[01:42] <gevreeen> does someone know of a good (but small) project hosted on launchpad where I can test my git setup?
[01:42] <gevreeen> (still hours until I could get an account from staging)
[01:42] <gevreeen> (just cloning)
[01:43] <gevreeen> never mind. found one
[02:24] <gevreeen> where is git.launchpad.net's public key posted?
[02:26] <gevreeen> no sshfp/tlsa/srv/txt record?
[02:26] <gevreeen> is it posted somewhere?
[02:34] <gevreeen> or rather, can someone just post it here?
[02:35] <gevreeen> my suggestion is for to define sshfp rr alongside rrsig for dnssec
[02:35] <gevreeen> *for launchpad/canonical to define*
[02:39] <gevreeen> (also, when could we use ed25519/ed448 keys?)
[02:44] <sarnold> work is underway on the ed25519 keys; I get the impression we're owners of that support in twisted, and some of the libraries involved are particular about which versions of python they'll run on, etc. it sounds frustrating...
[02:44] <gevreeen> rewt from #git gave this link > https://help.launchpad.net/SSHFingerprints
[02:44] <gevreeen> looks good thus far
[02:46] <sarnold> sigh I hate ssh-keygen stuff so much. ~/.ssh/known_hosts has a key but figuring out how to get a fingerprint from that is just annoying
[02:46] <sarnold> any chance you know the command off-hand? :)
[02:47] <sarnold> 2048 SHA256:UNOzlP66WpDuEo34Wgs8mewypV0UzqHLsIFoqwe8dYo git.launchpad.net,162.213.33.95 (RSA)
[02:47] <sarnold> that's what ssh-keygen -l -f ~/.ssh/known_hosts  gives me for git.launchpad.net, anyway :) I hope that's helpful
[02:49] <gevreeen> it's probably bxxdecode from base64, but lemme check
[02:50] <gevreeen> got "incorrect padding", but let's see
[02:52] <gevreeen> https://man.openbsd.org/sshd.8 says "Each line in these files contains the following fields: marker (optional), hostnames, keytype, base64-encoded key, comment. The fields are separated by spaces." on SSH_KNOWN_HOSTS_FILE_FORMAT. so b64decode should have worked
[03:01] <gevreeen> even the base64 util gives "invalid input", let's see whether I can do this the other around to obtain the true base64 through hexstring
[03:04] <gevreeen> https://stackoverflow.com/questions/15490728/decode-base64-invalid-input < this saved me
[03:05] <gevreeen> by appending padding "=" I also got base64.b64decode to work
[03:11] <sarnold> hmm, I thought I saw something about known_hosts skipping the padding, and almost suggested adding it, but when I checked mine, it had the proper padding, so I deleted what I wrote. hah.
[03:19] <gevreeen> the example just below https://man.openbsd.org/sshd.8#SSH_KNOWN_HOSTS_FILE_FORMAT has padding
[03:20] <gevreeen> so probably just add the padding(s) if the base64 characters do not come in quartets
[03:31] <gevreeen> sarnold: by the by, just wondering, whether ubuntu/canonical have some saas/paas service offerings similar to openshift/ibm cloud which allows, say, writing and hosting a golang/python web service with restricted lite-tiers?
[10:38] <cjwatson> gevreeen: Canonical does a lot of cloud work, but it's more in the vein of helping our customers run private clouds - we don't have our own public cloud offering like that
[10:39] <cjwatson> gevreeen: Ed25519 should I hope be later this week.  Just trying to get the final pieces landed at the moment
[10:40] <cjwatson> gevreeen: No plans for Ed448
[10:45] <cjwatson> (Ed448 isn't interestingly better than Ed25519 unless elliptic curve cryptanalysis comes along a fair bit more, and OpenSSH hasn't yet felt it important enough to implement it, so I don't currently see it being worth the fairly large amount of heavy lifting needed to get support for it into Twisted and then into our services)
[11:04] <juliank> I'm subscribed to grub2 bugs directly. But if someone assigns my team to the task, I'll start receiving emails with "because your team ... was subscribed to it"
[11:04] <juliank> Those in turn then skip the inbox because that's a load of emails I would not be able to read (and there's never been a relevant email for me, until today)
[11:05] <juliank> I wonder if this makes sense, that "team is assigned to bug" beats "direct subscription to package"
[16:09] <juliank> cjwatson: In https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/4783/+packages cross-toolchain-base does not seem to try to use the PPA at all to build, and hence fails to find the new make I want to build again, that seems unusual
[16:09] <juliank> (also help contact seems out of hours)
[16:10] <juliank> oh I guess only the source package is published yet
[16:11] <juliank> but empty Packages files at least exist
[16:11] <juliank> now I see debs, so I'll go retry it
[16:14] <cjwatson> juliank: Yeah, the archive is only included if it has any published binary packages
[16:14] <juliank> yeah, sorry for disturbing :)
[16:15] <cjwatson> juliank: and re grub2 bugs, seems worth a bug report on LP at least - I haven't dug through the slightly byzantine bug notification code to confirm that behaviour but it's possible
[16:16] <juliank> cjwatson: then the usual xkcd comic comes in where you break another user's flow, so I don't know, but I can report it :D
[16:18] <cjwatson> Always a problem with notification header changes.  Not sure which way round makes most sense here TBH