Mailing-List: contact crunch-dev-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: crunch-dev@incubator.apache.org
Received-SPF: pass (nike.apache.org: domain of jwills@cloudera.com designates
 209.85.213.175 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CAF-WD4SboOW7rnD=ZJzzFHEprR1DmHk5O7HZtVb0iBcDAxh5Hw@mail.gmail.com>
References: 
 <CAA5C_pvofK4DZjsXpFeEeS_C8nBMOCzkzi=OrxVicTC9EHp3KA@mail.gmail.com>
 <CAH29n6P8=BgYFSYSj+4T425Lkjf+wO+bqEDnfZQ3DJxBgi8dxA@mail.gmail.com>
 <CAOKEptZK2YCk=gCaFuXm1KoiHh0zwwzvozDP4G3j_mVophf+BA@mail.gmail.com>
 <CAF-WD4SboOW7rnD=ZJzzFHEprR1DmHk5O7HZtVb0iBcDAxh5Hw@mail.gmail.com>
From: Josh Wills <jwills@cloudera.com>
Date: Tue, 19 Jun 2012 14:19:46 -0700
Message-ID: 
 <CAH29n6Ow2VZT=H7Xjz6THicWgDfwu_qBoHnkSxd_8FtncV60aw@mail.gmail.com>
Subject: Re: Compiler warnings in Crunch
To: crunch-dev@incubator.apache.org
Content-Type: text/plain; charset=ISO-8859-1

On Tue, Jun 19, 2012 at 1:56 PM, Tom White <tom@cloudera.com> wrote:
> On Mon, Jun 18, 2012 at 6:42 PM, Robert Chu <robert@wibidata.com> wrote:
>> +1 to fixing compiler warnings. I would prefer proper usage of the wildcard
>> type to just suppressing the warnings.
>
> +1 Suppressing generics warnings should be the means of last resort,
> and should be done at the smallest possible scope.
>
> Regarding the serialization warnings, I think it's better not to add
> serial UIDs everywhere since they add clutter. You can turn off the
> warnings in Eclipse instead - would that acceptable?

I'm okay with that-- is it a setting we can add to the config settings
Gabriel just checked in?

>
> Cheers,
> Tom
>
>>
>> Robert
>>
>> On Mon, Jun 18, 2012 at 7:57 AM, Josh Wills <jwills@cloudera.com> wrote:
>>
>>> On Mon, Jun 18, 2012 at 6:24 AM, Gabriel Reid <gabriel.reid@gmail.com>
>>> wrote:
>>> > As prep for some other development I was going to do in Crunch, I was
>>> > considering cleaning up some (or all) of the compiler warnings that
>>> > are currently occurring (they make the right-side search ribbon in
>>> > Eclipse almost unusable right now).
>>> >
>>> > A significant portion of the compiler warnings are raw type generics
>>> > warnings, i.e. "xxx is a raw type. References to xxx should be
>>> > parameterized", where we're doing general operations with PCollections
>>> > and similar objects without knowing anything about their generic
>>> > types.
>>>
>>> There are also the warnings about not adding serialization UIDs to the
>>> built-in DoFns, which irritate me and are useless in the context of
>>> Crunch. Happy to volunteer to go around and add UID = 1; code all over
>>> the place if there are no objections.
>>>
>>> >
>>> > We could add wildcards (i.e. PCollection<?>) to each of these generic
>>> > objects in the methods where the warnings are occurring -- this would
>>> > be my preferred thing to do. On the other hand, I think that adding
>>> > wildcards make the code more difficult to read, while having any kind
>>> > of real added value.
>>> >
>>> > The other option we could take (less preferable to me) is to use
>>> > @SuppressWarnings("rawtypes") on some or all of the affected methods
>>> > -- it'll leave the code in a more readable state, but I'm not that
>>> > wild about just suppressing warnings.
>>>
>>> I'm a 0 on the approach here-- I always have a hard time getting the
>>> <?> stuff to compile when I'm casting the result, which is what often
>>> happens in Writables.java and Avros.java, but I agree that a working
>>> version of the wildcards is preferable to suppress warnings. We might
>>> say that we prefer <?> but add in SuppressWarnings when there is no
>>> other option for what we're trying to do.
>>>
>>> >
>>> > Anyone else care to weigh in on this?
>>> >
>>> > - Gabriel
>>>
>>>
>>>
>>> --
>>> Director of Data Science
>>> Cloudera
>>> Twitter: @josh_wills
>>>


-- 
Director of Data Science
Cloudera
Twitter: @josh_wills