3257
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3256
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3255
|
|
|
Andreas Pokorny |
8 years ago
|
|
|
3254
|
|
|
Daniel van Vugt |
8 years ago
|
|
|
3253
|
|
|
Kevin DuBois |
8 years ago
|
|
|
3252
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3251
|
|
|
Daniel van Vugt |
8 years ago
|
|
|
3250
|
|
|
Kevin DuBois |
8 years ago
|
|
|
3249
|
|
|
Andreas Pokorny |
8 years ago
|
|
|
3248
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3247
|
|
client: sort out the lifetime and ownership of MirBufferStreams. The streams' lifetimes are now managed by MirConnection, and SurfaceMap is where MirConnection stores the surfaces.
Previously, they were imperfectly managed by SurfaceMap, eg, the buffer streams were stored in SurfaceMap, but they were owned the Surface that created them. This led to scenarios where SurfaceMap wasn't really sure if the surface was alive or dead. Since NBS increases the amount of Events being sent to the surface/stream, this would tease out existing problems with this scenario around surface/stream destruction more.
Also eliminate the DefaultClientBufferStreamFactory, as it wasn't really all that useful.
Approved by PS Jenkins bot, Alberto Aguirre, Alexandros Frantzis, mir-ci-bot, Chris Halse Rogers.
|
Kevin DuBois |
8 years ago
|
|
|
3246
|
|
|
Daniel van Vugt |
8 years ago
|
|
|
3245
|
|
|
Andreas Pokorny |
8 years ago
|
|
|
3244
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3243
|
|
|
Andreas Pokorny |
8 years ago
|
|
|
3242
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3241
|
|
|
Alan Griffiths |
8 years ago
|
|
|
3240
|
|
|
Kevin DuBois |
8 years ago
|
|
|
3239
|
|
|
Alexandros Frantzis |
8 years ago
|
|
|
3238
|
|
|
Cemil Azizoglu |
8 years ago
|
|
|