Conversation
c5df1e4 to
08d273a
Compare
08d273a to
9239756
Compare
…cit-sync on wl_shm buffers. The only guaranteed-supported buffers are linux-dmabuf buffers. It's easy for Mir to support wl_shm buffers, too, but maybe other compositors will not support wl_shm here. Skip all the tests which rely on committing a wl_shm buffer when an `unsupported_buffer` error is returned
Clients are allowed to make empty or null commits to surfaces managed by a syncobj (as long as they don't also set syncpoints).
mattkae
left a comment
There was a problem hiding this comment.
This makes sense by me 👍
|
|
||
| namespace | ||
| { | ||
| class Fd |
There was a problem hiding this comment.
Nit: This could be nice to have in its own file for the future
AlanGriffiths
left a comment
There was a problem hiding this comment.
I'd like to avoid tests that sleep, especially for more than a few milliseconds.
Isn't there any scope for client.dispatch_until(...) or similar? Failures might still take a_long_time() but success should be fast.
| // We don't have a way to ensure that WM has settled down, so just wait a little | ||
| std::this_thread::sleep_for(1s); |
There was a problem hiding this comment.
This seems a very long time to be sleeping in a test - especially twice. (I'm no fan of sleeping in a test - that way leads to very slow feedback from the tests - imagine if every test case took at least a second...)
| // Again, we don't have a way to ensure that WM has settled down, so just wait a little | ||
| std::this_thread::sleep_for(1s); | ||
|
|
||
| EXPECT_THAT(surface,IsSurfaceOfSize(new_width, new_height)); |
There was a problem hiding this comment.
| // Again, we don't have a way to ensure that WM has settled down, so just wait a little | |
| std::this_thread::sleep_for(1s); | |
| EXPECT_THAT(surface,IsSurfaceOfSize(new_width, new_height)); | |
| client.dispatch_until([&]() | |
| { | |
| return surface_has_size(surface, new_width, new_height); | |
| }); |
|
@RAOF ping! |
|
Needs work, converting to draft. |
No description provided.