14:00:14 <stickster> #startmeeting Workstation WG
14:00:16 <zodbot> Meeting started Mon Dec  4 14:00:14 2017 UTC.  The chair is stickster. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:16 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
14:00:16 <zodbot> The meeting name has been set to 'workstation_wg'
14:00:17 <stickster> #meetingname workstation
14:00:17 <zodbot> The meeting name has been set to 'workstation'
14:00:20 <stickster> #topic Roll call
14:00:22 <stickster> .hello pfrields
14:00:23 <zodbot> stickster: pfrields 'Paul W. Frields' <stickster@gmail.com>
14:00:31 <juhp_> .hello petersen
14:00:32 <zodbot> juhp_: petersen 'Jens Petersen' <petersen@redhat.com>
14:00:36 <kalev> .hello kalev
14:00:37 <zodbot> kalev: kalev 'Kalev Lember' <klember@redhat.com>
14:00:51 <stickster> Cchair juhp_ kalev ryanlerch
14:00:59 <ryanlerch> .hello ryanlerch
14:01:00 <zodbot> ryanlerch: ryanlerch 'Ryan Lerch' <rlerch@redhat.com>
14:01:36 <stickster> Hmm, missing one more for quorum (although not sure what decision items we have in front of us)
14:02:10 <stickster> Anyone seen otaylor, mclasen, cschalle, rdieter, catanzaro?
14:02:53 <kalev> mclasen is travelling, don't know about others
14:03:13 <stickster> Ah right, I recall him mentioning that last week somewhere
14:03:55 <stickster> #link https://pagure.io/fedora-workstation/issues?tags=meeting -- agenda
14:04:35 <stickster> So I see only two things on the agenda currently. One, regarding SELinux and removing the troubleshooter, is in uncertain status. I need to take some time to figure out where that went off rails and see what's to do next.
14:05:03 <stickster> The other is on langpacks, and there was an update before the last meeting that I think didn't happen. Since kalev and juhp_ are here, maybe we should see where we are?
14:05:15 <juhp_> okay
14:05:34 <stickster> sorry, that last bit was boolean "before (last meeting that didn't happen)"
14:05:43 <stickster> not that no one updated :-)
14:05:49 <kalev> I haven't looked at langpack stuff at all, not really sure what's missing
14:05:55 <mcatanzaro> .hello mcatanzaro
14:05:58 <zodbot> mcatanzaro: Sorry, but you don't exist
14:06:00 <mcatanzaro> .hello catanzaro
14:06:01 <zodbot> mcatanzaro: catanzaro 'Michael Catanzaro' <mcatanzaro@gnome.org>
14:06:04 <stickster> aha, now we have quorum
14:06:09 <stickster> #topic Langpacks
14:06:11 <stickster> #link https://pagure.io/fedora-workstation/issue/27
14:06:15 <stickster> #chair mcatanzaro
14:06:15 <zodbot> Current chairs: mcatanzaro stickster
14:06:22 <stickster> #chair juhp_ ryanlerch kalev
14:06:22 <zodbot> Current chairs: juhp_ kalev mcatanzaro ryanlerch stickster
14:06:53 <otaylor> .here otaylor
14:07:17 <stickster> #chair otaylor
14:07:17 <zodbot> Current chairs: juhp_ kalev mcatanzaro otaylor ryanlerch stickster
14:07:22 <stickster> ah look, they're filtering in now ;-)
14:08:42 <stickster> kalev: juhp_: I think the status of this item is, we are looking for whether gnome-initial-setup could detect language installation with missing langpacks and then call Software to add langpacks when needed, so we didn't have to ship them in the default image
14:08:53 <stickster> please correct me if I'm wrong
14:09:11 * kalev nods.
14:09:29 <stickster> There is the question of what this means for Live images
14:09:37 <stickster> Since those are used by people globally
14:10:12 <juhp_> kalev: do you have ideas about it?
14:10:32 <stickster> Here's the rub... whatever is in the Live image is basically what ends up on hard disk install -- AIUI the image content (dirs/files) is basically copied out to the new disk
14:10:50 <kalev> juhp_: not really, I haven't thought about this at all yet
14:10:57 <juhp_> okay
14:12:11 <stickster> My bet is there will be folks resistant to the (admittedly hackish) idea of tinkering with the content + rpmdb during the install-to-hard-disk process or post-process
14:12:33 <juhp_> hmm yeah
14:12:46 <stickster> And if the Live image doesn't have langpacks installed, how does that affect, for instance, a JP user who boots it
14:13:08 <juhp_> it's bad for libreoffice users
14:13:55 <stickster> right
14:16:01 <stickster> Here's maybe a crazy idea - would we want to consider Libreoffice not being in the Live image, but having some sort of "featured install" for it as a g-i-s module specific to Fedora?
14:16:15 <juhp_> but still having the installer do something would be more desktop agnostic
14:16:56 <juhp_> stickster: that is also an interesting idea but I imagine some people might sad if if wasn't on live
14:17:03 <stickster> *nod
14:18:01 <stickster> I guess we don't have any way of judging the rate of migration to cloud-based productivity software; certainly the Linux userbase tends toward local more so than the average consumer
14:18:06 <juhp_> another crazy idea, but libreoffice specific: libreoffice could pull in the langpacks somehow at first startup
14:18:20 <juhp_> true
14:18:21 <ryanlerch> stickster: one case for it being on the live image is people installing from local media, and not having good interent access
14:18:29 <stickster> juhp_: That's less crazy -- how would that operate in the Live environment, though?
14:18:31 <ryanlerch> but not sure how many of our users fit that case
14:18:31 <juhp_> right
14:18:45 <stickster> ryanlerch: very true -- OTOH getting langpacks after the fact also doesn't help those folks, right?
14:19:02 <stickster> well, maybe it's not so bad -- one langpack is not the same as "all langpacks"
14:19:42 <stickster> these are mostly less than 10 MB total
14:19:54 <stickster> I don't know how bad that bad internet can be
14:19:55 <juhp_> anyway failing the installer, next stop would installing via gnome-initial-setup
14:20:53 <stickster> So the idea would be here, g-i-s detects language, confirms the lack of langpacks, and then asks Software to install them
14:21:16 <stickster> (I think we're still avoiding the question of how this looks in the Live image, but we can defer that for the moment)
14:21:16 <juhp_> it could
14:21:51 <stickster> kalev: mcatanzaro: otaylor: let us know if you have questions/objections
14:22:15 <juhp_> and ideally gnome-control-center would also install langpacks when switching desktop language
14:23:15 <juhp_> but that is probably lower priority
14:23:28 <otaylor> Would want to here from kalev about feasibility of having gnome-software do that sort of one-off task before commenting. Would g-i-s wait for the installation?
14:23:38 <juhp_> yes
14:24:10 <juhp_> otaylor: good question
14:24:41 <juhp_> that's the thing, I feel uncomfortable about it just happening in the background since it can take time to complete
14:25:51 <juhp_> my hack proposal in the ticket was just to launch gnome-software on the appropriate langpack meta package page - better via a button click so it is not so surprising
14:26:08 <juhp_> but maybe there are better ways to do it
14:27:40 <otaylor> I think it's not OK to accumulate extra windows during extra setup, or to leave the user in gnome-software at the end of setup.
14:27:48 <otaylor> s/during extra setup/during initial setup/
14:28:01 <juhp_> okay
14:28:06 <juhp_> true
14:28:15 <ryanlerch> +1 to otaylor on that one IMHO
14:29:51 <juhp_> guess exception would be yelp at the end of initial setup
14:30:27 <mcatanzaro> Well g-i-s would need to show a progress bar and some meaningful additional information "Downloading language packs" or "Installing language packs"
14:31:12 <mcatanzaro> I think there is a minor technical problem here because our primary API for installing extra packages launches GNOME Software as a matter of course, but this should be possible. From both g-i-s and g-c-c.
14:32:04 <mcatanzaro> Finding a developer to do the work might be the hardest part.
14:32:36 <mcatanzaro> But my understanding is that PackageKit already understands langpacks on Fedora. If I'm wrong about that, then some design work will also be required with hughsie and kalev.
14:33:02 <juhp_> yeah I haven't tested that
14:33:17 <juhp_> but at least it works through gnome-software
14:33:27 <juhp_> so I assume it is okay
14:33:37 <stickster> oops, "assume" :-)
14:34:35 <juhp_> lol
14:36:24 <juhp_> But the UI sounds a bit complicated...
14:36:57 <stickster> Is this able to happen in a headless way via software/pkcon ?
14:37:54 <juhp_> All this fuss is actually just to install a meta package
14:38:38 <juhp_> Just wondering if there were so simpler way to handle it, but i can't think of one now since it needs to interact with weak rich dependencies in the langpacks
14:39:17 <juhp_> evil grin
14:40:34 <stickster> Yeah, I'm trying to envision what the user sees (if anything?) when this happens and that's what's not clear to me.
14:40:53 <mcatanzaro> stickster: In kalev's absence, I'll try to answer: I think the headless API is a lot more complicated than the API that pops up GNOME Software. It will need some dedicated developer effort. But I think all the requests should be doable.
14:40:55 <stickster> So what specific action do we need to happen next here?
14:41:04 <juhp_> stickster: well coming up with an extra screen in the setup is not hard, but not sure it is ideal either
14:41:18 <mcatanzaro> It should be done on the existing language selection screen IMO.
14:41:24 <juhp_> like what mcatanzaro described
14:41:32 <stickster> juhp_: right, IME folks have been resistant to "one more screen," and doing this via/during the existing screen is ideal
14:41:34 <juhp_> I see
14:41:42 <stickster> *jinx
14:41:58 <juhp_> not sure if there is room on the language screen
14:42:09 <juhp_> need to think a bit
14:42:23 <juhp_> ditto for control-center
14:42:53 <mcatanzaro> The designers will find room... or maybe design a new screen... but I imagine all the languages disappearing once one is selected, and a progress bar appearing. And the heading on the page staying the same. (In g-i-s. Not sure about g-c-c.)
14:43:12 <juhp_> yes that sounds nice
14:43:26 <juhp_> also I feel it should be opt in
14:43:33 <juhp_> [Install langpacks]
14:43:59 <juhp_> or one could [Skip]
14:44:16 <juhp_> takes a while even to download the repodata
14:44:28 <stickster> *nod
14:45:25 <otaylor> If it's only 10mb, I don't see optional as being interesting. But is language selection before or after network?
14:45:43 <juhp_> If libreoffice could install it's own langpacks then we might not need this feature so much
14:45:57 <stickster> otaylor: the point isn't necessarily the metapackage size. It's getting all the repodata from Fedora in general, that's pretty sizable right?
14:45:57 <juhp_> otaylor: per language
14:46:29 <otaylor> stickster: yuck.
14:47:16 <otaylor> But, I don't think that changes the desirability of an opt-out button, just makes it a question whether making it a blocking step in g-i-s is a good idea.
14:47:54 <otaylor> Maybe just entirely in the background possibly with a notification
14:48:37 <juhp_> yeah if we do it inline it will increase the time in g-i-s quite a bit
14:49:09 <mcatanzaro> otaylor: Good point... language selection is before network.
14:49:15 <mcatanzaro> Has to be, or you won't be able to read the network page ;)
14:49:33 <mcatanzaro> And keyboard layout has to come before anything that requires typing. So maybe langpack installation needs to come last.
14:49:44 <juhp_> that's why I thought leveraging gnome-software also would avoid that
14:49:52 <juhp_> now we are in anaconda?
14:50:05 <mcatanzaro> I'm still talking about g-i-s
14:50:21 <mcatanzaro> I'm pretty sure we don't want gnome-software popping up during initial setup.
14:50:26 <stickster> OK, this sounds like a discussion that needs to continue after this meeting, and as you guys know I have a hard stop coming up for a meatspace meeting I have to moderate :-) -- can we move this to #fedora-workstation?
14:50:27 <juhp_> g-i-s has networking?
14:50:55 <juhp_> mcatanzaro: right
14:51:09 <mcatanzaro> juhp_: It has a network page for connecting to a Wi-Fi network.
14:51:19 <juhp_> okay I see - I forgot that
14:51:23 <stickster> I would like to #action juhp_ to drive this discussion with otaylor, kalev, and any other interested/concerned folks to figure out the right place to seek/inject devel work
14:51:34 <juhp_> okay sure
14:51:45 <juhp_> I will try to continue the discussion
14:52:00 <stickster> This has definitely been a worthwhile conversation to get unstuck, it's just that we're out of time here :-)
14:52:13 <stickster> appreciate juhp_ being here to help :-)
14:52:17 <juhp_> yes it is helpful to get good advice
14:52:39 <juhp_> thanks
14:53:22 <juhp_> mcatanzaro: I think I am usually not on wifi when running g-i-s...
14:56:33 <mcatanzaro> juhp_: langpack installation will fail if run before the wi-fi screen, unless the user is plugged in with Ethernet
14:56:44 <juhp_> mcatanzaro: indeed
14:57:23 <juhp_> Just saying that's why I was not aware of the screen
14:57:47 <stickster> OK, need to close out here folks
14:57:52 <mcatanzaro> juhp_: It doesn't appear at all unless you have a laptop (or a wi-fi card).
14:57:57 <mcatanzaro> Bye everyone!
14:57:58 <juhp_> nod
14:58:02 <stickster> Let's head to #fedora-workstation for more discussion
14:58:11 <stickster> Thanks for coming, everyone!
14:58:38 <stickster> #info LONG discussion on technical choices/avenues, needs more winnowing down to a process that will work in the UI with given tools
14:59:09 <stickster> #action juhp_ drive additional conversation in #fedora-workstation and on list to figure out where devel effort would really be needed
14:59:12 <stickster> #endmeeting