Conversation
petrutlucian94
changed the title
|
@petrutlucian94 I'll start reviewing this one while we wait for the other parts to merge. What's the first unique commit for this PR? |
That would be 842fc6b win32: avoid broken mingw rng Here's a git log, showing the PR structure: http://paste.openstack.org/raw/792975/ Thanks a lot for reviewing those patches. I'll take another look at the chain and try to avoid unnecessary changes, should be easier now after getting to know the code base better. |
petrutlucian94
force-pushed
the
windows.12
branch
10 times, most recently
from
7652884 to
564d0ae
Compare
petrutlucian94
force-pushed
the
windows.12
branch
from
1c56f4d to
1239448
Compare
petrutlucian94
force-pushed
the
windows.12
branch
2 times, most recently
from
ebe1a9e to
d01616e
Compare
petrutlucian94
force-pushed
the
windows.12
branch
2 times, most recently
from
8e8d942 to
e1dd05a
Compare
petrutlucian94
force-pushed
the
windows.12
branch
2 times, most recently
from
2d8ca1a to
3c29a6d
Compare
petrutlucian94
force-pushed
the
windows.12
branch
2 times, most recently
from
90c70fd to
1c108f0
Compare
petrutlucian94
force-pushed
the
windows.12
branch
from
1c108f0 to
aa5bd20
Compare
|
Can one of the admins verify this patch? |
petrutlucian94
force-pushed
the
windows.12
branch
3 times, most recently
from
46067f9 to
86f8706
Compare
petrutlucian94
force-pushed
the
windows.12
branch
3 times, most recently
from
4fcc634 to
395804f
Compare
petrutlucian94
force-pushed
the
windows.12
branch
3 times, most recently
from
a85fbb5 to
bb53642
Compare
|
This PR now includes submodule changes. See the "Unmodified Submodules" check |
My bad, I must've used "git commit -a" by mistake. Thanks for letting me know! |
petrutlucian94
force-pushed
the
windows.12
branch
from
bb53642 to
5643d50
Compare
|
jenkins test make check |
|
For test results please see #32776 (comment) |
We're porting to Windows some of the Ceph tests, mostly related to rados and rbd. Overall, this is about using compat.h functions and types as well as avoiding unsupported functionality. The systest framework will have to be limitted to threads instead of separate processes until we cover the process spawning and communication. Signed-off-by: Lucian Petrut <[email protected]>
We're porting to Windows some of the unit tests, mostly related to rados and rbd. This change filters out the tests that aren't supported on Windows. At the same time, we're defining some missing dependencies since unlike ELFs SOs, all DLL symbols have to be resolved at compile time. Signed-off-by: Lucian Petrut <[email protected]>
The windows pthread library doesn't support named semaphores, for which reason we're updating some tests to use unnamed semaphores. Signed-off-by: Lucian Petrut <[email protected]>
Some timer tests don't actually initialize the timer, for which reason the callbacks will never be executed, leading to an infinite loop. This change initializes the timer, stopping it after the tests finish. Signed-off-by: Lucian Petrut <[email protected]>
|
Latest test results are here @neha-ojha @dillaman If you could spare some time for these, we'd be much obliged. |
@jan--f none of the failures look related |
8 participants
Port tests to Windows
This PR ports some of the Ceph tests to Windows, providing significant test coverage for the client libraries.
Here are some test_results.zip.
For more details about the build process and current status, please check the
README.windows.rst file.
Other Windows related PRs:
Part 1: #31981Part 2: #32704Part 3: #32705Part 4: #32707Part 5: #32780Part 6: #32779Part 7: #32778Part 8: #32777ceph/fmt#2ceph/rocksdb#42#32027 librados: avoid symbol versioning on Windows
#32776 Update Windows build scripts
#34858 Add Windows CephFS support
#33750 Add Windows RBD support