Fixing resource leak in GZipContent.writeTo

Fix for http://code.google.com/p/google-http-java-client/issues/detail?id=61

[rmistry, 2012-01-25 15:26:18]

This is to fix
http://code.google.com/p/google-http-java-client/issues/detail?id=61

The deflator created by new GZIPOutputStream(out) needs to be end() otherwise
there is a resource leak.
I added Javadoc explaining why this is done. Also, since this class is package
private I think this is a non-breaking change.

[rmistry, 2012-01-26 17:39:49]

Added more detailed javadoc to HttpContent.writeTo

[yanivi, 2012-01-26 20:44:48]

LGTM

This is fine, but we also need to solve the more general problem.  We need to
make a pass at all of the callers of the writeTo method to make sure they are
manually closing the output stream, e.g. AbstractHttpContent.computeLength(). 
We should also make a pass at the implementations of writeTo and try to change
them to not close the output stream, e.g. JacksonFactory.writeTo.

http://codereview.appspot.com/5572060/diff/3/google-http-client/src/main/java...
File
google-http-client/src/main/java/com/google/api/client/http/HttpContent.java
(right):

http://codereview.appspot.com/5572060/diff/3/google-http-client/src/main/java...
google-http-client/src/main/java/com/google/api/client/http/HttpContent.java:48:
* should assume that this method does not close the output stream unless
otherwise specified.
2nd sentence doesn't make sense in general because a lot of times you don't know
what implementation you have because you just get an interface.  So a safer
assumption is that the callers should not assume whether or not the output
stream has been closed.

[rmistry, 2012-01-26 22:21:36]

I did a pass of the callers and the implementers and fixed them. I left tests
alone. Some though were not obvious-
ContentEntity.writeTo and NetHttpRequest.execute().

Please take another look.

http://codereview.appspot.com/5572060/diff/3/google-http-client/src/main/java...
File
google-http-client/src/main/java/com/google/api/client/http/HttpContent.java
(right):

http://codereview.appspot.com/5572060/diff/3/google-http-client/src/main/java...
google-http-client/src/main/java/com/google/api/client/http/HttpContent.java:48:
* should assume that this method does not close the output stream unless
otherwise specified.
On 2012/01/26 20:44:48, yanivi wrote:
> 2nd sentence doesn't make sense in general because a lot of times you don't
know
> what implementation you have because you just get an interface.  So a safer
> assumption is that the callers should not assume whether or not the output
> stream has been closed.

Done.

[yanivi, 2012-01-27 16:46:51]

Can we please merge this into http://codereview.appspot.com/5482047/ and then
I'll review it there?

[rmistry, 2012-01-27 16:59:30]

On 2012/01/27 16:46:51, yanivi wrote:
> Can we please merge this into http://codereview.appspot.com/5482047/ and then
> I'll review it there?

Done. Merged this into http://codereview.appspot.com/5482047/
I will go ahead and close this issue then.