[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Mark Thomas  changed:

   What|Removed |Added

 CC||benjamin.da...@gmail.com

--- Comment #11 from Mark Thomas  ---
*** Bug 59699 has been marked as a duplicate of this bug. ***

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #10 from Henning Brune  ---
I just verified the improvements made in v8.0.33 in our application: Running on
a current build the execution time on the relevant JSP parts is reduced to
1/10th.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Henning Brune  changed:

   What|Removed |Added

 OS|Linux   |Solaris
   Hardware|PC  |Sun
Version|8.0.18  |8.0.32

--- Comment #9 from Henning Brune  ---
(In reply to Mark Thomas from comment #7)
> I've implemented an improvement for 9.0.x (9.0.0.M4 onwards) and 8.0.x
> (8.0.33 onwards).
> 
> The improvement only works when Tomcat's JSP API implementation is used in
> conjunction with Tomcat's EL implementation.
> 
> In my local testing the time for a single request with a single attribute
> lookup was reduced from ~1ms to ~0.1ms.

Any improvement would be highly welcome. After our migration from Tomcat 7 to
8.0.32 many pages in our applications became more or less unusable. We make
heavy use of EL and custom tags and would have to rebuild large parts of our
applications, if the performance would stay on the current level.

Is there any information about the release date for 8.0.33?

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #8 from Konstantin Kolinko  ---
For reference: recent threads:
"Intermittent ClassNotFoundException in Jasper EL evaluation"
http://tomcat.markmail.org/thread/4a4x4ms63hrkpoix

"r1734418"
http://tomcat.markmail.org/thread/fdacbeusrgvlqd2w

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Mark Thomas  changed:

   What|Removed |Added

 Resolution|--- |FIXED
 Status|REOPENED|RESOLVED

--- Comment #7 from Mark Thomas  ---
I've implemented an improvement for 9.0.x (9.0.0.M4 onwards) and 8.0.x (8.0.33
onwards).

The improvement only works when Tomcat's JSP API implementation is used in
conjunction with Tomcat's EL implementation.

In my local testing the time for a single request with a single attribute
lookup was reduced from ~1ms to ~0.1ms.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Mark Thomas  changed:

   What|Removed |Added

 Status|RESOLVED|REOPENED
 Resolution|FIXED   |---

--- Comment #6 from Mark Thomas  ---
I think I have a way to address the performance issues without breaking the
spec.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #3 from Arne Bultman  ---
Maybe I'm missing something, but even with this cache we still have this
problem. Since a en ELContextImpl seems to be created with every jsp render,
the cache has very limited effect, we're still seeing dramatically slower
rendertimes with this implementation. Am I missing something?

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #5 from Arne Bultman  ---
Isn't that lookup relatively static, as in could the caching of lookups not be
shared across instances of ImportHandler rather than having a cache per
instance?

I need to read up on the spec more.

Thanks for your help, so far. Will look into adding the scopes.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #4 from Mark Thomas  ---
What you are missing is that with this change in the EL spec resolving of
unscoped attributes got a lot slower. We have fixed it as best we can and I am
not aware of any further improvements that could be made while staying spec
complaint. Suggestions for improvements are welcome but they do need to
maintain spec compliance.

The work-around is to add the appropriate scope to your attributes so use
${requestScope.foo}, ${applicationScope.foo} etc. rather than ${foo}.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Mark Thomas ma...@apache.org changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED

--- Comment #2 from Mark Thomas ma...@apache.org ---
Fixed in trunk and 8.0.x for 8.0.21 onwards.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

--- Comment #1 from Mark Thomas ma...@apache.org ---
In theory, you should be safe swapping API implementation classes but it is
likely you will see different behaviour as different implmentations will have
fixed different bugs. Personally, I wouldn't swap implementations. I'd rather
stick with a known problem than risk an untested, unsupported combination.

From a Tomcat PoV we are normally only going to support folks that are using
the Tomcat API implementations.

In this case the reason you see better performance is that Glassfish hasn't
added the required support for imports in the version you are using. Looking at
the Glassfish source it appears that this is partially fixed (statics don't
appear to be supported yet) in trunk which should be in the next release.

In terms of solving the performance problem I think we need to cache Not
Found results as well. I'll take a look.

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org

[Bug 57583] Performance issue evaluating EL in custom tags (tagx) due to inefficient calls to java.lang.Class.forName()

[bugzilla]

https://bz.apache.org/bugzilla/show_bug.cgi?id=57583

Yevgeny V jevge...@gmail.com changed:

   What|Removed |Added

 CC||jevge...@gmail.com

-- 
You are receiving this mail because:
You are the assignee for the bug.

-
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org