Add core types to be used to replace the use of jcr.Session in the kernel API #1107

acoburn · 2016-09-30T18:14:52Z

I have a complete implementation for this, but so as not to make the PR too massive, these are the new interfaces being added which will be used to replace instances of jcr.Session, jcr.Repository, and the existing org.fcrepo.kernel.api.services.TransactionService.

Begins to address: https://jira.duraspace.org/browse/FCREPO-1868

Supersedes #968

ajs6f · 2016-09-30T18:18:10Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+
+    public interface State {}
+
+    public enum RequiredState implements State {


Wait, what does it mean for a FedoraSession to be in ACTIVE as opposed to BATCH?

ACTIVE is something that is used in the current impl to test jcr.Session::isLive. BATCH is a marker to indicate that a particular session persists across multiple HTTP requests.

Do we need to do that here? Can't we just offer begin, commit, etc., and leave it at that? HTTP notions don't belong here, to my mind.

you mean remove ACTIVE? I have that here b/c the http layers test for that.

I'm saying I don't see the need for this state machinery at all.

ajs6f · 2016-09-30T18:19:11Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+
+    /**
+     * Determine whether the session has the provided state
+     * @param <T> an enum that extends the default State enum


You cannot extend a Java enum.

the comment is wrong -- State is an interface, not an enum.

Then what is the point of the generic? Doesn't it come to:

boolean hasState(State state);

?

yes, that's correct, I can simplify that signature.

ajs6f · 2016-09-30T18:20:21Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+     * @param prefix the prefix
+     * @param uri the URI
+     */
+    void setNamespacePrefix(String prefix, String uri);


We don't have any real data-handling methods, but we're still going to demand that you manage namespaces? I don't think this makes sense here.

yes, that part can go.

ajs6f · 2016-09-30T18:21:18Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+     * @return the default timeout value
+     */
+    public static Duration operationTimeout() {
+       if (System.getProperty(TIMEOUT_SYSTEM_PROPERTY) != null) {


How about justjava.lang.System.getProperty(String, String)?

yes, that would be much better

ajs6f · 2016-09-30T18:25:32Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+ * @author acoburn
+ * @since Sept 30, 2016
+ */
+public interface BatchService {


Is this a kind of "manage the connections between users and open transactions" thing?

There is currently an interface called TransactionService. This interface is a slight modification of that, mostly to remove some of the baggage of using the word "transaction"

Yeah. I wonder if we want this stuff in the kernel at all.

Partly, this is about the fact that I don't think we have any clear sense of why there are all these *Service guys.

The reason they are in the kernel is so that the HTTP layers can access these "services" via spring injection. Moving those out would be a structural change that goes way beyond the scope of this effort.

No, I'm not asking you to do that. I just don't want to make any effort to preserve them if we can avoid it.

Okay, leave it be for now. I'm certainly not trying to block this.

@ajs6f: related to the org.fcrepo.kernel.api.services.*Service types, I wonder whether they might not be better situated in a different package -- say: org.fcrepo.spi or org.fcrepo.kernel.spi (which could be independent of the fcrepo-kernel-api module). Once we have a full specification and remove all of the JCR concepts from the API, I think there's a certain amount of re-organization that can happen w/r/t the modules and package namespaces, but that's a conversation for later.

Yeah, let's get through this. I have yet to hear anyone give a purpose for those types. (I know you have OSGi in mind, but I don't think could have been our original purpose. If we decide to have these kinds of types for OSGi, they should be impl'd in a common OSGi-related module, not in the kernel-api forcing everyone to impl them directly.)

I'd argue that the purpose of these types is to make it possible for the HTTP layer (fcrepo-http-api and fcrepo-http-commons) to access those implementation-specific services that are injected via Spring (but that could be substituted with Blueprint for OSGi). Either way, I agree that they are not a necessary part of the kernel-api.

They do that, but they are only needed to do that on the assumption that model types (like FedoraResource) aren't managed by the bean container or other outer system. @awoods and I have talked in the past about whether that's a step we ought to contemplate taking. My point here is that whatever we do in the JCR impl, other people may want to play it out differently elsewhere and we shouldn't make their lives any harder.

awoods · 2016-10-01T00:28:19Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param session The session to use for this batch operation
+     * @param username the name of the {@link java.security.Principal}
+     */
+    void begin(FedoraSession session, String username);


From the FedoraSession interface, it would appear that a userId is already associated with sessions. How is this arg username reconciled with the session userId?

You are correct: the jcr.Session interface has a getUserID method, and the new FedoraSession interface has a getUserId method -- which in the impl just proxies to the jcr.Session::getUserID. The issue is that in the case of an anonymous user, the result of the jcr.Session::getUserID is not null, but rather <anonymous>. But we can't actually rely on that string, because that, too, is configurable in the repository.json file. This way, we are relying on the value of the java.security.Principal::getName value for looking up FedoraSession values.

Believe me -- my first stab at this excluded the username parameter on all of these methods, but relying on that conflicts with how transaction values are looked up.

awoods · 2016-10-01T00:30:40Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param username the name of the {@link java.security.Principal}
+     * @return the {@link FedoraSession} with this user
+     */
+    FedoraSession getSession(String txId, String username);


What is the expected result if the txId exists but not with the provided arg username?

I did not change the behavior of the existing TransactionServiceImpl, namely an Exception is thrown: https://github.com/fcrepo4/fcrepo4/blob/master/fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/TransactionServiceImpl.java#L137-L138

Personally, I'd prefer to return an Optional, but this PR is about removing the jcr.Session, not about changing everything.

awoods · 2016-10-01T00:31:50Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param username the name of the {@link java.security.Principal}
+     * @return the {@link FedoraSession} object for the defined user
+     */
+    boolean exists(String txid, String username);


Minor: change txid to txId for consistency.
..or the other way around.

I'd go further and remove the "tx" part altogether, just using sessionId.

@awoods I changed this and all subsequent txid names to sessionId.

awoods · 2016-10-01T00:32:31Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param txid the Id of the {@link FedoraSession}
+     * @return the {@link FedoraSession} object
+     */
+    default boolean exists(String txid) {


Minor: change txid to txId for consistency.
..or the other way around.

awoods · 2016-10-01T00:32:34Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param txid the id of the {@link FedoraSession}
+     * @param username the name of the {@link java.security.Principal}
+     */
+    void commit(String txid, String username);


Minor: change txid to txId for consistency.
..or the other way around.

awoods · 2016-10-01T00:32:42Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     *
+     * @param txid the id of the {@link FedoraSession}
+     */
+    default void commit(String txid) {


awoods · 2016-10-01T00:32:47Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param txid the id of the {@link FedoraSession}
+     * @param username the name of the {@link java.security.Principal}
+     */
+    void abort(String txid, String username);


awoods · 2016-10-01T00:32:53Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     *
+     * @param txid the id of the {@link FedoraSession}
+     */
+    default void abort(String txid) {


awoods · 2016-10-01T01:38:45Z

👍

acoburn · 2016-10-02T20:27:33Z

@awoods, @ajs6f I added the modeshape implementation for these new interfaces. Once I get the green light on this, I'll add the rest of the commits (which actually remove the use of jcr.Session from the kernel-api).

ajs6f · 2016-10-03T00:48:35Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraRepositoryImpl.java

+     *
+     * @return the JCR Repository
+     */
+    public Repository getRepository() {


This naming is confusing; maybe getJcrRepository?

That works for me. Also, I should double check the use of this method, because we may be able to do away with this altogether.

@ajs6f I renamed the method here and in FedoraSessionImpl (getSession -> getJcrSession). The FedoraRepositoryImpl::getJcrRepository method is only used for unit tests -- I don't know if that's a good reason to keep it, but generally, I tried to keep existing unit/integration tests in place. All things being equal, I would probably remove that test and therefore that method, too. @awoods: thoughts?

acoburn · 2016-10-03T14:30:32Z

@awoods @ajs6f let me know when you are ready for the full implementation. Barring any changes with the code so far, the next commit will be 76 files changed, 1019 insertions(+), 2175 deletions(-)

ajs6f · 2016-10-05T13:49:28Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+    Instant getCreated();
+
+    /**
+     * Get the date this session expires


We've talked about immortal sessions. Either there should be a note explaining what to use for that "value" or this should be Optional<Instant> (I like the latter).

I updated the FedoraSession interface to return an Optional<Instant>

ajs6f · 2016-10-05T13:49:52Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

+     * @param key the key
+     * @param value the value
+     */
+    void setSessionData(String key, String value);


What about multi-valued keys?

Multi-valued keys aren't supported. Should they be?

As a little background, the reason this method is here is so that the HTTP layers can record the Host and UserAgent headers in such a way that those are available to the eventing machinery: for example: https://git.io/vPZrr and https://git.io/vPZoe. In the current impl, the data is packed into a JSON structure and stored as a String. We could do that here as well (make the signature into setSessionData(String data), but I thought this provided a little more structure which could be relied on more than hoping the JSON string is well formatted, which is what we have now.

I'm thinking about what I am going to do with this, not what you have already done with it: we're writing an API. IOW, I can definitely think of multi-valued use cases, but I'm not asking you to support them, just to write this method's signature in such a way as to avoid disallowing them. E.g.

void addSessionData(String key, String value); Collection<String> getSessionData(String key); void removeSessionData(String key, String value); default removeSessionData(String key) { getSessionData(key).forEach(v -> removeSessionData(key, v)); }

or maybe just commit to Guava's MultiMap

// @return a _mutable_ multimap of session-related data MultiMap<String, String> getSessionData();

ajs6f · 2016-10-05T13:50:42Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraRepositoryImpl.java

+                throw new RepositoryRuntimeException(ex);
+            }
+        }
+        throw new IllegalArgumentException("Credentials are of the wrong type");


Can this be a more specific subtype?

ajs6f · 2016-10-05T13:51:20Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraRepositoryImpl.java

+        if (repository instanceof FedoraRepositoryImpl) {
+            return ((FedoraRepositoryImpl)repository).getJcrRepository();
+        }
+        throw new IllegalArgumentException("FedoraRepository is of the wrong type");


Ditto-- maybe a more specific exception?

ajs6f · 2016-10-05T13:52:53Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+
+    public static final String TIMEOUT_SYSTEM_PROPERTY = "fcrepo.session.timeout";
+
+    private final Session session;


This name is a little confusing: maybe call it jcrSession or something?

ajs6f · 2016-10-05T13:54:01Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+        if (session instanceof FedoraSessionImpl) {
+            return ((FedoraSessionImpl)session).getJcrSession();
+        }
+        throw new IllegalArgumentException("FedoraSession is of the wrong type");


Maybe a more specific exception type?

ajs6f · 2016-10-05T13:59:14Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

+
+    @Override
+    public FedoraSession getSession(final String sessionId, final String username) {
+        final FedoraSession tx = transactions.computeIfAbsent(getTxKey(sessionId, username), s -> {


I like Java 8 but this seems a bit fancy-shmancy. How about

String txKey = getTxKey(sessionId, username); if (!transactions.containsKey(txKey)) { throw new SessionMissingException("Batch session with id: " + txKey + " is not available!"); } return transactions.get(txKey);

The reason for that structure has to do with the fact that another thread may remove the session after a containsKey check. We could, however, do something like:

FedoraSession s = transactions.get(txKey); if (s == null) { throw new SessionMissingException("..."); } return s;

The Javadocs for Map::computeIfAbsent say The default implementation makes no guarantees about synchronization or atomicity properties of this method. so you are relying on ConcurrentHashMaps impl-specific behavior here.... but do we have any understanding anyway of the atomicity of *Service? I don't think this is coherent.

ajs6f · 2016-10-05T14:01:16Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

+
+    @Override
+    public void commit(final String sessionId, final String username) {
+        final FedoraSession session = transactions.remove(getTxKey(sessionId, username));


If the session.commit() fails, don't you not want to have removed it from the transactions? The client might want to retry.

ajs6f · 2016-10-05T14:01:52Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

+
+    @Override
+    public void abort(final String sessionId, final String username) {
+        final FedoraSession tx = transactions.remove(getTxKey(sessionId, username));


Same as above-- if the expire fails, shouldn't the session still be available for further detective work?

ajs6f · 2016-10-05T14:02:35Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

+    }
+
+    private static String getTxKey(final String sessionId, final String username) {
+        return username == null ? sessionId : username + ":" + sessionId;


Maybe it's just me, but I would actually prefer anonymous sessions to have the colon.

acoburn · 2016-10-05T18:55:17Z

@ajs6f I addressed the latest round of comments in that last commit.

ajs6f · 2016-10-05T19:03:28Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/FedoraSession.java

@@ -77,16 +80,31 @@
    Map<String, String> getNamespaces();

    /**
-     * Set session-specific data
+     * Add session-specific data


Can we have a note here that it is up to the impl whether multiple values per key are afforded, and a note on the immediate impl that they aren't?

ajs6f · 2016-10-05T19:04:08Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

    private final String id;
    private final Instant created;
-    private final Map<String, String> sessionData;
+    private final ConcurrentHashMap<String, String> sessionData;


If we aren't claiming that this is thread-safe, what's the point of ConcurrentHashMap?

because if there are multiple concurrent requests in the context of the same transaction, there will be several threads writing to that Map. The question about whether the class is thread-safe at all has to do with how the jcr.Session handles concurrency.

I don't get that. If multiple threads are going to hit a single instance of this class, how is it okay for this class to not be thread-safe?

I think it ought to be thread-safe. I just don't know if the jcr.Session actually is. In the current code, we're treating the JCR Session as if it is thread-safe, so one option is to keep the ConcurrentHashMap (or use a Collections::synchonizedMap) and remove the comment about thread-safety.

https://docs.jboss.org/author/display/MODE/Repository+and+Session?_sscc=t#RepositoryandSession-Threadsafety

Grumble...

I don't think we should make guarantees, and if we don't, we should use the cheapest Map that does the job.

ajs6f · 2016-10-05T19:04:46Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

-            throw new SessionMissingException("Batch session with id: " + s + " is not available");
-        });
-        return tx;
+        System.out.println(transactions.keySet().toString());


Looks like some debugging juice?

acoburn · 2016-10-05T21:32:28Z

@ajs6f more comments addressed in the latest commit.

awoods · 2016-10-12T16:13:26Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+    void begin(FedoraSession session, String username);
+
+    /**
+     * Create a new FedoraSession for the anonymous user and add it to the currently open ones


I am confused by: "Create a new FedoraSession..." since a FedoraSession is being passed into this method.

awoods · 2016-10-12T16:14:15Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param username the name of the {@link java.security.Principal}
+     * @return the {@link FedoraSession} with this user
+     */
+    FedoraSession getSession(String sessionId, String username);


What happens if no such session for the provided username exists?

A SessionMissingException is thrown.

awoods · 2016-10-12T16:16:31Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+    }
+
+    /**
+     * Refresh an existing session using an implementation-defined default


What is the difference between "Refresh an existing session" here and FedoraSession.updateExpiry? https://github.com/fcrepo4/fcrepo4/pull/1107/files#diff-2a974e96c1ebe0bcd554be7d704b6955R48

The BatchService implementation will typically involve an in-memory (possibly distributed) map of FedoraSessions. BatchService::refresh will update the expiry of one of those batch sessions. I.e., BatchService::refresh will probably call FedoraSession::updateExpiry, but as for how that is done is up to the implementation.

awoods · 2016-10-12T16:17:43Z

fcrepo-kernel-api/src/main/java/org/fcrepo/kernel/api/services/BatchService.java

+     * @param sessionId the id of the {@link FedoraSession}
+     * @param username the name of the {@link java.security.Principal}
+     */
+    void commit(String sessionId, String username);


There seems to be some overlap between this interface and FedoraSession. How are we reconciling this commit with FedoraSession.commit?

The BatchService::commit will call FedoraSession::commit, but it may also (depending on the implementation) remove the persistent session from the (distributed) session map.

awoods · 2016-10-12T16:21:45Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java

+     */
+    private static Map<String, FedoraSession> transactions = new ConcurrentHashMap<>();
+
+    public static final long REAP_INTERVAL = 1000;


This can be private.

this is public (I believe) to facilitate testing. I can add the appropriate guava annotation.

awoods · 2016-10-12T16:28:55Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+public class FedoraSessionImpl implements FedoraSession {
+
+    // The default timeout is 3 minutes
+    public static final String DEFAULT_TIMEOUT = Long.toString(ofMinutes(3).toMillis());


Can be private.

same as above...

awoods · 2016-10-12T16:29:01Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+    // The default timeout is 3 minutes
+    public static final String DEFAULT_TIMEOUT = Long.toString(ofMinutes(3).toMillis());
+
+    public static final String TIMEOUT_SYSTEM_PROPERTY = "fcrepo.session.timeout";


Can be private.

same as above...

awoods · 2016-10-12T16:30:57Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+    private static final ObjectMapper mapper = new ObjectMapper();
+
+    /**
+     * Create a Fedora session with a JCR session and a username


This javadoc should probably be updated to remove reference to username.

awoods · 2016-10-12T16:34:57Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+     *  @param key the data key
+     *  @param value the data value
+     *
+     *  Note: while the FedoraSession inteface permits multi-valued


Typos: inteface and assciated

awoods · 2016-10-12T16:36:29Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

+     *  @param key the data key
+     *  @param value the data value
+     *
+     *  Note: while the FedoraSession inteface permits multi-valued


Have you verified that the current codebase does not require appending values to an existing UserData key?

awoods · 2016-10-12T17:14:19Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

@@ -51,8 +52,10 @@
 public class FedoraSessionImpl implements FedoraSession {

    // The default timeout is 3 minutes
+    @VisibleForTesting
    public static final String DEFAULT_TIMEOUT = Long.toString(ofMinutes(3).toMillis());


This is not actually used in any testing code... and can presently be safely private.

awoods · 2016-10-12T17:14:33Z

fcrepo-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/FedoraSessionImpl.java

    public static final String DEFAULT_TIMEOUT = Long.toString(ofMinutes(3).toMillis());

+    @VisibleForTesting


This is not actually used in any testing code... and can presently be safely private.

awoods · 2016-10-12T17:15:05Z

...po-kernel-modeshape/src/main/java/org/fcrepo/kernel/modeshape/services/BatchServiceImpl.java


+    @VisibleForTesting


This is not actually used in any testing code... and can presently be safely private.

awoods · 2016-10-12T17:35:19Z

This is ready to go from my perspective. @ajs6f, have all of your comments been addressed?

ajs6f · 2016-10-13T13:26:59Z

@awoods I have no idea. That's not @acoburn's fault at all-- I find this new Github "review" mode terribly confusing and unhelpful and I can no longer keep track of this conversation. Frankly, if @acoburn gives us his word that he addressed everything, that's fine with me.

acoburn · 2016-10-13T13:30:20Z

And I find that this "review mode" makes it difficult to keep track of issues that I've resolved. As far as I know, I have addressed all of the issues brought up in this discussion, but it has also involved a lot of back and forth, and it is entirely possible that something has fallen between the cracks.

awoods · 2016-10-13T13:42:42Z

It sounds like, to the best of our collective knowledge, this PR is ready to go. Feel free to merge it @ajs6f .

ajs6f · 2016-10-13T13:57:15Z

Am I to close and reopen the ticket?

acoburn · 2016-10-13T13:58:14Z

@ajs6f just reopen the Jira ticket.

ajs6f reviewed Sep 30, 2016

View reviewed changes

awoods reviewed Oct 1, 2016

View reviewed changes

ajs6f reviewed Oct 3, 2016

View reviewed changes

acoburn added 7 commits October 3, 2016 12:49

initial types for replacing jcr.Session in kernel-api

e9c1328

code review

023398c

missed some changes with earlier commit

f0d8bff

code review

7410e8b

add initial implementation

2552b0d

rename methods

629b0de

add support for event-based data

bdec501

ajs6f suggested changes Oct 5, 2016

View reviewed changes

Code review

8e82574

ajs6f reviewed Oct 5, 2016

View reviewed changes

acoburn added 3 commits October 5, 2016 15:25

more code review

4fb0626

remove thread-safety comments

b323e34

more code review

1911fe3

acoburn mentioned this pull request Oct 7, 2016

FCREPO-1868: remove dependency on javax.jcr.Session in kernel-api #968

Closed

awoods reviewed Oct 12, 2016

View reviewed changes

code review

811c8a2

awoods reviewed Oct 12, 2016

View reviewed changes

more code review

eaacdc3

ajs6f approved these changes Oct 13, 2016

View reviewed changes

ajs6f merged commit a11360d into fcrepo:master Oct 13, 2016

acoburn deleted the fcrepo-1868 branch December 27, 2016 21:59


		public interface State {}

		public enum RequiredState implements State {


		public static final String TIMEOUT_SYSTEM_PROPERTY = "fcrepo.session.timeout";

		private final Session session;

		public static final String DEFAULT_TIMEOUT = Long.toString(ofMinutes(3).toMillis());

		@VisibleForTesting

Add core types to be used to replace the use of jcr.Session in the kernel API #1107

Add core types to be used to replace the use of jcr.Session in the kernel API #1107

Conversation

acoburn commented Sep 30, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ajs6f Sep 30, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

acoburn Oct 1, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

awoods Oct 1, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

awoods commented Oct 1, 2016

acoburn commented Oct 2, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

acoburn commented Oct 3, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

acoburn commented Oct 5, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

acoburn commented Oct 5, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

acoburn commented Sep 30, 2016 •

edited

ajs6f Sep 30, 2016 •

edited

acoburn Oct 1, 2016 •

edited

awoods Oct 1, 2016 •

edited

acoburn commented Oct 5, 2016 •

edited