[ jEdit-users ] Re: CodeHelp 0.2/0.3 performance

[Jesper J. <Jes...@ge...>]

I agree, currently SpeedJava is the best plug-in for code completion.
CodeHelp seems very slow even with just rt.jar in the cp. SpeedJava
manages to pop-up fields and methods much faster than CodeHelp without
hindering continuous typing.=20
I couldn't start jedit with CodeHelp 0.3 as apparently my cp was too
long as a shorter cp did fix this start-up problem.
also, I find that the method pop-up does not work correctly in either
SpeedJava or CodeHelp, eg I have jms.jar in he JCompiler cp and neither
plugins pop-ups any methods on the JMS classes/interfaces..
Also, the method pop-ups should work on jar files and source code that
are not jar'ed but in the cp..

For now, I'll use SpeedJava because it's faster until someone was
cracked the code completion and have a fully working plugin that does
method pop-up on all classes/interfaces in the cp..

Jesper



I get the following exception continuously when using codehelp 0.3 with
jedit 3.2.2
If my claspath is small (i.e. only rt.jar) it comes up a few times on
startup and then a few more each time I press a key.

At the moment I have been unable to use it. Not only does the popup
never
appear it takes for ages to release and type the actual character.
At the moment I'm using windose 98 (92Mb ram). I've tried waiting for it
to
parse the jar the first time. Then reloading the buffer and trying again
but
there's no difference If I have my normal classpath I can't even load up
jedit.

I have jcompiler classpath set up as follows...:
C:\jdk1.3\jre\lib\rt.jar;$basepath\src\java

In fact I've never had any success with any of the codehelp versions. Am
I
doing something wrong?

Up to know I happiest with speedjava as a code completion plugin.


[error] java.lang.NullPointerException
[error]         at
codehelp.UrlInfo.findAllClassesRecusiv(UrlInfo.java:161)
[error]         at codehelp.UrlInfo.findAllClasses(UrlInfo.java:155)
[error]         at codehelp.UrlInfo.parseUrl(UrlInfo.java:114)
[error]         at
codehelp.PackageInfoCache.parseAndPutInCache(PackageInfoCache
.java:73)
[error]         at
codehelp.PackageInfoCache.getAllClassesInPackage(PackageInfoC
ache.java:54)
[error]         at
codeaid.popup.PopupKeyHandler.somethingTyped(PopupKeyHandler.
java:158)
[error]         at
codehelp.CodeHelpPlugin$SomethingPopupCommand.execute(CodeHel
pPlugin.java:904)
[error]         at
codehelp.CodeHelpPlugin$CommandRunnable.run(CodeHelpPlugin.ja
va:780)
[error]         at
java.awt.event.InvocationEvent.dispatch(InvocationEvent.java:
154)
[error]         at
java.awt.EventQueue.dispatchEvent(EventQueue.java:317)
[error]         at
java.awt.EventDispatchThread.pumpOneEvent(EventDispatchThread
.java:103)
[error]         at
java.awt.EventDispatchThread.pumpEvents(EventDispatchThread.j
ava:93)
[error]         at
java.awt.EventDispatchThread.run(EventDispatchThread.java:84)



----- Original Message -----
From: Jesper Jorgensen <Jes...@ge...>
To: <jEd...@li...>
Sent: Monday, November 26, 2001 11:08 AM
Subject: [ jEdit-users ] Re: CodeHelp 0.2/0.3 performance


I get a big slow-down every time, not just the first time...
eg. HashMap map =3D new HashMap();
map.

takes too long for CodeHelp to either figure out the field name ("map")
or the methods in this class.
This slow-down occurs every time I re-type "map".
I tried to reduce my cp to only rt.jar which does speed things up a
lot...

Perhaps you should try with a very long classpath to re-produce the
problem.

Jesper

ps. my cp in JCompiler:
c:\iplanet\ias6\ias\apps;c:\iplanet\ias6\ias\lib\java\kadmin.jar;c:\ipla
net\ias6\ias\lib\java\naf.jar;c:\iplanet\ias6\ias\lib\java\toolutil.jar;
c:\iplanet\ias6\ias\lib\java\jdbc20.jar;c:\iplanet\ias6\ias\lib\java\jav
ax.jar;c:\iplanet\ias6\ias\classes\java\xml.jar;c:\iplanet\ias6\ias\clas
ses\java\servlet.jar;c:\iplanet\ias6\ias\lib\java\nasdm60.jar;c:\iplanet
\ias6\ias\usr\java\lib\tools.jar;c:\iplanet\ias6\ias\usr\java\jre\lib\rt
.jar;c:\iplanet\ias6\ias\lib\java\kfcjdk11.jar;c:\iplanet\ias6\ias\lib\j
ava\ktjdk11.jar;d:\dev_one\libs\ant.jar;d:\dev_one\libs\ebilling.jar;d:\
dev_one\libs\httpclient.jar;d:\dev_one\libs\j2ee.jar;d:\dev_one\libs\jak
arta-regexp-1.1.jar;d:\dev_one\libs\jcert.jar;d:\dev_one\libs\jms.jar;d:
\dev_one\libs\jnet.jar;d:\dev_one\libs\jsse.jar;d:\dev_one\libs\ldapfilt
.jar;d:\dev_one\libs\classes12.zip;c:\progra~1\jedit
3.1pre3\jedit.jar;d:\downloads\java\protomatter-1.1.5.jar;d:\downloads\j
ava\jdom-B6.jar;C:\java\junit3.6\junit.jar;e:\g2\libs\jdmk-hack.jar;e:\g
2\libs\jdmkrt.jar;e:\g2;d:\dev_one\libs\protomatter-1.1.5.jar;d:\dev_one
\libs\jdom-B6.jar;D:\java\jdk1.3.1\jre\lib\rt.jar




Hi ,
if you activate the popups the first time during a jedit session,
all jars of your classpath are parsed and a cache is created , which
contains
for every jar(and dir), the class names it contains. (just the name, no
further reflection is used).
This can take a long time. (A few secons in my case).
I dont have a lot of jars in my classpath, but rt.jar, which is quite
big.

If you use the help later, the help should come quite quick, beause the
information comes from the cache.
Can you confirm, that its geting much faster the second time ?? (I get
no waiting time at all here)

I plan to implement a second cache, which contains the fields and
methods for a class
, but only after help for a class was requested the first time.

This is a bit like creating the CodeAid databases on the fly.

On the question about having javaDocs like codeaid does:
I just parsing classes/class-jars, which does not contain any javadoc.

I think, the best for the moment is, to have the possibility to
peacefully have codeAid and  CdeHelp running the same time. So with one
codeStrike you could request the codeAid help (with docs), and if there
is none, do an other code strike and you will get the CodeHelp help
(without docs, but you dont need the db) . But it should not be
complicatet to combine both,because I use a lot of the codeAid code,
th parser, he popup code and so on. But I improved the Codeaid code,
like giving help in the "middle of a method",for example :
'aHashtabe.cont'
gives you "contains" .





Carsten

--
-----------------------------------------------
jEdit Users' List
jEd...@li...
https://lists.sourceforge.net/lists/listinfo/jedit-users


--=20
-----------------------------------------------
jEdit Users' List
jEd...@li...
https://lists.sourceforge.net/lists/listinfo/jedit-users

[prev in list] [next in list] [prev in thread] [next in thread]=20


  Log in / Log out
  Configure Your Environment
  About MARC
  We're Hiring!
  Want to add a list? Tell us about it.
  The AIMS Group