Roll back the write transaction on schema init if nothing changed #1281
Conversation
|
I think I would prefer the simpler approach if possible rather than trying to change a bunch of stuff at once. I think I see at least one place where there could now be a race condition if initializing a Realm from multiple threads simultaneously, and I'm not convinced that we aren't breaking things by changing where and how our synchronization blocks/mutexes/exclusive writes are used. I'm also not thrilled with the new factoring, and if we want to go ahead with these changes we may want to move/rename some stuff so that the initialization code is more consolidated an the logic is more compact. |
tgoyne
force-pushed
the
tg-interprocess-init
branch
from
4e5f4a5 to
597ab4f
Compare
tgoyne
changed the title
|
Now does way less stuff since all the other bits have been done better in different PRs. |
| } | ||
| } | ||
|
|
||
| if (oldVersion != newVersion) { |
There was a problem hiding this comment.
This is the same as migrating. This should be combined with the previous block, with if (migrationBlock) in a nested code block.
|
Other than the code factoring issues 👍 |
tgoyne
force-pushed
the
tg-interprocess-init
branch
from
597ab4f to
5a99bb0
Compare
Eliminates spurious change notifications when opening a Realm flle in multiple processes.
Roll back the write transaction on schema init if nothing changed
Eliminates spurious change notifications when opening a Realm flle in multiple processes.