Refactor IOSimPOR #114
Merged
Refactor IOSimPOR #114
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This allows us to use `tasty-1.5`.
coot
force-pushed
the
coot/refactor-io-sim-por
branch
4 times, most recently
from
a71aef7 to
135e42a
Compare
With unsafeInterleaveSTconjoinParST :: TestableNoCatch prop => [ST s prop] -> ST s Property
conjoinParST sts = do
ps <- sequence (unsafeInterleaveST <$> sts)
return $ conjoinPar ps
Note: on a machine ith 12 physical cores, 24 CPU threads. Without unsafeInterleaveSTconjoinParST :: TestableNoCatch prop => [ST s prop] -> ST s Property
conjoinParST sts = do
ps <- sequence sts
return $ conjoinPar ps
|
coot
force-pushed
the
coot/refactor-io-sim-por
branch
from
135e42a to
5aec9e9
Compare
|
Any idea why the with/without |
coot
force-pushed
the
coot/refactor-io-sim-por
branch
from
5aec9e9 to
5bd2d80
Compare
|
No, it's a bit surprising. When I run it now I get similar results. Different test case, and previously I was running with Par
Seq
|
coot
force-pushed
the
coot/refactor-io-sim-por
branch
from
27c6367 to
a976591
Compare
bolt12
reviewed
Sep 25, 2023
Added `BlockedOnDelay` and `BlockedOnThrowTo` instead of `BlockedOnOther`.
Instead of using `forkIO` & `threadDelay` it's more robust to test with `async` and `wait`.
This allows to show more compact information about execution. It's useful for debugging IOSimPOR.
We report `EventTxWakup` in the thread that waken up the thread, but we report the `ThreadId` of the awaken thread. Including step of the executed thread is misleading.
This is only useful for debugging `IOSimPOR` issues.
The original type clashes with `ThreadId` type class from `io-classes` which is in particular annoying when pasting `ControlSchedules` in `ghci` or counterexamples.
Consider the following sequence of events x ⋮ y ⋮ z If x is racing with z, we will reverse x and z (e.g. postpone x after z). If y is also racing with x, then we should not include in the schedule any event which happens after it. Originally we used a policy to remove y's thread from the list of racing threads with the x's thread. But this turns out to be too week. While we are visiting new events we need to remove their threads as soon as the event happens after any of the so far discovered races. Otherwise we can end with a schedule which contains events that can only happen after the racing event was executed, and thus likely will block and violate internal invariant: all steps which we follow should be in the runqueue (e.g. should not be blocked).
Note: `deschedule` is traced by the caller, not by itself.
This makes it much easier to debug ones. In the future we might turn more assertion failures into `FailureInternal`.
Unlike `exploreSimTrace` it allows to run an `ST` action in the context of each schedule, and thus allows to share state between schedules. For this reason it evaluates each schedule sequentially, while `exploreSimTrace` evaluates each schedule in parallel. This patch also makes the API more uniform by exposing: * `exploreSimTrace` & `exploreSimTraceST` * `controlSimTrace` & `controlSimTraceST` This patch also removes parallel evaluation of different schedules. Furthermore the tests are updated and they no longer use `unsafePerformIO` as well. The `traceCounter` property was removed as it wasn't used and doesn't make much sense, since `IOSimPOR` is forcing to evaluate each schedule only once by using a cache.
This should be exposed as part of the `newTimeout` API.
This can be added to `.hlint.yaml` file (we do not checkout it to the git repo, but one can add it to one of the `.gitignore` files).
coot
force-pushed
the
coot/refactor-io-sim-por
branch
from
a976591 to
5d15651
Compare
|
@bolt12 I addressed your review. I accepted most of your comments, the one that I challenged I left opened. |
bolt12
approved these changes
Sep 26, 2023
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A long list of self contained patches. See
CHANGELOG.mdfor a short description of all the changes.