Conversation
| try { | ||
| ref = queue.remove(); | ||
| try { | ||
| while(!shutdownLatch.await(100, TimeUnit.MILLISECONDS)) { |
There was a problem hiding this comment.
Do you have a reasoning as to why 100ms?
There was a problem hiding this comment.
Not against a timeout, but 100ms looks way too small.
|
Not really. |
| throw td; | ||
| } catch (Throwable t) { | ||
| t.printStackTrace(); | ||
| } finally { |
There was a problem hiding this comment.
You should release the hatch at the finally block, before setting queue to null:
shutdownLatch.countDown();
|
@jjYBdx4IL I see that you modified the pull request after my comments, just wanted to point out that I'm not a Batik developer nor am I related to the ASF in any way. Thus, my comments do not represent any "official" statement from the Batik developers, the PMC nor the ASF. |
jjYBdx4IL
force-pushed
the
CleanerThreadInterrupt
branch
2 times, most recently
from
8ac91be to
e0f777e
Compare
* de-initialize CleanerThread variables on thread shutdown * don't run CleanerThread as daemon
jjYBdx4IL
force-pushed
the
CleanerThreadInterrupt
branch
from
e0f777e to
1a3a005
Compare
|
Up to Java 18 a workaround existed to stop the thread from outside (eg. application code). This was necessary for re-deploying the app in a JEE environment without having leaks. Now, after Java 21 LTE is mainstream, no more workaround exists (Thread.stop is unsupported). Since this is a real problem for companies and forks exists, may I kindly ask why this won't be included in the original Batik distribution? Can I help getting this integrated? |
I am not a Batik developer but instead the maintainer of one of the forks that you mention: I hope that this PR will not be merged because the patch is incorrect. A BTW my fork has had a correct fix since three years ago :-) |
|
@carlosame since you mentioned you have a fix for this, can you contribute it to this repo so we can fix this 😄? |
I've had another PR opened for 5 years now, in case you missed it. Even if I was confident that the PR would be merged (nope), the fix was already published in EchoSVG with my copyright. As you know, contributing to Batik implies assigning the copyright to the ASF. So there is a copyright issue (it is the same reason why I had to close an unrelated Batik PR as I had put the thing in EchoSVG by that time). My counter-proposal: why don't you migrate to EchoSVG which has loads and loads of improvements over Batik? 😁😇 |
4 participants
proper fix for:
https://issues.apache.org/jira/browse/BATIK-939
instead of using or suppressing interrupts, a simple CountDownLatch is used to check whether the thread should shut down or not.