=== chihchun is now known as chihchun_afk | ||
=== tmpRAOF is now known as RAOF | ||
=== chihchun_afk is now known as chihchun | ||
duflu | That's scary. I'm seeing documentation that shows the C++14 ways of doing things won't still work in C++17 | 04:06 |
---|---|---|
RAOF | duflu: Which particular bits? I seem to remember seeing some edge-case stuff that didn't seem to be terribly unreasonable. | 05:19 |
duflu | RAOF: unique_ptr and destructors I think it was | 05:57 |
=== chunsang is now known as chunsang-away | ||
chrisccoulson | RAOF, are you still around? I've got something you might be able to help me with :) | 07:35 |
RAOF | chrisccoulson: I am, briefly. | 07:35 |
RAOF | chrisccoulson: What can I do you for? | 07:35 |
chrisccoulson | RAOF, how familiar are you with the graphics architecture in oxide + chromium? | 07:36 |
RAOF | chrisccoulson: I think you might be after racarr. | 07:36 |
RAOF | chrisccoulson: On the basis that he did most of the Mir backend for chromium. | 07:37 |
chrisccoulson | RAOF, possibly, but you don't know what I'm going to ask next :) | 07:37 |
RAOF | I am not particularly familiar with the graphics arcitectures you're talking about :) | 07:37 |
chrisccoulson | RAOF, the short story is - Oxide has a webiew compositor that composites to a texture via chromium's GPU service thread. We then create an EGLImage from that which gets passed to the QML scenegraph compositor | 07:38 |
RAOF | Sensible enough. | 07:39 |
chrisccoulson | RAOF, but we want to move the GPU service thread out of process, as it is with Chrome / Chromium | 07:39 |
RAOF | Also sensible enough. | 07:39 |
chrisccoulson | so the EGLImage we pass to QML has to be backed by something that lives in another process | 07:39 |
RAOF | Welcome to nonstandardland. | 07:40 |
chrisccoulson | I know how to approach this on X + GLX (we have a different compositing path for X currently) | 07:40 |
RAOF | There's no simple way to get an EGLImage across process boundaries. | 07:40 |
chrisccoulson | Yeah, that's what I was expecting | 07:40 |
RAOF | In the not *too* distant future EGLStream is what you're after. | 07:41 |
chrisccoulson | Having different implementations for mesa / hybris isn't really too much of an issue | 07:41 |
RAOF | For mesa you can use... EGL_EXT_image_dma_buf_import and EGL_MESA_drm_image, I think. | 07:43 |
RAOF | For hybris... | 07:43 |
RAOF | ...have you perhaps considered becoming a Mir compositor? :) | 07:44 |
chrisccoulson | Hah, I haven't. I'm not entirely sure what that would involve (we're pretty much tied to Chromium's compositor) | 07:45 |
RAOF | Oh, you'd just use the Mir backend for that :) | 07:45 |
chrisccoulson | RAOF, For mesa, I was looking at libgbm. Is that not the right thing to use? | 07:45 |
RAOF | Not really, no. | 07:46 |
chrisccoulson | ah, ok | 07:46 |
* RAOF heads out. | 07:48 | |
RAOF | Sorry I couldn't be of more help! | 07:48 |
chrisccoulson | RAOF, oh, you've been helpful enough. Thanks! :) | 07:49 |
=== c74d is now known as Guest42134 | ||
=== marcusto_ is now known as marcustomlinson_ | ||
=== marcustomlinson_ is now known as marcustomlinson | ||
=== benonsoftware is now known as clockwork | ||
=== clockwork is now known as benonsoftware | ||
=== tmpRAOF is now known as RAOF | ||
=== alan_g is now known as alan_g|lunch | ||
=== alan_g|lunch is now known as alan_g | ||
=== ahayzen is now known as Guest21985 | ||
=== ahayzen_ is now known as help | ||
=== help is now known as ahayzen_ | ||
=== ahayzen_ is now known as ahayzen | ||
=== 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 | ||
=== alan_g is now known as alan_g|EOD |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!