IOLocal propagation for unsafe access
#3636
Conversation
armanbilge
force-pushed
the
topic/thread-local-iolocal
branch
from
dc9b74c to
0b88c01
Compare
| @@ -43,4 +43,6 @@ final class IOFiberConstants { | |||
| static final byte CedeR = 6; | |||
| static final byte AutoCedeR = 7; | |||
| static final byte DoneR = 8; | |||
|
|
|||
| static final boolean dumpLocals = Boolean.getBoolean("cats.effect.tracing.dumpLocals"); | |||
There was a problem hiding this comment.
Bikesheddable configuration for opting-in. So the rest of us don't have to pay the penalty 😇
There was a problem hiding this comment.
Is this specifically "tracing", even if that's the most obvious use case?
There was a problem hiding this comment.
Oh woops, this was a very lazy copy-pasta. I copied it from the system properties we use to configure fiber tracing. We should rename it anyway, dumpLocals is not quite right I think 😅
There was a problem hiding this comment.
What about cats.effect.localContextPropagation similar to Monix's monix.environment.localContextPropagation?
There was a problem hiding this comment.
Thanks, I liked that! I went with cats.effect.ioLocalPropagation.
| var locals: IOLocals = null | ||
| if (dumpLocals) { | ||
| locals = new IOLocals(localState) | ||
| IOLocals.threadLocal.set(locals) | ||
| } |
There was a problem hiding this comment.
I did this just for delay, but I guess blocking and interruptible would want it too.
| // TODO handle defaults and lenses. all do-able, just needs refactoring ... | ||
| final class IOLocals private[effect] (private[this] var state: IOLocalState) { |
There was a problem hiding this comment.
Ok, so I was extremely lazy with implementing this thing. But the main idea of this wrapper API is that it should only give the user access to IOLocals that they know about i.e. they should not able to clear out other locals that happen to be present.
| val fiber = new IOFiber[A]( | ||
| Map.empty, | ||
| if (IOFiberConstants.dumpLocals) unsafe.IOLocals.getState else Map.empty, |
There was a problem hiding this comment.
We can even go in the opposite direction for IO#unsafeRun* 😁
It's less clear if/how to do this for fibers started in a Dispatcher, since they should be inheriting locals from the fiber backing the Dispatcher.
| @@ -43,4 +43,6 @@ final class IOFiberConstants { | |||
| static final byte CedeR = 6; | |||
| static final byte AutoCedeR = 7; | |||
| static final byte DoneR = 8; | |||
|
|
|||
| static final boolean dumpLocals = Boolean.getBoolean("cats.effect.tracing.dumpLocals"); | |||
There was a problem hiding this comment.
Is this specifically "tracing", even if that's the most obvious use case?
| // defined in Java since Scala doesn't let us define static fields | ||
| final class IOFiberConstants { | ||
| public final class IOFiberConstants { |
There was a problem hiding this comment.
This is not good. To avoid this we'll either have to replicate it at both the cats.effect and cats.effect.unsafe levels, or move the thread-local IOLocals accessors into cats.effect.
|
I apologise for jumping in and asking this question but, can it be used as an alternative to Local from Monix? |
|
@kevin-lee no problem! At a glance, yes, this does look like an alternative/replacement to that Monix feature. Perhaps @alexandru can confirm :) |
armanbilge
changed the title
IOLocalsIOLocals
@armanbilge Thank you. That's great! |
@kevin-lee the For an example integration, see this PR which implements a Java SPI using this |
@armanbilge Oh... got it. It looks very promising then. Thank you! |
|
Finished benchmarking. @armanbilge let's see your armchair performance reasoning explain this result. I'm at a loss. BeforeAfter |
|
Finally scraped together the time needed to shave a few dev environment yaks and test this PR with propagation enabled. The performance hit is about 25% on microbenchmarks involving So I think the summary of this PR, from a performance standpoint, is the following:
So enabling this is analogous to enabling cached tracing, but of course would be cumulative with that flag as well, so technically turning both of them on would bop your microbenchmark scores by something like 55%. It's very unclear whether this type of performance impact matters in practice. We would probably need to run some end-to-end scale tests with propagation enabled in order to see. Given that this is default-off and the impact when disabled is zero (outside of the still bizarre WSTP microbenchmark mirages), I'm comfortable merging this as-is and we can do some further testing. We should also probably document (or at least, write a ticket to document) the performance effects so it doesn't surprise anyone. |
|
I added some documentation.
Discussed on Discord, but to clarify there is no reason to expect that |
|
@djspiewak Please forgive my impetuousness - do you have a guesstimate of when this might be going in please? |
|
Oh-em-gee. Exciting. |
|
@armanbilge Conflict resolution time! |
|
can someone drop a message here when this PR ends up in a release? |
|
@NthPortal I'd recommend subscribing to releases additionally/instead: 
|
|
We're working to get an RC out asap! |
Still needs quite a bit of work, but wanted to sketch the basic idea.
Goal: to expose a fiber's
IOLocals asThreadLocals within a side-effecting block, so that they can be accessed and modified in unsafe land.Motivation: basically telemetry Java interop.
Constraints: to do this as safely as possible 😅