DisplayDriverServer : Create IPv6 endpoint #1395
Conversation
|
LGTM. This greatly speeds up the new test in GafferHQ/gaffer#5541 that inspired this change. It goes from about 15 seconds to nominal with this change. A cursory look at the Arnold tests gives me hope that it will also help those pass without needing to extend the wait time for pixels. |
Great! There's also a flag you can set to say "this v6 endpoint will also accept v4 connections". It was on by default when I tested, but I'm not sure that is always the case. I'll update the PR to set that too, and do a bit of testing on Linux. |
johnhaddon
force-pushed
the
ipV6DisplayDriver
branch
from
6ce6919
to
2d1cf2f
Compare
|
I reckon we should get this merged and included in Gaffer 1.4. I've updated the PR to explicitly accept v4 connections as well as v6 and to update the changes file. Still looking good on Windows @ericmehl? |
|
LGTM, still getting a very nice performance boost on those tests, and maybe a bit with starting an interactive render. |
This avoids delays on Windows, where the ClientDisplayDriver was first trying a V6 connection and then only making a V4 connection if that failed (after a timeout).
johnhaddon
force-pushed
the
ipV6DisplayDriver
branch
from
2d1cf2f
to
c357ea1
Compare
This avoids delays on Windows, where the ClientDisplayDriver was first trying a V6 connection and then only making a V4 connection if that failed (after a timeout). I don't know enough about this stuff yet to know if this is a good thing to do generally, or if it could cause problems somewhere. But I'm making a draft PR in the hope that @ericmehl can at least confirm that it improves things for him too. For me the difference is very noticeable when running
gaffer test GafferImageTest.CatalogueTest.