ISPN-3400 Upgrade to JSR-107 (JCache) 1.0.0-PFD version #2187
Conversation
galderz
commented
Nov 1, 2013
galderz
commented
- Upgraded to Public Final Draft version of the JSR-107 specification.
- TCK has expanded loadAll() and expiry policy usage tests, which in turn has required more complex logic to deal with all the expectations in the TCK. For example, the logic for loadAll() is now different based on whether a JCache cache loader has been configured, or the Infinispan configuration has a native cache loader configured.
- Tweaked a test util method to provide more generic friendly returns that avoids the need for cast.
|
This needs some serious rebasing... |
|
Rebased and tests are passing :) |
| @@ -246,7 +244,7 @@ private boolean shouldInvoke(Object event, boolean isLocalNodePrimaryOwner) { | |||
| private Throwable getRealException(Throwable re) { | |||
| if (re.getCause() == null) return re; | |||
| Throwable cause = re.getCause(); | |||
| if (cause instanceof CacheException || cause instanceof RuntimeException) | |||
| if (cause instanceof CacheException || cause instanceof RuntimeException || cause instanceof Error) | |||
There was a problem hiding this comment.
CacheException instanceOf RuntimeException, I guess you can simplify here.
There was a problem hiding this comment.
Yeah, it can be simplified.
| try { | ||
| return policy.getExpiryForCreation(); | ||
| } catch (Throwable t) { | ||
| return Duration.ETERNAL; |
There was a problem hiding this comment.
Why Duration.Eternal here and null in other cases?
There was a problem hiding this comment.
It should probably be Eternal in all, let me try...
There was a problem hiding this comment.
I've opened an issue in the TCK. The TCK expects that if access or modified expiration callbacks throw exceptions, the entries are left untouched (as opposed to what happens with creation time), so the code just matches what the TCK expects here. Let's see how the issue is resolved and change afterwards if necessary.
|
This is a big freaking PR. I've tried to look for fishy things from the outset as much as possible. Hope you found it helpful Galder |
|
Updated PR addressing comments made. |
|
I've rebased now. |
|
Mircea would you please integrate this PR? I don't have a fresh infinispan source tree as I have spent most of the time on hrcpp. Cheers |
* Upgraded to Public Final Draft version of the JSR-107 specification. * TCK has expanded loadAll() and expiry policy usage tests, which in turn has required more complex logic to deal with all the expectations in the TCK. For example, the logic for loadAll() is now different based on whether a JCache cache loader has been configured, or the Infinispan configuration has a native cache loader configured. * Tweaked a test util method to provide more generic friendly returns that avoids the need for cast.
|
It's rebased again now. |
|
@vblagoje sure I will :-) |
|
@galderz integrated, thanks! |
