18:01:35 <smooge> #startmeeting EPEL 18:01:35 <zodbot> Meeting started Wed Sep 13 18:01:35 2017 UTC. The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:35 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:01:35 <zodbot> The meeting name has been set to 'epel' 18:01:35 <centbot> Meeting started Wed Sep 13 18:01:35 2017 UTC. The chair is smooge. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:35 <centbot> Useful Commands: #action #agreed #help #info #idea #link #topic. 18:02:16 <smooge> #meetingname EPEL 18:02:16 <zodbot> The meeting name has been set to 'epel' 18:02:16 <centbot> The meeting name has been set to 'epel' 18:02:16 <smooge> #topic aloha 18:02:16 <smooge> #chair avij bstinson Evolution nirik smooge 18:02:16 <zodbot> Current chairs: Evolution avij bstinson nirik smooge 18:02:16 <centbot> Current chairs: Evolution avij bstinson nirik smooge 18:02:21 <smooge> we will do it here 18:02:33 <avij> we'll do it live 18:02:40 <smooge> <snicker> 18:02:51 <bstinson> heh 18:02:58 <nirik> and in production! 18:03:28 <smooge> #topic Meeting TIme 18:03:53 <smooge> ok wanted to get an idea of who can make meetings here and if this timeslot is still good 18:04:06 <smooge> also where I am supposed to update the meeting calendar :)? 18:04:17 <nirik> https://apps.fedoraproject.org/calendar/list/epel/ 18:04:24 <nirik> this time is still fine with me 18:05:25 <bstinson> works for me too 18:05:34 <avij> this is an appropriate meeting time for me. it's evening (9pm), but that suits my schedules better than normal "office hour" times. 18:08:25 <smooge> ok I am putting in. OK for me to 'end' the series at the end of this year or make it longer? 18:09:40 <smooge> ok thanks for the info. I will finish up and start inviting after this meeting 18:10:04 <smooge> #topic Announcements 18:10:04 <smooge> #info CentOS-7.4 issues 18:10:04 <smooge> #info ??? 18:10:36 <avij> as for announcements.. CentOS 7.4.1708 will be released today. 18:10:40 <smooge> So we have still been having CentOS CR issues reported but I think it is ending soon. 18:10:48 <smooge> avij, thanks for the info on that 18:11:06 <nirik> cool. :) 18:11:25 <smooge> #info CentOS 7.4.1708 will be released today. 18:11:50 <smooge> #info FLOCK last week. Lots of talks. My brain is full syndrome on catching up with all the items 18:11:51 <avij> hopefully this will make the life of everyone a bit easier 18:12:31 <smooge> yeah.. now it will just be the people who think that EPEL somehow provides a link to older releases we see every minor release 18:13:21 <smooge> bstinson, nirik was there anything for announcements that you knew of? 18:13:45 <nirik> not off hand 18:13:58 <bstinson> nothing from me 18:14:53 <avij> nothing apart from the above 18:15:57 <smooge> okie dokie 18:16:14 <smooge> #topic FLOCK 2017 recap 18:16:48 <smooge> So there were 2 major meetings and a lot of hallway meetings. I have been trying to get all of them written down in a blog post but have had a bunch of irqs that ate up that time 18:17:15 <smooge> I am putting it down as a deliverable for next weeks meeting that I have the writeups done to a blog and to the list. 18:17:50 <smooge> Major items that looked 'interesting' was that modularity is going to make things very different in the future for EL 18:18:25 * nirik nods 18:19:18 <avij> yes, I can imagine.. 18:20:53 <smooge> Next we talked about things we could try and do in the coming year.. which will hopefully be easier with the additional branching in Fedora sources 18:21:50 <smooge> that would be to allow branches for specific releases and to move EPEL into a rawhide type channel built against CentOS and a EL release cycle built against RHEL. 18:22:05 <smooge> Again I need to write down the particulars and will do so before meeting 18:22:21 <smooge> Was there anything from those who could make it? 18:23:06 <nirik> there was your opt parallel installable packages idea... 18:23:30 <nirik> which I am not sure I followed all the details of. I think I was fighting fires or lack of sleep about then. ;) 18:24:12 <smooge> both 18:25:03 <smooge> you had to step out. Basically I am hoping to get /opt/epel approved so we can make rules for putting packages which 'collide' with RHEL package filesystem space 18:26:33 <smooge> as in /opt/epel/bin/httpd could be in a httpd24 rpm on RHEL6 so that a packager does not need to make sure the binary and all its sub-items are named so as it could be /usr/bin/httpd24 18:26:44 <smooge> or python 18:28:31 <smooge> again it needs some policy written and more than 2 lines in irc for why and how it is supposed to work 18:28:49 <nirik> yeah, I wonder if it's worth the trouble now... but probibly 18:29:08 <avij> an interesting concept, but I'm not sure I know all the implications of that proposal 18:29:38 * carlwgeorge pops in 18:29:53 <smooge> since EPEL-6 is our largest and still growing user space I was figuring it made sense for the next 2 years and similar to EL- 18:30:08 <smooge> avij, understood. 18:30:24 <carlwgeorge> smooge: i remember you mentioned the goal was to provide one newer version, not multiple. is that still the case? 18:31:34 <smooge> yes this is for single LTS streams which people want to focus on versus the other /opt/ thing which my brain cant remember the term for 18:31:57 <bstinson> the idea is not to replicate SCLs 18:32:02 <smooge> thanks 18:32:06 <smooge> god I am old today 18:32:45 <carlwgeorge> would python stick with the current python3_pkgversion stuff or move to this new competing space? 18:33:19 <nirik> IMHO it should just stick with the current thing... 18:33:26 <nirik> since there's a bunch of packages now that depend on it 18:33:32 <smooge> it is a depends thing. Currently (and far as I know) python3 is not shipping with RHEL7 18:33:48 <smooge> so we aren't running 2 different versions on the same system. 18:33:51 <carlwgeorge> ah, so it doesn't fit the competing definition 18:34:18 <carlwgeorge> hmm, well it competes with the stock python2... 18:34:33 <smooge> on EL-6 we have python2.6 as the /usr/{bin,lib,share} and people want python2.7 because stuff doesnt work in 2.6 18:34:41 <Sketch> smooge: correct, though it is in epel. 18:35:33 <nirik> people always want what they don't have. ;) 18:35:37 <Sketch> and SCL 18:35:46 <carlwgeorge> either way, i would be fine if the competing stuff is limited to things that do not easily cooperate with parallel installs. python works great with it's make altinstall. 18:35:48 <Sketch> (for 7. not sure about 6.) 18:36:29 <smooge> I am using python as it was the one I remember being a pain in the ass for EPEL-5 18:37:05 <smooge> but fill in the bucket for the software which works better 18:37:36 <carlwgeorge> back then python may not have had altinstall in their makefile yet 18:38:16 <carlwgeorge> what about naming scheme? would using names like httpd24 cause too much confusion with similarly named packages in epel like python34? 18:40:36 <smooge> I don't understand why people would be confused. I chose the name so it would be similar to how we do python34 and such. The /opt/epel/ choice was mainly that it is a path namespace which we could have which would not collide with current ones 18:40:56 <smooge> it is basically an /usr/local for the project 18:41:37 <smooge> however I am also likely blind to something 18:42:08 <carlwgeorge> my thinking is that users install python34 and expect python3.4 or similar in their path, but if they install httpd24 they will open issues about httpd2.4 or similar not being in their path. 18:44:17 <smooge> ah ok 18:44:48 <smooge> thanks 18:44:54 <smooge> yeah that does make sense 18:45:08 <carlwgeorge> with scl it's a bit more explicit with their long names (rh-<collectionname>-<packagename>), so the expectation is different i think 18:46:18 <bstinson> maybe that's something we can highlight when we create the proposal 18:46:21 <carlwgeorge> i'd like to say i have an ideal name scheme to suggest, but naming things is hard. how about competing-<packagename>? i.e. competing-httpd, competing-redis, etc. 18:47:15 <smooge> ok will do so 18:47:17 <carlwgeorge> that would also solve the problem of future updates invalidating a versioned name, i.e. competing-httpd could update from 2.4 to 2.6 18:47:34 <nirik> but then we get back to when would that be allowed/make sense... 18:47:53 <nirik> vs just stopping maintaining something. :) 18:48:36 <Evolution> sorry I'm late. 18:48:38 <carlwgeorge> iirc smooge wanted these competing packages to roll with the latest upstream lts 18:48:47 <smooge> my proposal was that we only maintain things which are either upstream LTS or supported by RHEL 18:48:57 <smooge> so lets take the case of the mythical EL-next 18:49:11 <smooge> let us say it comes with python-3.9 18:49:30 <smooge> (that isn't a real version on purpose) 18:51:11 <smooge> then we could have a python39 on the os's which don't have it. If for some reason EL-7.5 came with python-3.pi (again made up) we could have one python39 in a noncoliding space with the python3 18:51:20 <smooge> the more likely space for this is tetex 18:51:30 <smooge> but I didn't want to drive you all to madness 18:53:11 <carlwgeorge> ah, so more of "backport the latest EL spec file to older EL's", versus "backport the latest Fedora spec file to EL's" 18:53:18 <nirik> ok, I see... probibly will be more clear with a written up proposal/examples. ;) 18:53:28 <smooge> yes. it hopefully will 18:53:46 <carlwgeorge> sorry to derail with so many questions 18:53:50 <smooge> no it was needed 18:54:22 <carlwgeorge> several of these have been on my mind since flock tbh 18:54:59 <smooge> I will try to get this written up after a) fix nagiso pacakging b) get download-ib moved c) find a big pot of coffee 18:55:04 <nirik> questions are good. :) helped clarify it to me... 18:56:38 <carlwgeorge> so if name contains the version, in the httpd example, would httpd26 obsolete httpd24? or is it more likely to just lock onto upstream's major verison number? 18:58:27 <smooge> I would say from the perspective of what I was writing about it would obsolete 18:59:54 <smooge> I am trying to make it simple for packagers 19:00:07 <smooge> who mostly want to fire and forget 19:00:20 * nirik nods 19:00:41 <smooge> I would like to make it so it is quite well laid out so that a normal spec file can be made easily to fit into the other name space without a lot of headaches 19:01:14 <smooge> supposedly it should be easy. If it isn't then this dies a fast death 19:01:23 <tibbs> It appears that I've missed a very interesting meeting. 19:01:36 <carlwgeorge> would a convenience symlink in the PATH be allowed? something like /usr/bin/httpd24 -> /opt/epel/bin/httpd? 19:02:14 <smooge> I would think so as long as that did not break the app 19:02:41 <smooge> aka if /usr/bin/httpd24 crashes because it needed to be run from /opt/epel/bin/httpd 19:03:18 <smooge> The last part of the proposal would be to have a epel-opt.sh script which added /opt/epel/bin ot PAT and /opt/epel/lib to LIB 19:03:27 <carlwgeorge> fair enough 19:04:00 <smooge> so anyway.. time to get to writing that. 19:04:56 <smooge> thanks for the qustions 19:05:10 <smooge> so the next agenda item 19:05:17 <smooge> #topic Next Meeting Agenda 19:05:51 <smooge> so I need help here.. are there any tickets or items we need for this? If something comes up in the next week please email/ping me also 19:06:49 <nirik> I had 1 thing (I guess it could be open floor).... we had scheduled a work day, but I and I think everyone else completely forgot... so we should rescheule one sometime. 19:07:22 <smooge> yeah it was last friday 19:07:33 <smooge> my apologies on that 19:08:07 <nirik> mine too. I saw the reminder, but was swamped. 19:09:20 <smooge> so yes. we need an activity day and what to do on it. 19:09:31 <smooge> Next week we will go over what items need to be worked on 19:09:39 <smooge> Then advertise and do it 19:09:46 <smooge> And maybe cookies 19:10:02 <nirik> punch and pie? 19:10:17 <smooge> mmmm 19:10:20 <smooge> pie 19:10:33 <smooge> so ok I will have that for agenda next meeting 19:10:49 <smooge> I think we are running into other peoples meetings so 19:10:53 <smooge> #topic Open Floor 19:11:01 <smooge> I can close out in 2 minutes 19:11:04 <smooge> or less 19:11:53 <smooge> Shoot 19:11:57 <smooge> I just remembered something 19:12:51 <smooge> https://lists.fedoraproject.org/archives/list/epel-announce@lists.fedoraproject.org/message/PBEYGLJXWY33ANB5XTKHIJVXEIYW2PSI/ 19:13:38 <smooge> We have a bunch of packages which were in EPEL but aren't. tyll made a list 19:14:01 <smooge> We need to work out what to do with this and get it done. I would like some ideas on the list 19:14:19 <smooge> #info need ideas to deal with retired packages due to pkgs change 19:15:09 <smooge> ok with that 19:15:14 <smooge> #endmeeting