--- Log opened Thu Feb 13 00:00:39 2020 03:04 -!- Hildegard4Rau [~Hildegard@ns334669.ip-5-196-64.eu] has joined #bitcoin-builds 03:31 -!- jonatack [~jon@2a01:e0a:53c:a200:bb54:3be5:c3d0:9ce5] has quit [Ping timeout: 240 seconds] 05:41 -!- jonatack [~jon@2a01:e0a:53c:a200:bb54:3be5:c3d0:9ce5] has joined #bitcoin-builds 06:40 < MarcoFalke> jonasschnelli: I think https://bitcoin.jonasschnelli.ch/ is still missing the new SDK in the inputs folder 06:42 -!- Hildegard4Rau [~Hildegard@ns334669.ip-5-196-64.eu] has quit [Ping timeout: 240 seconds] 07:31 < jonasschnelli> MarcoFalke: thanks. I'm currently replacing that server. So ignore the site for now. 08:50 -!- mfoolb [~mfoolb@gateway/tor-sasl/mfoolb] has quit [Remote host closed the connection] 08:50 -!- mfoolb [~mfoolb@gateway/tor-sasl/mfoolb] has joined #bitcoin-builds 09:15 -!- mfoolb [~mfoolb@gateway/tor-sasl/mfoolb] has quit [Remote host closed the connection] 09:15 -!- mfoolb [~mfoolb@gateway/tor-sasl/mfoolb] has joined #bitcoin-builds 10:40 < jonatack> dongcarl: hi :) wrt guix, are #17595 and #17920 the current tip of progress? 10:43 < dongcarl> jonatack: Yes! The things they are based on might have changed and I haven't rebased, but they are definitely the tip of progress! 10:44 < dongcarl> Hmmm... wondering if I can get gribble here to pull up PR descriptions... 10:44 < jonatack> dongcarl: thanks! yes, will check back soon to see if i can manage to test, just getting up to speed a bit on this. 10:48 < dongcarl> jonatack: No worries :-) For some context: 10:48 < dongcarl> 1. For #17595, we've got everything working, but reproducibility has been elusive. That's the last thing we need to accomplish before we merge. 10:48 < dongcarl> 2. For the #17920 lineage of commits, you're better off testing #18072, which requires a macOS SDK with libc++ headers built in (SDK generation instructions in the PR description). 10:48 < dongcarl> Thanks in advance if you find the time to help! 10:50 < jonatack> dongcarl: super helpful, thanks! 10:50 < dongcarl> :-) 11:39 < jonasschnelli> is there a way how I can speedup the gitian install (install.log) process? It feels like its installed the same tools over and over (for my nighly build server) 11:39 < jonasschnelli> A cache or so? 11:39 < jonatack> +1 ... it's slow 11:56 < jonasschnelli> My previous gitian installation was much faster (on a slower server). So I wonder. 11:56 < jonasschnelli> I guess it was mainly the install process 12:07 -!- achow101 [~achow101@unaffiliated/achow101] has quit [Ping timeout: 240 seconds] 12:59 < dongcarl> fanquake: When you added `-Wl,--no-insert-timestamp` in your builds, you did this in contrib/guix/libexec/build.sh? Like adding a HOST_LDFLAGS for mingw? 13:06 < fanquake> dongcarl iirc yea. Just added a HOST_LDFLAGS 13:07 < dongcarl> thanks! 13:07 < fanquake> See also #18135 13:08 < fanquake> I'd like to add that to regular builds, so that consecutive bitcoinds are deterministic 13:10 < ryanofsky> is dongcarl/guix a temporary fork? do the tweaks get upstreamed? https://github.com/dongcarl/guix/commits/b3a7c72c8b2425f8ddb0fc6e3b1caeed40f86dee 13:12 < dongcarl> ryanofsky: Yes, my tweaks are slowly being upstreamed, the last two is stalled because they're a bit ugly... See discussion here: https://lists.gnu.org/archive/html/guix-patches/2019-12/msg00206.html 13:13 < dongcarl> fanquake: I'm not sure we should implicitly add these flags... Perhaps a explicit configure time flag for determinism would be nice (just like binutils does) 13:13 < fanquake> dongcarl: why not? 13:14 < fanquake> I had considered a flag that would bundle reproducibility flags together. 13:15 < dongcarl> fanquake: because I think we should only implicitly add flags that more or less critical to the correct and secure operation of the binary, and determinism flags don't seem to be. I'd have no objections with a flag that bundles the reproducibility things though 13:16 < fanquake> I'd argue that determinism falls pretty close to the "secure" category. 13:17 < fanquake> I mean we've added a bunch of flags, or for macOS, that aren't necessarily security related. More performance/binary size specific. 13:17 < fanquake> I will investigate a --enable-reproducibility or similar option though 13:18 < dongcarl> I see your point, but not totally convinced... Will think on it! 13:21 < wumpus> jonasschnelli: there is some trick to run the apt upgrade in the base image, instead of the target image, cutting the upgrade time for each image copy 13:21 -!- ghost43 [~daer@gateway/tor-sasl/daer] has joined #bitcoin-builds 13:21 < wumpus> but yes that step takes a lot of time 13:22 < dongcarl> o/ ghost43 13:22 < ghost43> hi all 13:23 < dongcarl> ghost43 fixed the NSIS reproducibility in electrum, and I think we have a lot to share between the two projects, great to have you in the chat! 13:29 < ghost43> >>> 0xcd81 - 0xcaa3 13:29 < ghost43> 734 13:29 < ghost43> >>> 15272833 - 15272099 13:29 < ghost43> 734 13:29 < ghost43> https://github.com/bitcoin/bitcoin/pull/17595#issuecomment-582696467 13:29 < ghost43> that first diff is just the file size 13:30 < fanquake> Any idea why they are different sizes though? 13:31 < fanquake> I can dump a full diff somewhere if required. 13:31 < dongcarl> Hmmmmm 13:32 < dongcarl> I'm running uncompressed right now 13:32 < dongcarl> I think it'll give clues 13:32 < dongcarl> Idk how lzma works 13:32 < dongcarl> But perhaps a smaller diff can cascade to size diff once compressed? 13:36 < fanquake> I think there is another optimization flag you could turn off as well 13:36 < fanquake> Setdatablockoptimize or similar 13:39 < dongcarl> I'm seeing same size right now 13:39 < dongcarl> let me see what happens with the block optimize 13:40 < dongcarl> Oh I don't think we use the datablock optimize 13:40 < fanquake> We don't set it explicitly, but it's on by default I'm pretty sure p 13:41 < dongcarl> Ah I see 13:41 < dongcarl> ghost43: did you guys end up having to use faketime? 13:43 < ghost43> no. we just have a couple lines of "find -exec touch -d '2000-11-11T11:11:11+00:00' {} +" here-and-there 13:44 -!- aqua42 [~aqua42@amsterdam3.jp.net] has joined #bitcoin-builds 13:44 < dongcarl> ghost43: Okay good to know! 13:47 -!- achow101 [~achow101@unaffiliated/achow101] has joined #bitcoin-builds 13:48 < ghost43> actually looking at the docs now, "SetDateSave off" might work too? have not tried that one. 13:48 < dongcarl> Oh! That seems... important... 13:48 < ghost43> maybe this doc link is better: https://nsis.sourceforge.io/Docs/Chapter4.html 13:48 < dongcarl> ghost43: did you guys set that? 13:48 < ghost43> what fanquake said is also here under "4.8.2.6 SetDatablockOptimize" 13:49 < ghost43> dongcarl: no we don't set SetDateSave, I have just found about it myself :D maybe we could do away with all the "touch" timestamp modifications with that 13:50 < dongcarl> ghost43: gotcha... will try it out, and that's a good chapter to look for things 13:50 * dongcarl gunna head home while things build 13:51 < fanquake> Guess I should get up and start doing some builds then 14:24 -!- achow101 [~achow101@unaffiliated/achow101] has quit [Ping timeout: 265 seconds] 14:25 -!- achow101 [~achow101@unaffiliated/achow101] has joined #bitcoin-builds 14:27 -!- achow101 [~achow101@unaffiliated/achow101] has quit [Client Quit] 14:31 -!- achow101 [~achow101@unaffiliated/achow101] has joined #bitcoin-builds 19:05 < dongcarl> I think I've got reproducible builds windows builds working... Just experimenting and minimizing the diff now 19:05 < fanquake> dongcarl: nice! 19:06 < dongcarl> :-) 19:07 < fanquake> Look forward to running some builds after you push something up 19:07 < dongcarl> will do! 19:57 < dongcarl> fanquake: ready! 19:57 < fanquake> any chance you can rebase? :o 19:57 < fanquake> I'll run some builds this arvo 19:59 < dongcarl> Oh over master? 19:59 < dongcarl> will do if so 19:59 < fanquake> yea 20:00 < fanquake> conflict in guix-build.sh 20:36 < dongcarl> done! 20:36 < fanquake> dongcarl: awesome. Will take a look! --- Log closed Fri Feb 14 00:00:40 2020