:32etnguyen03 (etnguyen03) joins
16:16:35hooway joins
17:14:32hooway quits [Client Quit]
17:21:36hooway joins
17:40:07tzt joins
20:13:30Doranwen quits [Ping timeout: 252 seconds]
20:15:19Doranwen joins
20:23:59<themadpro>Random question from someone who has no idea how .zip compression works:
20:24:16<themadpro>Is it possible to (in theory) crawl the contents of a file without uncompressing?
20:24:32<themadpro>Assuming the .zip archive is homogenous (all text files or all .html files)
20:25:46<OrIdow6>What do you mean, "crawl the contents"?
20:28:45<Sanqui>it's possible to "stream" a zip file without having to decompress on disk yes.
20:30:05<@JAA>Any decompression tool basically reads the data, decompresses in memory, then writes the decompressed version to disk. You can always get rid of the last step with sufficient motivation and just do something with the decompressed data in memory.
20:32:15<themadpro>Orldow6: Sorry for the abrupt afk, basically the Save Yahoo Groups/Yahoo Geddan crew decided to try hunting for Mediafire links (#mediaonfire) and out of curiosity someone asked if it might be possible to search for URL strings without unzipping a zip archive
20:33:24<themadpro>So that's a regular expression that goes a little something like r`http://mediafire.com/file/[0-9a-z]+`
20:33:54<themadpro>The streaming idea seems interesting, how would that work?
20:34:23<themadpro>If we wanted to search for matches while "streaming" a zip?
20:34:33<OrIdow6>-p
20:35:44<OrIdow6>On the "unzip" that gets bundled with (I assume to be) most Linux distros
20:41:10<OrIdow6>By the way, I think Pyxia in #mediaonfire gave the format of some sort of short MF link - might want to include that as well
20:42:43<thuban>mfi.re
21:02:43<atphoenix>themadpro, is that conversation on the SYG discord?
21:27:15<themadpro>yeah
21:27:23<themadpro>feel free to take it there!
21:27:30<themadpro>(I'm pretty busy rn)
21:43:04tzt quits [Ping timeout: 240 seconds]
22:59:45godane quits [Client Quit]
23:57:21Connection closed.