[
https://issues.jboss.org/browse/JBIDE-19820?page=com.atlassian.jira.plugi...
]
Viacheslav Kabanovich commented on JBIDE-19820:
-----------------------------------------------
Unfortunately, it did not work. It turns out, that JDT does most computations at the very
first call to IType, results are cached so that next calls are much faster. For
experiment, I left only call IType.isInterface() in a check for artifact, and removed
everything else. The scanning time reduced only by 15%. Then I measured time of scanning
for CDI (that does not need looking into super types). The scanning time is the same 15%
less than Batch, which means that once you call IType it does not make much difference how
much you query it after that. No patch, sorry.
Batch feature makes wildfly import way slower
---------------------------------------------
Key: JBIDE-19820
URL:
https://issues.jboss.org/browse/JBIDE-19820
Project: Tools (JBoss Tools)
Issue Type: Bug
Components: batch
Affects Versions: 4.3.0.Alpha2
Reporter: Rastislav Wagner
Assignee: Viacheslav Kabanovich
Priority: Critical
Fix For: 4.3.0.Beta1
I made a comparison between Eclipse M6 and Eclipse M6 with batch feature installed while
importing wildfly. WF is imported as maven project, maven is set to offline mode and local
repository contains all needed dependencies.
I did a 4 runs with following results:
M6 was able to import WF 628,575,577,612 (in seconds) = average 598s
M6 with batch feature: 843,832,924,877 (in seconds) = average 869s
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)