tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig R. McClanahan" <>
Subject Re: TC 4.0.1: classes being cached - requires restart to update
Date Fri, 09 Nov 2001 23:20:44 GMT

On 9 Nov 2001, Brett B. Tofel wrote:

> Date: 09 Nov 2001 16:54:55 EST
> From: Brett B. Tofel <Brett.B.Tofel@Dartmouth.EDU>
> Reply-To: Tomcat Users List <>
> To:
> Subject: Re: TC 4.0.1: classes being cached - requires restart to update
> --- You wrote:
> Could you please try Remy's test case?
> > thanks,
> > brett
> >
> Craig
> --- end of quote ---
> ah, actually i hadn't realized Remy was suggesting that i try what he had as a
> test case, ok.
> I did modify the jspcalendar bean class and did find that the log did show the
> webappClassLoader noting that the resource was modified -- _only after i rebuilt
> the class by hand_ -- saving the source file change was not enough. then
> reloading the jsp reflected the change.

This is a really stupid question, but, are you expecting class reloading
to happen automatically when you save the Java *source* of a class you are

It doesn't work that way -- you *must* compile the class -- because the
class loader that detects the changes is only looking at the class files

> but I change my class and recompile and no such notification to the log, no
> change to the jsp output, until i have the manager reload the context, that
> combo, which is much better than a restart of tomcat, rebuild class by hand,
> reload context does make the change show up.
> I have not noticed that waiting 15 secs helped anything. manual reload did the
> trick.
> Tom Drake had suggested my class had session context, if by that he means is the
> class using session variables, no, but the jsp that is importing the class is,
> so overall the compiled servlet wouldbe using session elements. perhaps this is
> my difficulty. I need to use the session attributes to maintain state though...

The only way to get any further on debugging this is if you can provide a
small test case that we can reproduce.  There is clearly something
different from the way you are doing things than the way we are doing
things (class reloading works for me to as long as I follow all the rules)
-- and we cannot tell what it is just from your descriptions.

> -brett


To unsubscribe:   <>
For additional commands: <>
Troubles with the list: <>

View raw message