--- Log opened Tue May 10 00:00:19 2022
00:01 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has quit [Ping timeout: 248 seconds]
00:38 < luke-jr> fanquake: FWIW, dropping #25099 commit into v23.0 results in mismatching output files
00:38 < gribble> https://github.com/bitcoin/bitcoin/issues/25099 | guix: bump time-machine to 998eda3067c7d21e0d9bb3310d2f5a14b8f1c681 by fanquake · Pull Request #25099 · bitcoin/bitcoin · GitHub
00:39 < fanquake> well yes, because you've changed the build environment
00:39 < fanquake> if you point me to the branch you're building i'll build it, and you can compare
00:40 < fanquake> but given that it's producing i'll take that to mean the build is working
00:40 < fanquake> which is the actual outcome we were after
00:41 < fanquake> *producing output files
00:41 < luke-jr> right (not 100% done yet tho, or I'd have commented)
00:41 < luke-jr> I didn't make a branch because I wanted it to match the tag - but all I did was paste in the new commit hash
00:42 < luke-jr> I would think we want old tags to be reproducible? not just the next one working when it's time..
00:42 < fanquake> Old tags are reproducible
00:43 < fanquake> There is a difference between reproducibility over time, which we have, and the build just not working. Your build doesn't work.
00:43 < fanquake> In 6 months time, people can go back and build v23.0 and the outputs will still match
00:44 < fanquake> That is the whole point of the time-machine setup
00:45 < luke-jr> if the build fails, there's no outputs to match :p
00:45 < fanquake> if you reported actionable information earlier, maybe you would have been able to build 23.0
00:46 < fanquake> amazing that 2-3 days after opening an issues we've found the guix fix, it's basically already solved in our repo, and the ultimate issue has a PR open to fix it upstream
00:46 < luke-jr> I assume at the time 23.0 was released, it worked
00:46 < luke-jr> unless literally nobody else is doing guix from source
00:47 < luke-jr> btw I'm not complaining, I'm trying to figure out a solution
00:48 < fanquake> and I am happy to keep working through this until we've got one
00:48 < fanquake> glad we seem to be nearly there
00:50 < fanquake> also the claim about "literally no-one else doing guix from source" is hardly correct
00:50 < luke-jr> see context..
00:50 < fanquake> most builders would have built (and run the unit tests) and then cached the libgit2 package likely many many months ago
00:51 < luke-jr> ah, good point
00:51 < fanquake> so only users building after the point in time that the time dependent unit test started failing would see the build failure
00:53 < luke-jr> diffing bitcoin-cli, it looks like the only change is the debug file checksum
00:54 < luke-jr> and the debug files have basically nothing in common XD
00:55 < luke-jr> probably compressed or something
01:04 -!- jonatack [~jonatack@user/jonatack] has joined #bitcoin-core-builds
03:10 -!- belcher [~belcher@user/belcher] has quit [Quit: Leaving]
03:15 -!- belcher [~belcher@user/belcher] has joined #bitcoin-core-builds
03:26 -!- jonatack [~jonatack@user/jonatack] has quit [Ping timeout: 248 seconds]
04:46 -!- davidbakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has joined #bitcoin-core-builds
04:47 -!- david-bakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has quit [Ping timeout: 246 seconds]
04:58 -!- jonatack [~jonatack@user/jonatack] has joined #bitcoin-core-builds
06:51 -!- ghost43 [~ghost43@gateway/tor-sasl/ghost43] has quit [Remote host closed the connection]
06:52 -!- ghost43 [~ghost43@gateway/tor-sasl/ghost43] has joined #bitcoin-core-builds
07:02 -!- lukedashjr [~luke-jr@user/luke-jr] has joined #bitcoin-core-builds
07:02 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 240 seconds]
07:03 -!- lukedashjr is now known as luke-jr
08:01 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has joined #bitcoin-core-builds
08:06 -!- jonatack [~jonatack@user/jonatack] has quit [Quit: Connection closed]
08:51 < dongcarl> Okay I made the wrong conclusion yesterday because I didn't realize luke-jr was using Guix System (Guix as an OS) so I'd like to clear things up. The current time-machine commit on bitcoin's master uses libgit2 1.3.0, which does not contain the faulty revparse date test (that was introduced in libgit2 1.4.0). The reason why luke-jr encountered the
08:51 < dongcarl> faulty test was because his Guix System is based on a newer version of Guix, which uses libgit2 1.4.x, which DOES contain the faulty revparse date test.
08:53 < dongcarl> So it is not absolutely necessary to bump the time-machine for the libgit2 problem itself (but we might still want to do so for clang 14).
08:53 < dongcarl> sipa: GuixSD was the old name for Guix System, which is "Guix as an OS"
08:56 < fanquake> dongcarl: confusing that the error output in #25082 shows the build of 1.3.0 as failing then? Shouldn't that be 1.4.0 failing to build?
08:56 < gribble> https://github.com/bitcoin/bitcoin/issues/25082 | Guix build w/o substitutes failing at libgit2 check · Issue #25082 · bitcoin/bitcoin · GitHub
08:56 < luke-jr> dongcarl: that's not correct AFAIK
08:57 < luke-jr> the issue is in 1.3.0, and I'm pretty sure the one I have installed is 1.1
08:59 < fanquake> The commit I mention in #25099 (https://git.savannah.gnu.org/cgit/guix.git/commit/?id=f5fe0082abe4547f3fb9f29d8351473cfb3a387b) is also a patch to libgit2 when it is version 1.3.0
08:59 < gribble> https://github.com/bitcoin/bitcoin/issues/25099 | guix: bump time-machine to 998eda3067c7d21e0d9bb3310d2f5a14b8f1c681 by fanquake · Pull Request #25099 · bitcoin/bitcoin · GitHub
09:01 < dongcarl> Oh you're right
09:02 < dongcarl> Urgh okay I'm gunna stop confusing people any more. In any case the many fixes are already out there :-)
09:34 < luke-jr> dongcarl: none that fixes the v23.0 tag build ? or am I wrong?
09:57 < dongcarl> luke-jr: If you're comfortable messing with the system time, you can turn back time 1 year, build the derivation, then turn it back
09:57 < dongcarl> Not an ideal fix, no. Hard to have a good fix until Linux has good time namespace support
10:00 < luke-jr> libfaketime has been around for years <.<
10:00 < luke-jr> idk why it should have to be in the kernel
10:01 < luke-jr> but an ideal fix is probably to patch out or skip the test, without changing the hashes everywhere that shouldn't care
10:26 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has quit [Ping timeout: 248 seconds]
10:41 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has joined #bitcoin-core-builds
12:27 -!- lukedashjr [~luke-jr@user/luke-jr] has joined #bitcoin-core-builds
12:29 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 260 seconds]
12:29 -!- lukedashjr is now known as luke-jr
14:52 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has quit [Ping timeout: 248 seconds]
15:13 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has joined #bitcoin-core-builds
15:53 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has quit [Ping timeout: 240 seconds]
16:01 -!- davidbakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has quit [Remote host closed the connection]
18:48 -!- david-bakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has joined #bitcoin-core-builds
21:44 -!- ghost43 [~ghost43@gateway/tor-sasl/ghost43] has quit [Quit: Leaving]
21:44 -!- ghost43_ [~ghost43@gateway/tor-sasl/ghost43] has joined #bitcoin-core-builds
22:18 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has joined #bitcoin-core-builds
22:33 -!- david-bakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has quit [Remote host closed the connection]
22:34 -!- david-bakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has joined #bitcoin-core-builds
22:56 -!- david-bakin [~david-bak@c-174-61-163-5.hsd1.wa.comcast.net] has quit [Ping timeout: 256 seconds]
23:53 -!- sipsorcery [~sipsorcer@2a02:8084:6180:500::12b] has quit [Ping timeout: 248 seconds]
--- Log closed Wed May 11 00:00:20 2022