FSA Extract inheritFrom and inheritanceType handling from AclBuilder.

This removes inheritFrom and inheritanceType handling from
AclBuilder and moves that responsibility to FsAdaptor, which
has knowledge of ACL inheritance structure.  This refactoring
changes the API to the AclBuilder methods used.  This also
requires significant changes to AclBuilderTest.

[ejona, 2014-02-05 21:37:59]

Looks great.

I reviewed it pretty deeply checking for accidental behavior modifications and
didn't spot any.

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
File src/com/google/enterprise/adaptor/fs/AclBuilder.java (right):

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/AclBuilder.java:65: public Acl getAcl()
throws IOException {
On all of these, it seems like it would be nicer for FsAdaptor if they returned
the Acl.Builder. I can't identify anything we gain by returning Acl instead of
Acl.Builder.

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/AclBuilder.java:85: public Acl
getShareAcl() throws IOException {
So this method is no longer strictly necessary. Given how much the filtering has
changed I think it could be reasonable to keep it, but things do seem to have
stabilized such that I think it is a good idea to remove it and use getAcl()
instead.

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
File src/com/google/enterprise/adaptor/fs/FsAdaptor.java (right):

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/FsAdaptor.java:433: private Acl
newFolderAcl(Acl acl, DocId inheritFrom) {
I think once AclBuilder returns Acl.Builder instead of Acl, getting rid of these
two methods isn't too painful.

Although it is redundant, it is really nice to have the sets explicit on the
usage.

[Brett, 2014-02-06 19:31:04]

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
File src/com/google/enterprise/adaptor/fs/AclBuilder.java (right):

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/AclBuilder.java:65: public Acl getAcl()
throws IOException {
On 2014/02/05 21:37:59, ejona wrote:
> On all of these, it seems like it would be nicer for FsAdaptor if they
returned
> the Acl.Builder. I can't identify anything we gain by returning Acl instead of
> Acl.Builder.

Done.

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/AclBuilder.java:85: public Acl
getShareAcl() throws IOException {
On 2014/02/05 21:37:59, ejona wrote:
> So this method is no longer strictly necessary. Given how much the filtering
has
> changed I think it could be reasonable to keep it, but things do seem to have
> stabilized such that I think it is a good idea to remove it and use getAcl()
> instead.

Done.

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
File src/com/google/enterprise/adaptor/fs/FsAdaptor.java (right):

https://codereview.appspot.com/54820046/diff/1/src/com/google/enterprise/adap...
src/com/google/enterprise/adaptor/fs/FsAdaptor.java:433: private Acl
newFolderAcl(Acl acl, DocId inheritFrom) {
On 2014/02/05 21:37:59, ejona wrote:
> I think once AclBuilder returns Acl.Builder instead of Acl, getting rid of
these
> two methods isn't too painful.
> 
> Although it is redundant, it is really nice to have the sets explicit on the
> usage.

Done.

[Brett, 2014-02-06 19:47:26]

Eric's Feedback

[ejona, 2014-02-06 20:22:07]

LGTM.

It would probably be reasonable for Miguel to take a cursory look to make sure
no stepping on toes, but I don't expect any strong issues.

[Brett, 2014-02-08 03:44:06]

On 2014/02/06 20:22:07, ejona wrote:
> LGTM.
> 
> It would probably be reasonable for Miguel to take a cursory look to make sure
> no stepping on toes, but I don't expect any strong issues.

I would like this to go in before his DFS share ACL incremental update change.

[ejona, 2014-02-10 23:51:01]

Miguel, could you do a quick review of this? I feel confident with the
correctness, but it just seems appropriate that we hear from you.

On 2014/02/08 03:44:06, Brett wrote:
> I would like this to go in before his DFS share ACL incremental update change.

I have no problem with that. The two of you should talk to prevent any issues.

[mifern, 2014-02-11 00:42:20]

LGTM. Everything looks good and I would add cleaner.

[Brett, 2014-02-11 03:07:05]

Committed 10 February 2014 
To https://code.google.com/p/plexi.fs/
   2de0df5..6c240cd  master -> master