ARROW-17178: [R] Support head() in arrow_dplyr_query with user-defined function#13706

Merged

paleolimbot merged 34 commits intoapache:masterfrom

paleolimbot:udf-head

Sep 16, 2022

Commits on Sep 15, 2022

consolidate exec plan reader into a dedicated reader class
paleolimbot
committed
fix the head thing
paleolimbot
committed
undo some kludges introduced
paleolimbot
committed
explicitly close readers where possible
paleolimbot
committed
just kidding don't do that
paleolimbot
committed
just a shot in the dark
paleolimbot
committed
try a new strategy for stopproducing
paleolimbot
committed
try a more raw approach to stoppinh production
paleolimbot
committed
always set status
paleolimbot
committed
another approach to stopping production
paleolimbot
committed
more sane but still deadlocking stop_producing
paleolimbot
committed
uncomment
paleolimbot
committed
go back to the other approach
paleolimbot
committed
remove unnecessary Close()
paleolimbot
committed
maybe fix python test
paleolimbot
committed
with ExecPlanReader R6
paleolimbot
committed
undo python workaround
paleolimbot
committed
with tests to enforce laziness
paleolimbot
committed
document the rbr subclasses
paleolimbot
committed
exec plan reader head() + tests
paleolimbot
committed
remove collect with udf kludge
paleolimbot
committed
nix the custom head method
paleolimbot
committed
wait for idle thread pool on valgrind job
paleolimbot
committed
remove thread pool stuff (it didn't work)
paleolimbot
committed
simplify arrow_dplyr_query to table
paleolimbot
committed
add test for lazy head()
paleolimbot
committed
don't use R function call recordbatchreader just now
paleolimbot
committed
add a note about lazy stopping but don't test it quite yet
paleolimbot
committed
Apply suggestions from code review

paleolimbot
and
pitrou
committed
make the test less funny and less confusing
paleolimbot
committed
attempt to simplify the call to StopProducing()
paleolimbot
committed
comment to remind myself why the complicated callback thing needs to happen
paleolimbot
committed
update for C++17
paleolimbot
committed
clang-format
paleolimbot
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ARROW-17178: [R] Support head() in arrow_dplyr_query with user-defined function#13706

ARROW-17178: [R] Support head() in arrow_dplyr_query with user-defined function#13706
paleolimbot merged 34 commits intoapache:masterfrom
paleolimbot:udf-head

Commits on Sep 15, 2022

consolidate exec plan reader into a dedicated reader class

fix the head thing

undo some kludges introduced

explicitly close readers where possible

just kidding don't do that

just a shot in the dark

try a new strategy for stopproducing

try a more raw approach to stoppinh production

always set status

another approach to stopping production

more sane but still deadlocking stop_producing

uncomment

go back to the other approach

remove unnecessary Close()

maybe fix python test

with ExecPlanReader R6

undo python workaround

with tests to enforce laziness

document the rbr subclasses

exec plan reader head() + tests

remove collect with udf kludge

nix the custom head method

wait for idle thread pool on valgrind job

remove thread pool stuff (it didn't work)

simplify arrow_dplyr_query to table

add test for lazy head()

don't use R function call recordbatchreader just now

add a note about lazy stopping but don't test it quite yet

Apply suggestions from code review

make the test less funny and less confusing

attempt to simplify the call to StopProducing()

comment to remind myself why the complicated callback thing needs to happen

update for C++17

clang-format