crunch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Micah Whitacre (JIRA)" <>
Subject [jira] [Commented] (CRUNCH-192) Document and enforce the semantics around reducer-based Iterables
Date Mon, 08 Apr 2013 21:27:16 GMT


Micah Whitacre commented on CRUNCH-192:

If this is dumping onto the bug I can log another one, but can you also clarify documentation
for the appropriate use of values from inside the Iterable inside of DoFn and MapFn?

As an example we've gotten bitten by the case where we were storing off the individual items
inside the Iterable to then do processing once we've read all the values in.  

    public Foo map(final Pair<Bar, Iterable<Bat>> input) {
        List<Bat> bats = ...;
        for(Bat b: input){
        return new Foo(bats);

When this gets ran during a reduce, the list bats will end up with a single item instead of
multiple items.  For this to work properly we actually have to make a copy of each item in
the iterable.  Making the javadoc more clearly state this behavior would help consumers to
write the MapFn/DoFn correctly the first time.
> Document and enforce the semantics around reducer-based Iterables
> -----------------------------------------------------------------
>                 Key: CRUNCH-192
>                 URL:
>             Project: Crunch
>          Issue Type: Bug
>            Reporter: Gabriel Reid
>         Attachments: CRUNCH-192.patch
> As reported on by Chad Urso McDaniel:
> BLUF: The Iterable parameter to CombineFn.process implies you can iterate multiple times
when you cannot and this leads to surprising behavior.
> As many of you probably know, the signature of CombineFn.process is 
> ---
> process(Pair<K, Iterable<V>> input, Emitter<Pair<K, V>> emitter)
> ---
> The corresponding Hadoop Reducer signature is
> ---
> reduce(K2 key, Iterator<V2> values, OutputCollector<K3,V3> output, Reporter
> ---
> I assume the Crunch use of Iterable is for convenient use in "for" loops.
> Unfortunately, the behavior of this Iterable seems to return the same Iterator object
each time Iterable.iterator() is called. 
> This makes sense to me based on the underlying hadoop mapreduce, but violates what I
think most expect from the Iterable interface.
> I understand that it's too late to change the interface, but could we at least have an
javadoc or an exception thrown if the Iterable is used more than once?

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message