| === walbon is now known as gwalbon_ | ||
| ginggs | seb128: note libzstd was merged from 1.3.3+dfsg-2 in Debian NEW queue, which was rejected, and a different 1.3.3+dfsg-2 was uploaded. missing at least the udeb stuff | 08:03 |
|---|---|---|
| seb128 | ginggs, k, then a merge or dropping those changes from Ubuntu as well? | 08:05 |
| ginggs | seb128: hmm, depends how attached xnox is to those udebs | 08:05 |
| === zumbi_ is now known as zumbi | ||
| cpaelzer | Skuggen: sil2100: hiho | 11:08 |
| Skuggen | o/ | 11:08 |
| cpaelzer | sil2100: I'd want for Skuggen to be able to use bileto | 11:08 |
| cpaelzer | I know the bileto-users LP group | 11:08 |
| cpaelzer | but there was something else that my brain doesn't want to remember | 11:08 |
| cpaelzer | sil2100: do you know what else he'd need to fully use it? | 11:08 |
| rbasak | For background, Skuggen works for MySQL upstream at Oracle. He generally looks after MySQL packaging in Debian and Ubuntu. He's not an uploader currently but I hope he'll get MySQL PPU soon. | 11:09 |
| sil2100 | Hey! Give me a minute guys | 11:30 |
| === nOOb__ is now known as LargePrime | ||
| sil2100 | cpaelzer: sorry it took so long - so, I guess Skuggen would need to do manual uploads to Bileto, right? | 13:23 |
| sil2100 | Skuggen: do you have any upload rights to the archive right now? Could I get your LP id? | 13:23 |
| sil2100 | Since it's a bit tricky right now, we don't have any hard-written policies who should be able to use Bileto, but we have some rules of the thumb - I actually wanted to bring up the topic of Bileto for a while, but we're still not really certain about its fate | 13:24 |
| sil2100 | Maybe it's time for me to start the topic anyway | 13:24 |
| sil2100 | For general Bileto MP-based usage we mostly only require ubuntu membership, but for manual uploads - well, those we usually only gave to people with at least any upload rights to the archive | 13:26 |
| sil2100 | I think Bileto membership should be also reviewed and decided by the DMB, IMO, and that's also one thing I wanted to discuss | 13:26 |
| cpaelzer | sil2100: yes he'd want to upload to the bileto ppas then | 13:27 |
| sil2100 | But since Bileto is still in bare maintenance mode, well, I guess we'd first have to decide if we want to invest the DMB's time in Bileto membership control | 13:27 |
| cpaelzer | rbasak: do you know the current state of Skuggen 's upload permissions | 13:28 |
| cpaelzer | I know more are about to come, but what does he have like now? | 13:28 |
| sil2100 | In case he has no upload permissions, if he really wants to have Bileto upload rights ASAP then I'd recommend attending a DMB meeting for this | 13:29 |
| sil2100 | And maybe actually following the procedures as per any other upload rights | 13:29 |
| cpaelzer | sil2100: he might have some, but no core-dev or anything similar wide-reaching | 13:29 |
| cpaelzer | sil2100: this is mostly to allow him to quickly iterate on uploads and trigger dep8 tests | 13:30 |
| cpaelzer | sil2100: would there be any alternative way to get the "LP base autopkgtest exec of all deps" that Bileto provides to a non Bileto ppa? | 13:31 |
| rbasak | He currently has no upload access. | 13:34 |
| rbasak | He intends to apply for MySQL PPU imminently. | 13:34 |
| rbasak | https://launchpad.net/~lars-tangvald | 13:34 |
| sil2100 | rbasak, cpaelzer: ok, I think in this case I'd defer to the DMB, I think we'd actually need a normal, formal review as for any other PPU access | 13:35 |
| rbasak | What are the implications of Bileto access in terms of the Ubuntu archive? | 13:35 |
| sil2100 | rbasak: first of all, the user gains access to devirt builders | 13:36 |
| sil2100 | rbasak: also, since there is no ticket-specific ACL in Bileto, having upload access to Bileto PPAs means anyone can affect any other people's ongoing tickets | 13:37 |
| rbasak | access to devirt builders> I'd say that's more of a matter for infrastructure maintainers - Canonical and therefore Launchpad admins maybe, rather than the DMB? | 13:38 |
| sil2100 | Meaning a melavolent user could upload something to someone's other PPA, which (even though audited) could go unnoticed and therefore get uploaded to the archive | 13:38 |
| rbasak | That's a bit tricker. | 13:39 |
| rbasak | trikier | 13:39 |
| rbasak | trickier | 13:39 |
| rbasak | \o/ | 13:39 |
| tkamppeter | cyphermox, hi | 13:40 |
| Skuggen | I don't think it's very urgent, so maybe try to get MySQL PPU access first | 13:40 |
| rbasak | Yeah that seems the easiest. | 13:40 |
| sil2100 | The unwritten rule so far was that anyone that has at least PPU access is 'trustworthy' enough to use Bileto | 13:41 |
| rbasak | We would still need to decide if PPU uploaders are permitted to upload to Bileto in general. | 13:41 |
| rbasak | OK | 13:41 |
| rbasak | Then we can just do that. | 13:41 |
| rbasak | Thanks | 13:41 |
| Skuggen | Right. Thanks :) | 13:41 |
| sil2100 | Skuggen: sorry about that, hope to see your PPU application soon! | 13:42 |
| tkamppeter | I have a problem with ppc64el. The build of ghostscript gets stuck on it and the build server kills the process after 2 hours, see https://launchpad.net/ubuntu/+source/ghostscript/9.23~dfsg+1-0ubuntu1 | 14:01 |
| tkamppeter | slangasek, could you help on this? | 14:01 |
| acheronuk | tkamppeter: ah. was just asking in #ubuntu-release about that | 14:11 |
| acheronuk | being retried, though I am not sure that will help | 14:11 |
| slangasek | in any event, it means the logs are gone now so I can't guess at the problem | 14:12 |
| ginggs | tkamppeter: ppc64el in ubuntu defaults to building with -O3, I suggest you try building in your PPA forcing -O2 and seeing what happens | 14:14 |
| acheronuk | slangasek: drat. build currently hung at the same point, which is then killed at the 150 min timeout without much diagnostic to be honest | 14:16 |
| acheronuk | slangasek: I have the last log open in a browser tab if you want it? | 14:18 |
| ginggs | acheronuk: the URL might help | 14:18 |
| ginggs | i don't think they are removed right away | 14:18 |
| acheronuk | https://launchpadlibrarian.net/382631754/buildlog_ubuntu-cosmic-ppc64el.ghostscript_9.23~dfsg+1-0ubuntu1_BUILDING.txt.gz | 14:19 |
| Laney | It'd already failed at least one retry, so this one is quite optimistic. | 14:33 |
| acheronuk | :/ | 14:34 |
| Laney | "10/08 15:33:54 <ubottu> Review: quiet '$~a' set on Wed Aug 1 14:41:29 2018 in #ubuntu-meeting, link: https://ubottu.com/bans.cgi?log=78631" | 14:34 |
| Laney | Why did the bot just PM me that? | 14:34 |
| Laney | Oh, it's asking me to review a channel mode change since I'm an op? | 14:35 |
| Laney | Yes, I approve of this change ✅ | 14:36 |
| tkamppeter | ginggs, I had used PPAs for some tests some years ago, but they only built for amd64 and i386. how can one make them build for ppc64el? | 14:49 |
| ahasenack | tkamppeter: it's a checkbox in the details page | 14:51 |
| ginggs | tkamppeter: go to your PPA in launchpad, click on 'Change details', scroll down to Processors and click on the checkboxes | 14:52 |
| ahasenack | sil2100: hi, I saw the conversation about bileto earlier, and I would also like bileto rights. I have PPU for landscaoe-client, and ubuntu server packageset. my lp id is ~ahasenack | 15:24 |
| tkamppeter | ahasenack, ginggs, thanks. | 15:34 |
| sil2100 | ahasenack: hey! Ok, I guess I can work with that, one moment | 15:52 |
| blackboxsw | hrm, anyone know why I might be getting Error checking signature during a dput operation for a package upload (line 24) https://pastebin.ubuntu.com/p/Wq3jqvTmgx/? This is my first time trying to upload with this key (well and a cloud-init upload attempt yesterday) | 16:00 |
| smoser | blackboxsw: i think it is probalby the local portion of 'dput' that is complaining there. | 16:04 |
| smoser | it *does* seem to upload. | 16:05 |
| smoser | blackboxsw: what happens if you just 'gpg --verify ../out/<that-dsc-file>' | 16:05 |
| blackboxsw | ahhh | 16:07 |
| blackboxsw | gpg: Good signature from "Chad Smith <chad.smith@canonical.com>" [unknown] | 16:07 |
| blackboxsw | gpg: WARNING: This key is not certified with a trusted signature! | 16:07 |
| blackboxsw | gpg: There is no indication that the signature belongs to the owner | 16:07 |
| smoser | right. you dont trust yourself. | 16:07 |
| blackboxsw | I bet I need to edit my signature and add trust | 16:07 |
| smoser | thats something maybe elizabot can help you with :) | 16:07 |
| smoser | but if that was the only issue with your cloud-init upload yesterday, i would have expected it to go through | 16:08 |
| smoser | and it did not seem to. | 16:08 |
| smoser | curtin did | 16:08 |
| smoser | https://launchpad.net/ubuntu/+source/curtin/18.1-44-g2b12b8fc-0ubuntu1 | 16:08 |
| blackboxsw | hrm ok just cloud-init didn't then | 16:10 |
| blackboxsw | https://launchpad.net/ubuntu/+source/cloud-init/18.3-24-gf6249277-0ubuntu1 | 16:10 |
| blackboxsw | which was prior to me associating by new key with my launchpad account | 16:10 |
| blackboxsw | ok so I might be able to upload a new version of cloud-init to check | 16:11 |
| smoser | yeah. that makes sense. but id' have thoguht you'd have gotten a rejection or something. | 16:13 |
| tkamppeter | ginggs, I have compared the end of the log of the ppc64el build of ghostscript with the log of the build on my amd64 machine. It does not hang on a compiler action biut on an action of creation of simple text files (./soobj/aux/echogs call, echogs source is ./base/echogs.c), so -O3/-O2 replacement on gcc calls is most probably not the solution. | 16:17 |
| ginggs | tkamppeter: so is echogs.c compiled with -O3 or -O2? | 16:19 |
| cjwatson | blackboxsw,smoser: you don't get a rejection if the key isn't registered in Launchpad | 16:29 |
| cjwatson | I mean, not a rejection email | 16:29 |
| tkamppeter | ginggs, this I have to see when the full log gets released and I succeed to catch it before someone else hits the retry button. | 16:29 |
| blackboxsw | cjwatson: thanks that explains why I didn't see any email about it | 16:38 |
| tkamppeter | ginggs, slangasek, ahasenack: The ghostscript build finished and I downloaded the log now. Compilation of base/echogs.c (the first compailation at all) uses BOTH -O2 and -O3, what is actually used then? | 16:48 |
| tkamppeter | Example gcc command line: | 16:48 |
| tkamppeter | gcc -fPIC -O2 -Wdate-time -D_FORTIFY_SOURCE=2 -Wall -Wstrict-prototypes -Wundef -Wmissing-declarations -Wmissing-prototypes -Wwrite-strings -fno-strict-aliasing -Werror=declaration-after-statement -fno-builtin -fno-common -Werror=return-type -DHAVE_STDINT_H=1 -DHAVE_DIRENT_H=1 -DHAVE_SYS_DIR_H=1 -DHAVE_SYS_TIME_H=1 -DHAVE_SYS_TIMES_H=1 -DHAVE_INTTYPES_H=1 -DHAVE_LIBDL=1 -DGX_COLOR_INDEX_TYPE="unsigned long long" -D__USE | 16:48 |
| tkamppeter | _UNIX98=1 -g -O3 -fdebug-prefix-map=/<<BUILDDIR>>/ghostscript-9.23~dfsg+1=. -fstack-protector-strong -Wformat -Werror=format-security -DUSE_LIBPAPER -I/usr/include/powerpc64le-linux-gnu -fno-strict-aliasing -Wl,-Bsymbolic-functions -Wl,-z,relro -DHAVE_POPEN_PROTO=1 -DHAVE_POPEN_PROTO=1 -I./base -o ./soobj/aux/echogs ./base/echogs.c -lm -ldl -lidn -lpaper -ltiff -rdynamic -lfontconfig -lfreetype -lfreetype -lz | 16:48 |
| ahasenack | is that what is failing? | 16:49 |
| slangasek | tkamppeter: later -O option wins | 16:50 |
| cjwatson | If you use multiple '-O' options, with or without level numbers, the | 16:50 |
| cjwatson | last such option is the one that is effective. | 16:50 |
| cjwatson | says info gcc | 16:50 |
| ginggs | tkamppeter: i think the second one overrides the first | 16:50 |
| tkamppeter | Thanks all, so it is actually -O3. | 16:57 |
| tkamppeter | ahasenack, the shown gcc command line is not failing. It is the one which builds ./sobin/aux/echogs and what is failing is a call of ./sobin/aux/echogs. | 16:59 |
| tkamppeter | ahasenack, ginggs, slangasek, the failing echogs line is | 17:00 |
| tkamppeter | ./soobj/aux/echogs -e .dev -a- ./soobj/xpswrite -include ./soobj/vector -includ | 17:00 |
| tkamppeter | e ./soobj/libtiff | 17:00 |
| tkamppeter | It is always this one, which simply hangs indefinitely and gets killed after one and a half hour by the build server. | 17:01 |
| tkamppeter | There are tons of other echogs lines before which do not fail. | 17:02 |
| ginggs | tkamppeter: so what happens if echogs is built with -O2 ? | 17:04 |
| tkamppeter | ginggs, yes, this I could test via PPA. Need to find out now how to override the system-injected -O3 ... | 17:08 |
| Laney | tkamppeter: export DEB_CFLAGS_MAINT_APPEND = -O2 in debian/rules probably does it | 17:19 |
| tkamppeter | Laney, thanks. | 17:34 |
| tsimonq2 | Hah, I'm guessing the advice above can help with what I was just about to ask about... | 23:53 |
| tsimonq2 | https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=905868 | 23:53 |
| ubottu | Debian bug 905868 in gcc-8 "gcc-8: miscompiles vec_sl at -O3 with -fwrapv on ppc64el" [Normal,Open] | 23:53 |
| tsimonq2 | First time but probably not the last I'm going to have to set that flag... | 23:53 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!