[05:02] <mwhudson> yay systemd migrated
[10:11] <xnox> fixing 6 digit bug report is nice
[10:11] <xnox> my irc is broken.
[10:11] <xnox> maybe not.
[10:11] <xnox> fixing 6 digit bug report is nice =) in my dput upload.
[10:22] <ginggs> xnox: good job!
[16:05] <TJ-> Is it known that phased updates and by-hash/...SHA256/$hash  can break fetching Package.{xz,gz} - are hashes really file-system symlinks?
[16:07] <cjwatson> They are not symlinks
[16:07] <cjwatson> Precisely because that would break
[16:11] <TJ-> cjwatson: got a user in #ubuntu curl-ing a focal-updates Package.xz by hash and it's not matching the hash he expects; difference is the Phasing percentage. I presume they are linked and the user has got hashes from a different phase?
[16:13] <cjwatson> TJ-: Can we get more exact details?
[16:13] <cjwatson> TJ-: Could be a busted mirror or overly-smart proxy or ...
[16:15] <TJ-> I'll copy the relevant comments:
[16:15] <TJ-> 16:50 <Rogach> curl -s --resolve archive.ubuntu.com:80:91.189.88.152 http://archive.ubuntu.com/ubuntu/dists/focal-updates/main/binary-amd64/by-hash/SHA256/125d5d9e5514f816011b6ba86ea5574caa09f2587502618e763e3ffcfac77a9f | sha256sum
[16:15] <TJ-> 16:50 <Rogach> 79d6950134a801405612775563ad754367e4b94366d6f9de2e92d6bf2c6cef80  -
[16:15] <cjwatson> TJ-: sha256sum matches for me
[16:15] <TJ-> cjwatson: me too!
[16:15] <cjwatson> TJ-: so to me that implies a proxy in the way
[16:16] <cjwatson> TJ-: may be explicit (http_proxy) or "transparent"
[16:16] <TJ-> they report no problems on other hosts, we suggested a proxy - only difference is the Phasing value
[16:16] <cjwatson> TJ-: curl -v may give some clues
[16:16] <TJ-> cjwatson: I suspect the transparent but wanted to be sure I correctly understood the Phasing operations
[16:16] <cjwatson> TJ-: phasing is unlikely to be very directly relevant - that probably just means that a proxy has cached something recent, and the most recent change to that Packages file happened to involve phasing
[16:17] <cjwatson> TJ-: if adding -H 'Pragma: no-cache' fixes it, then it's a buggy proxy
[16:17] <TJ-> cjwatson: right... thanks for the confirmation :)
[16:18] <cjwatson> TJ-: by-hash uses different URLs for each hash partly to minimize the chance of caching problems, but if a proxy is egregiously violating the spec and caching results from some different URL, there's not much we can do
[16:19] <TJ-> cjwatson: thanks ... that must be it. It's affecting apt as well as curl. Role on HTTPS :)
[21:35] <jawn-smith> Is any core dev available to retry https://autopkgtest.ubuntu.com/request.cgi?release=impish&arch=ppc64el&package=cwltool&trigger=curl/7.74.0-1.2ubuntu4
[21:43] <bdmurray> oooh ooh me!
[21:58] <jawn-smith> thanks bdmurray! It's always you haha
[21:59] <bdmurray> maybe nobody else trusts you?
[22:09] <jawn-smith> well the test passed, so did I earn +1 trust unit?
[23:43] <mwhudson> hey can someone review https://code.launchpad.net/~mwhudson/livecd-rootfs/+git/livecd-rootfs/+merge/406272 pls
[23:44] <mwhudson> (it's pretty simple)
[23:45] <sarnold> shell quoting .. simple .. dunno :)
[23:45] <sarnold> is that outer layer of quotes useful? helpful?
[23:46] <mwhudson> yeah i guess the results of command substitution don't get split do they?
[23:46] <sarnold> I don't *think* so but .. shell :)
[23:46] <mwhudson> If the substitution appears within double quotes, word splitting and pathname expansion are not performed on the results.
[23:46] <mwhudson> from man bash
[23:46] <mwhudson> so maybe they do?
[23:48] <mwhudson> please doctor it hurts https://paste.ubuntu.com/p/ypJDvh58jF/
[23:49] <mwhudson> oh ok so the rhs of an assignment is not split
[23:50] <mwhudson> (obviously it can only contain spaces to split on after some kind of expansion)
[23:52] <mwhudson> so the outer quotes are not needed but i'm inclined to leave them in anyway
[23:52] <sarnold> alright, cool, thanks for indulging my curiosity/concern :)
[23:53] <cjwatson> They're not needed, but they're harmless and IMO often best left in for pedagogical purposes (i.e. to reinforce the notion that you should have one layer of "" per variable expansion unless you have a damn good reason not to)
[23:53] <cjwatson> LGTM for whatever my out-of-date livecd-rootfs brain is worth
[23:53] <mwhudson> cjwatson: thanks
[23:54] <sarnold> cjwatson: ah that's a solid reason :) thanks