[00:29] Pharaoh_Atem, yes, that was what I was thinking :( [00:51] robert_ancell: I have a rediffed version of the patch (http://copr-dist-git.fedorainfracloud.org/cgit/ngompa/Mir/libinput.git/tree/libinput-1.5.0-touch-point-orientation.patch), but there's literally no way I can ever *hope* to consider getting Mir in Fedora proper while patches are required in libinput and mesa [00:53] robert_ancell: though I suspect the Mesa people aren't particularly happy with the idea that Mir turns Mesa GPLv3 [00:53] which might explain why the platform support patch hasn't made it in [00:54] Pharaoh_Atem, what's the GPLv3 issue? [00:54] the community doesn't particularly care for copyleft licenses at all [00:54] Wayland is licensed under the same terms as Mesa and Xorg (X11 Expat license) [00:55] they're more in the "copyfree" (or as I call it, "pushover") camp [00:56] robert_ancell: that particular community is made up of Linux, BSD, Windows, and various other platform people [00:57] with the exception of Linux, all of them prefer permissive licensing [00:57] Pharaoh_Atem, but how is Mir affecting the Mesa license? [00:57] in order for Mir to function, Mesa must have a platform module built into it [00:57] which links the Mir libraries [00:58] thus, Mesa, when compiled with Mir support, goes from X11/MIT to GPLv3 [01:00] Pharaoh_Atem, the Mir client library is LGPL, so that shouldn't be an issue [01:00] No; it goes to LGPLv3 [01:01] Or, rather, it remains X11/MIT as long as you don't statically link libmirclient in. [01:01] ah, so the mesa client libs are LGPLv3 then? [01:01] libmirclient is LGPLv3. [01:02] Anything you need for writing clients (or drivers) is LGPL; if you want to write a server, those libraries are GPL. [01:03] hmm [01:03] The reason why the platform support patch hasn't made it in is that we need to blow it up in the immediate future. [01:03] blow it up? [01:03] Change the API. [01:04] Change the types you hand to EGL. [01:04] is that happening soon? [01:04] Yes. [01:05] The Mir plumbing required to do that has either landed or is in a reveiewable MP, I think. [01:06] The final step is to publish that API and migrate the platform patch. [01:06] (And, obviously, to migrate all the downstreams. Yay!) [01:07] well, the downstreams aren't exactly plentiful, so that's not nearly as big of an issue [01:07] Right. *Particularly* when the patch isn't upstream. [01:07] * RAOF is also somewhat glad he never quite got around to implementing EGL_KHR_platform_mir support, so we can update that spec before it's released. [01:09] Incidentally, thanks for doing the Fedora packaging work. [01:10] unfortunately, it's all broken [01:10] Mesa in Fedora is now at v13 [01:10] and that patch is not trivial to rebase [01:10] RAOF: it seems like *technically* it's possible to use Mir without the mesa patch [01:11] so the *bare minimum* for me being willing to put Mir through review in Fedora would be to get the libinput patch upstreamed [01:11] as it won't compile without it [01:11] It is. You can't run GL clients, but you can run GL shells and software-rendered clients. [01:11] eugh [01:12] not a great experience then [01:12] Yes. [01:12] EGL unavoidaby requires platform-integration for different display servers. [01:13] (At least until someone proposes and implements a vendor-neutralish EGL with platform integration bits, like Android). [01:13] well, that would require the big Linux companies and the big GL consumers to come together to agree on something [01:14] we can't have that! [01:15] Heh. glvnd now exists, vulkan has a mostly-sensible ICD system. [01:15] Things are getting better :) [01:15] glvnd *barely* made it into Fedora 25 [01:15] so we have that fully enabled [01:16] well, technically, it made it to Fedora 23+ [01:16] but we're starting to use it with Fedora 25 [01:16] And in another 10 years or so maybe we can remove the previous Linux OpenGL ABI! [01:16] yeah, no [01:16] you're asking for way too much [01:17] even Apple and Microsoft haven't done that [01:18] Yeah, I know it's not going to happen. [01:19] my main interest in Mir has been to see if I can accomplish what many of my predecessors say is impossible: get Unity working properly on Fedora, without downgrading half the world [01:19] and since Unity 7 is on life support, I've looked toward Unity 8 [01:20] Unity8 should be helpful there, as it mostly doesn't rely on GNOME patches. [01:20] (And Qt is friendlier to downstream platform integration) [01:21] yes [01:22] it helps that since I'm more of a KDE guy myself, I'm more familiar with that stack [01:22] Although I also think that much of our Qt stuff is upstream. [01:23] Although not qtmir, because it's a fast moving target. [01:23] it is, but it's disabled in Fedora since we don't have the scaffolding for it [01:23] i.e. Mir itself [01:23] and various unity libraries [01:23] Heh [01:23] well, we recently brought back libunity and libappindicator, since Plasma 5 now uses them [01:24] I've never been happy with how those projects are managed, though [01:24] we're forced to yank tarballs from the ubuntu archive because the lp projects are dead wastelands with no updated releases ever made [01:25] Embrace the rolling release + CI pipeline! [01:25] It Is Life [01:26] well, that would require things like mir's gmock tests to actually work [01:27] (Or, more seriously, is hostile to downstreams) [01:27] right [01:27] I've always had a bit of lethargy for working with Launchpad-hosted projects, because they tend to do this to me [01:28] irony of ironies, my very first package in Fedora is of a piece of software hosted on Launchpad [01:28] There was a time when our projects would do actual releases, but not now. Mir is an outlier in that we *do* do releases. [01:28] (Because we don't have a stable C++ ABI, so updating Mir requires rebuilding the world, so we don't do that every commit ☺) [01:29] On the plus side, you *should* be able to take a random trunk commit and have it work properly. [01:29] nope [01:29] I won't trust that until I can run gmock tests [01:30] What's broken there? We don't patch googletest. [01:30] unfortunately, for some reason I cannot fathom, Mir's CMake script can't detect gmock and gtest [01:30] Does fedora just lack the latest gtest? [01:30] Huh. Maybe you install it in an odd place? [01:30] https://apps.fedoraproject.org/packages/gtest [01:31] https://apps.fedoraproject.org/packages/gmock [01:31] Alternatively, maybe you don't install the sources and expect projects to link to prebuilt libraries? [01:31] for gmock, it's sources [01:31] for gtest, it's libraries [01:31] https://apps.fedoraproject.org/packages/gmock/sources/spec/ [01:32] https://apps.fedoraproject.org/packages/gtest/sources/spec/ [01:32] Well, we have googletest 1.8.0 in zesty. But we don't require it. [01:32] Right. Our gtest detection won't work unless you've got the gtest sources unpacked somewhere. [01:32] (As is recommended by upstream) [01:32] upstream dropped that recommendation when they fully moved over to cmake [01:33] the apt guys told me the same thing... [01:34] * RAOF heads out. [01:36] RAOF: https://github.com/Debian/apt/commit/99ba7cc1901c761c97d67775f23858b86594f2ba === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [09:03] Pharaoh_Atem: it was proposed but not yet added because a way to calibrate/scale the reported touch contact size unit into something meaningful was missing.. === chihchun_afk is now known as chihchun === dandrader is now known as dandrader|afk === marcusto_ is now known as marcustomlinson === hikiko is now known as hikiko|ln === hikiko|ln is now known as hikiko === dandrader|afk is now known as dandrader === shuduo is now known as shuduo-afk === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader === chihchun is now known as chihchun_afk === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader