[not-for-merge] Towards statically type-checking parsl by benclifford · Pull Request #1676 · Parsl/parsl

Remove broken hard-coded usage tracking IP address.

benclifford

committed

upgrade mypy

benclifford

committed

Remove unused codepath from HighThroughputExecutor scale_in

benclifford

committed

this test only tests the behaviour from simple strategy, so turn it into a simple strategy test

benclifford

committed

test partial scaling-in (the scaling in the comes from one block being idle, not from the whole executor being idle)

benclifford

committed

uses human readable names not numeric indices for scale in code

benclifford

committed

make scale in pick longest idle blocks in preference to least idle blocks

benclifford

committed

scale-in should consider all known blocks, not just blocks that have at least one manager registered

benclifford

committed

fiddle with submiterror

benclifford

committed

make storage access a sequence for type-checker goodness

benclifford

committed

fiddling with wqtaskfailrue exception

benclifford

committed

connected workers is used in strategy (and exposed as part of executor interface) only for printing a status line. this is awkward to statically typecheck, so remove it - and its awkward to typeche…

benclifford

committed

serialization performance test client

benclifford

committed

Prohibit adding duplicate-labelled executors

benclifford

committed

Add type annotations to common test configurations

benclifford

committed

htex doesn't shut down its queue management thread at shutdown

benclifford

committed

Form WQ worker_command in init to make type visible to mypy

benclifford

committed

add invocation type annotations to see how well type checking of decorators works

benclifford

committed

Add typing for dfk.submit()

benclifford

committed

more wq typechecking - add asserts to reflect assumed behaviour

benclifford

committed

move port mailbox into init so it can be seen by typechecker

benclifford

committed

Squash benc-mypy history

benclifford

committed

Add type annotations to common test configurations

benclifford

committed

remove None-type from provider API

benclifford

committed

marker for what i really expect to work

benclifford

committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[not-for-merge] Towards statically type-checking parsl#1676

Commits on Feb 17, 2024

Remove broken hard-coded usage tracking IP address.

upgrade mypy

Remove unused codepath from HighThroughputExecutor scale_in

this test only tests the behaviour from simple strategy, so turn it into a simple strategy test

test partial scaling-in (the scaling in the comes from one block being idle, not from the whole executor being idle)

uses human readable names not numeric indices for scale in code

make scale in pick longest idle blocks in preference to least idle blocks

scale-in should consider all known blocks, not just blocks that have at least one manager registered

fiddle with submiterror

make storage access a sequence for type-checker goodness

fiddling with wqtaskfailrue exception

connected workers is used in strategy (and exposed as part of executor interface) only for printing a status line. this is awkward to statically typecheck, so remove it - and its awkward to typeche…

serialization performance test client

Prohibit adding duplicate-labelled executors

Add type annotations to common test configurations

htex doesn't shut down its queue management thread at shutdown

Form WQ worker_command in init to make type visible to mypy

add invocation type annotations to see how well type checking of decorators works

Add typing for dfk.submit()

more wq typechecking - add asserts to reflect assumed behaviour

move port mailbox into init so it can be seen by typechecker

Squash benc-mypy history

Add type annotations to common test configurations

remove None-type from provider API

marker for what i really expect to work