This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Summary: | Tomcat 4.0 Server module has ModuleInstall performance degradation | ||
---|---|---|---|
Product: | serverplugins | Reporter: | fomenko <fomenko> |
Component: | Tomcat | Assignee: | Damian Frach <dfrach> |
Status: | RESOLVED WONTFIX | ||
Severity: | blocker | ||
Priority: | P1 | ||
Version: | -FFJ- | ||
Hardware: | Sun | ||
OS: | Solaris | ||
Issue Type: | DEFECT | Exception Reporter: |
Description
fomenko
2002-03-29 04:15:59 UTC
The reason is that we need to copy part of tomcat instalation into userdir because in general user should not access any file in IDE instalation. The question is whether this can be done lazily (other modules can built on top of our stuff). Also this performance regression affects only first start. Next opening doesn't perform such huge operation. Tomcat 4 plugin is not part of NetBeans yet. this problem was caused by fixing 4621234. and we expected performance regression. the second and other starts of IDE should be faster then first, but will be longer then 28 ms. I did today some other performance improvements for second and other restarts. I discussed the solution with Jasse, but this fix is only possible. Jasse uses similiar technik, when he copies sample-dir application into the user-home during first start. But he does it during sample-dir FS mounting, so you can not measure it as his module startup time. But it is part of startup time, because FSs are mounted during start up. So I do not see any possibility, how to reduce this performance regression. So, I think, this bug should be closed as "will not fix". If you do not agree, reopen please this bug and add your comments. I measured method "restored" in TomcatModuleInstall object. I got these data: - fresh IDE 1.start 625 ms 2.start 156 ms 3.start 109 ms 4.start 110 ms - again fresh IDE 1.start 672 ms 2.start 110 ms 3.start 94 ms 4.start 110 ms So second and other starts should be about 5-6 times faster then first. |