21:00:02 #startmeeting Rust SIG (2017-03-08) 21:00:02 Meeting started Wed Mar 8 21:00:02 2017 UTC. The chair is ignatenkobrain. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:02 Useful Commands: #action #agreed #halp #info #idea #link #topic. 21:00:02 The meeting name has been set to 'rust_sig_(2017-03-08)' 21:00:07 #meetingname rust-sig 21:00:07 The meeting name has been set to 'rust-sig' 21:00:11 #chair ignatenkobrain jistone 21:00:11 Current chairs: ignatenkobrain jistone 21:00:14 #topic Agenda 21:00:20 #link https://docs.pagure.org/fedora-rust.sig/meetings/2017-03-08.html 21:00:23 #info (1) Roll Call 21:00:26 #info (2) Packaging Guidelines 21:00:28 #info (3) Open Floor 21:00:31 #topic Roll Call 21:00:33 .hello ignatenkobrain 21:00:34 ignatenkobrain: ignatenkobrain 'Igor Gnatenko' 21:00:35 .hello 21:00:37 jistone: (hello ) -- Alias for "hellomynameis $1". 21:00:40 .hello jistone 21:00:41 jistone: jistone 'Josh Stone' 21:00:51 not so much stuff to discuss today 21:00:54 .hello ngompa 21:00:58 Pharaoh_Atem: ngompa 'Neal Gompa' 21:01:16 so should be quick ;) 21:01:21 #topic Packaging Guidelines 21:01:28 #link https://pagure.io/fedora-rust/sig/pull-request/9 21:02:43 comments are welcomed 21:05:24 (for the meeting logs -- comments are happening on the issue) 21:06:59 ignatenkobrain: threw in a bunch of comments about this 21:07:10 so seems like naming is main issue 21:07:14 yeah 21:08:19 having rust-* for all source packages ensures we don't have accidental collisions with non-Rust things 21:08:31 perhaps the use of rich deps needs to be discussed too 21:08:50 technically, this is a rule for golang, Python, rubygems, and other similar things 21:09:03 however, we *could* allow on a case-by-case basis to not require it being rust-* 21:09:11 rich deps are cool, but limiting where we can deploy this 21:09:13 so maybe it should be "strongly preferred" 21:09:46 and the generators should do it always, and if the packager wants to change it... well, they'll have to justify in the package review 21:11:07 I'm fine with preferring rust-%{crate} 21:11:53 so then rust-ripgrep and Provides: ripgrep ? 21:12:03 for the srpm that is, then rust-%{crate}-devel and/or %{crate} as appropriate 21:12:06 yep 21:12:15 ignatenkobrain: what jistone said 21:12:30 don't have to have a rust-ripgrep.arch.rpm at all 21:12:33 note that you can't have main package name as %{crate} 21:12:53 because primary package name should be same as srpm name 21:13:03 %package -n %{crate} as a subrpm 21:13:15 if the primary package has no %files section, it doesn't get produced 21:13:27 that's the trick for decoupling 21:13:38 yeah, we can do that 21:14:39 #agreed all packages should be prefixed with "rust-", even binary ones 21:15:00 #undo 21:15:00 Removing item from minutes: AGREED by ignatenkobrain at 21:14:39 : all packages should be prefixed with "rust-", even binary ones 21:15:13 all source packages should be prefixed with "rust-" 21:15:29 #agreed all source packages should be prefixed with "rust-", even which produce packages with binaries (applications) 21:15:32 yeah, poor wording 21:15:36 much better 21:15:52 as for rich deps, I'm strongly for leaving them in, but if we want to "bend" it for EPEL or earlier Fedora (assuming Fedora doesn't get rich dep support backported), we *could* 21:16:19 the generator could have a "legacy" mode to resolve them to their actual crate packages with the rich dep version commented out 21:16:41 for now I would stick to rich deps, but in couple of weeks we will see how it goes with rele g 21:16:43 releng* 21:16:52 I'm fine with that 21:17:11 and that's f27 at the earliest 21:17:18 right? 21:17:20 and Mageia 7 21:17:34 yup 21:18:13 so the alternate thing I offered to ignatenkobrain was to encode semver in the Provides 21:18:15 Provides: crate(num-0.1) = 0.1.37 21:18:24 Requires: crate(num-0.1) > 0.1.20 21:18:51 yeah, this is the backup 21:18:53 if we want to do that, I would rather that be a *very* short term thing 21:18:57 this doesn't allow bigger ranges, >0.1,<0.4 21:19:24 Pharaoh_Atem, this would be for a "legacy" mode as you mentioned 21:19:24 it assumes the range is within the same semver block 21:19:31 right 21:19:31 yes 21:20:02 *very* short term --> until epel7 goes away 21:20:03 :P 21:20:12 jistone: that's not even "short" 21:20:22 it's basically more than 10years 21:20:24 :D 21:20:24 that's just making our lives miserable :( 21:20:30 (that's the joke) 21:20:46 well, Mageia 6 expires 3 months after Mageia 7 releases 21:20:56 hopefully it won't take a year to get there :/ 21:21:45 jistone: I wonder if it might just be cleaner to not do alternate provides and just do package spec generation time resolution 21:22:05 so it'd look like this: 21:22:09 Pharaoh_Atem: then you will need to rebuild it every time 21:22:23 ignatenkobrain: that's not really going to matter for EPEL7 21:23:17 EL7 is sufficiently horrible anyway 21:23:37 update policy for epel is hard 21:23:50 at this point I'm very interested what releng will say about richdeps 21:23:54 but this will take a bit of time 21:24:07 they've got use-cases for php, rust and ruby 21:24:13 well, dgilmore already told me the earliest he can review the pungi PR is next week 21:24:21 link? 21:24:34 and actually there are more use-cases for SCLs and Modularity 21:24:42 Pharaoh_Atem: porting pungi is not enough 21:24:42 https://pagure.io/pungi/pull-request/526 21:24:47 apparently 21:24:58 no, we need koji ported too 21:25:06 and bodhi needs to move to using koji rather than mash 21:25:22 he said that much more stuff will need to be touched 21:25:24 I stopped my work on mash once koji signed repos was rebased and noted as complete 21:25:27 so... quite a lot then 21:25:41 well, the problem is that not many people are motivated to fix this besides ignatenkobrain and myself 21:25:41 :-P 21:26:07 Pharaoh_Atem: maxamillion is interested as well 21:26:39 if we can get together for a weekend or something, maybe we can just port Koji :P 21:27:08 let's decide about this stuff on next meeting 21:27:27 rpm merged the fix upstream today 21:27:42 libdnf and dnf-plugins-core still need their changes merged in 21:27:48 then the whole feature needs to be backported 21:27:53 it's not enough 21:27:59 "with" deps are not implemented yet 21:28:08 libdnf still hangs on me, should do more fixes there 21:28:14 hmm 21:28:16 dnf-plugins-core... need to fix tests 21:29:18 for RPM I've merged support for depgens which emit rich deps 21:30:13 okay, naming discussed. version stuff has been moved for next week... any major thngs to discuss about packaging? 21:30:55 Hi all. Bit late due to classes. 21:31:10 hi luke_nukem 21:31:50 #topic Open Floor 21:32:23 I didn't do much last week 21:32:29 luke_nukem: did you have any progress on rust + opensuse? 21:32:35 although keeping crates up2date 21:32:41 today updated rustfmt to 0.8.0 21:32:53 most of patches were merged in upstream 21:32:53 I've got rustbuild going in dist-git jistone/beta for 1.16 21:33:06 jistone: have copr repo? 21:33:21 https://copr.fedorainfracloud.org/coprs/jistone/rust-beta/ 21:33:28 perhaps I should do those in the sig group 21:33:38 yeah 21:33:38 Pharaoh_Atem: Haven't had much time as of yet. Just in 2nd week of University and getting dates/schedules sorted now. I'll be making time to get things rolling in SUSE this weekend. 21:33:42 ignatenkobrain: oh, did you speak with copr guys about double root mock? 21:33:55 oh right 21:34:06 they deployed latest mock for me 21:34:10 on dev instance 21:34:14 so we can try stuff there 21:34:23 where's dev copr? 21:34:26 that was my plan for today, but -ENOTIME 21:34:39 ain't that the truth :) 21:34:50 jistone: How is rustbuild for you so far? I didn't have any issues when I tried using it in packaging a few weeks ago. 21:35:12 copr-fe-dev.cloud.fedoraproject.org 21:35:36 luke_nukem, not too bad, just makes the bootstrap srpm even more ridiculously huge 21:35:51 jistone: how big is it now? 21:36:15 hmm, I think I said ~700MB or so 21:37:11 it goes from ~40MB tarballed binaries per target to ~100MB 21:37:15 wow 21:37:25 jistone: How do you do bootstrap currently? I have the openSUSE one split out from the main rust package 21:37:31 it's reaching texlive sizes 21:37:55 Pharaoh_Atem: cm'on, spec file is how many MiBs? 21:37:56 ;) 21:37:57 luke_nukem, all in one srpm, with a %bootstrap_arches control variable 21:38:18 8.4M texlive.spec 21:38:20 ignatenkobrain: okay, that one is a bit extreme 21:38:27 16K rust.spec 21:38:30 not even close :P 21:38:31 luke_nukem, we never ship a bootstrapped binary build to users 21:38:35 I can't even open that spec file in most text editors 21:38:54 luke_nukem, so I set %bootstrap_arches, do a build, then turn it off and rebuild 21:39:02 luke_nukem: I'm surprised that OBS doesn't have bootstrap control flag support 21:39:03 jistone: the bootstrap for openSUSE is only used in OBS ;) 21:39:20 and I'm guessing you disable publishing of the bootstrap package :) 21:39:27 Pharaoh_Atem: it does in the form of flags that use a %bcond 21:39:45 Pharaoh_Atem: yes disabled. 21:41:19 oh, I also rebuilt rust in f25-llvm39 for the upcoming llvm rebase 21:41:33 well, now that Factory is on rpm 4.13, you can probably fork rpm, libsolv, and libzypp to add support for "with" richdep 21:42:23 Pharaoh_Atem: I'll have to get on that pretty soon I think. This weekend anyway. 21:42:54 let me know if you need any help :) 21:43:07 I think you might want to talk to adrianS and mls before you start this weekend though 21:43:40 jistone: are you rebootstrapping on each release or something? 21:43:43 (adrianS is Adrian Schroeter, who manages OBS, and mls is Michael Schroeder, who maintains perl-BSSolv, libsolv, rpm, etc.) 21:43:50 mich181189, no, but I make sure we *can* 21:44:09 jistone: ah ok makes sense. Good thing to test I guess 21:44:11 Pharaoh_Atem: thanks, I'll get in touch with them soon. 21:44:14 luke_nukem: I suggest hanging out in #rpm-ecosystem and #opensuse-buildservice 21:44:42 Already in build-service. But I'll pop in to rpm-aecosystem too 21:44:53 emailing them specifically may help, too... adrian@suse.de and mls@suse.de 21:45:00 be sure to CC myself and ignatenkobrain to those emails 21:45:18 your emails are? 21:45:23 mine is ngompa13@gmail.com 21:45:44 I dunno which one ignatenkobrain would want you to use 21:45:58 my @fp.o 21:46:05 ignatenkobrain@fedoraproject.org 21:46:08 :P 21:46:24 Thanking you! 21:46:56 I was considering proposing this all as a Google Summer of Code project... But, not sure of the chances of success. 21:47:18 hey, you can 21:47:39 either under the Fedora org or the openSUSE organization (if openSUSE has one) 21:47:48 I think we're still taking GSoC proposals in Fedora 21:48:06 https://fedoraproject.org/wiki/GSOC_2017 21:48:34 I've nearly finished my Gnome proposal, maybe I could try submitting two in-case? I don't think I saw any restriction there.. 21:50:10 SUSE looks a bit more restrictive on their gsoc stuff: http://101.opensuse.org/gsoc/ 21:50:22 errm openSUSE 21:52:08 luke_nukem: if you want to propose a project to do under the Fedora banner, ignatenkobrain might be willing to mentor you 21:52:25 since he works on the DNF team, he intersects with a lot of different, relevant groups 21:53:01 Pharaoh_Atem: Yeah not so sure I want to try with openSUSE. I like working with the team I know of so far, but trying to do the rust stuff will be hard as there is really only me working on it for them - no mentors. 21:53:31 ignatenkobrain would be an excellent mentor :) 21:53:53 Pharaoh_Atem: Okay, might be worth a try then. I'll see if I can get a proposal done this weekend. 21:54:15 and you can probably squeeze in some libzypp/zypper love too :) 21:54:25 Lol! 21:54:44 well, at some point soonish, I'll get zypper into Fedora, so it'll be there and "relevant"-ish 21:54:53 So toss up between gnome-polari, and rusting the fedora. 21:55:08 well, corroding fedora sounds more fun, doesn't it? :) 21:56:30 it... does... But I've also got a great project lined up for Polari too - Tracker integration for power-searching of chat logs. 21:56:30 that might actually be the first time I've heard "rusting" said in a positive sense! (well, positive to those here anyway!) 21:56:42 mich181189: :D 21:57:22 mich181189: Hey, rat-rodders (hot-rodders) like rust too. It's decorative. ;) 21:58:00 note that we have only 3 more minutes =) 21:58:12 ignatenkobrain: Pharaoh_Atem: What's a quick run down of what still needs to be done for rust in fedora? As I understand it, it's mostly rpm issues? 21:58:32 luke_nukem: mostly organizational and releng things 21:59:00 probably I will be able to re-bootstrap cargo using only system deps at some point soon :D 21:59:18 there are only 4-5 deps are missing 21:59:18 some porting work for koji, adding new richdep support to rpm, libsolv, libdnf, dnf, etc. 21:59:27 ignatenkobrain: Actually, could you do a draft of the relevant things with links? I'll need to refresh myself for the work + GSoC proposal. 21:59:40 luke_nukem: will do 21:59:46 Thanks! 22:00:04 anything else? 22:00:09 I'm good 22:00:19 Good here too. 22:00:30 jistone: mich181189: ^ 22:00:39 good 22:00:44 cool! 22:00:44 #endmeeting