"4.5 RC IV" - for real?
Are these real release candidates? I don't think it's ready for release, there are several stopship bugs for me:
- Cannot use custom components in UI designer palette in 1.5 mode (without running IDEA under JDK5.0)
(http://intellij.net/tracker/idea/viewSCR?publicId=36370)
- Cannot use bytecode instrumentation in UI designer in 1.5 mode no matter what
(http://intellij.net/tracker/idea/viewSCR?publicId=36279)
Some other important bugs for me, but maybe not blocking release of 4.5:
- "Replace for loop with foreach loop" quick fix always produces uncompilable code for Iterator loops
(http://intellij.net/tracker/idea/viewSCR?publicId=36452)
- Problems with UI designer and "maximum size":
(http://intellij.net/tracker/idea/viewSCR?publicId=36427,
http://intellij.net/tracker/idea/viewSCR?publicId=28922)
- Missing components in default UI Designer palette
(http://intellij.net/tracker/idea/viewSCR?publicId=35304)
I hope at least the two bugs I said were stopship are fixed before 4.5 is released, otherwise developing with the UI designer in JDK5.0 is not so much fun.
请先登录再写评论。
I agree that this thing is far from ready for release. I hope that at least IntelliJ continues the EAP program actively, right after the release is out. I've reported several glaring CVS and other issues which, while probably not at the stop-ship level, they are irritating. I wish I could get my hands on the CVS plugin code to fix it myself.
see:
35452 CVS history date sort wrong
35619 CVS Commit file, tag commited files not working
34614 "ant" ant task shows errors incorrectly when using a property for the "dir" attribute
Keith Lea wrote:
I suppose you do have an idea how this should be fixed, don't you?
For those who haven't seen the stacktrace the problem is as follows: IDEA tries to load custom component class into its
own JVM (running under 1.4 JRE) in order to show up it inside the form and gather the avalibale properties etc. Of
course it fails since class itself have been compiled with 1.5 compiler and reports class file version that couldn't be
tracked by 1.4 JRE.
So, the only workaround I can suggest is to compile your custom components with target 1.4.
--
Maxim Shafirov
IntelliJ Labs / JetBrains Inc.
http://www.intellij.com
"Develop with pleasure!"
I have a few ideas for fixes.
1. Integrate RetroWeaver (http://retroweaver.sourceforge.net/) to attempt to transform the classfiles from 5.0 to 1.4
2. Don't load and render custom components if an exceptions happens while loading (just display as an empty panel, this is better than not being able to use them)
3. Automatically attempt compile UI components with target 1.4 and store in a secret temporary folder, if being used as custom components.
Personally I like #1, with a fallback on #2 (because some classes can't be transformed because they use methods only available in JDK 5.0). Perhaps if a classfile is transformed with RetroWeaver, a little icon could appear in the upper left corner of the component in the UI designer, and when you mouse over it, a tooltip says "This component may not look correct because it has been converted to a Java 1.4 class file so it can be previewed in the UI designer."
I think that integrating RetroWeaver would take much more effort than simply displaying an empty panel if an exception occurred when loading or displaying the component. So, I think for now, for 4.5 release, #2 should be implemented. This solution would also fix SCR 35326 (http://intellij.net/tracker/idea/viewSCR?publicId=35326 ).
6 of my J2EE bugs are marked open when in fact they've been fixed, so that's not a bad complaint I suppose.
My current favourite stopshopper though has to be the 'flash once per ejb module' bug:
http://www.intellij.net/tracker/idea/viewSCR?publicId=29724
This bug actually went away for a while, but is now back (and before anyone feels smug that it's an OSX only bug, it now happens on all platforms).
Since the first 2000 build I have been switching between two filters in the tracker; one that lists the issues reported specifically against the latest released build and another that lists the fixed issues in later builds soon to be released.
Only once did a 2000-series build solve more problems than it introduced; that in my opinion is also an indication that the product is fairly far from being ready for a commercial release.
Shouldn't there be at least several EAP build releases that solve the majority of reported issues without introducing a bunch of new ones?
I have followed this EAP (used the builds) but not participated much because the apps I currently work on are so simple; the EAP builds have worked fine for this, but I think JetBrains owe it to the many truly involved EAP users to use the abundance of gathered information about issues to really fine-tune and polish IDEA 4.5 before it's released.
Continuing to have 6-7000 issues still to be fixed, 3-4000 open issues, or openly postponing currently 77 bugs for IDEA 5.0 - some of which originate from the 4.0/4.1 versions, could damage JetBrains' reputation in the long run.
It may not be considered proper to invite people to get involved in an EAP only to take their reported issues and fix them in a 5.0 release that - I'm guessing, correct me if I'm wrong - will require these users to pay for the upgrade from their 4.x license.
IDEA continue to be my favorite but my bosses would be happy if they could find some good reason for forcing my colleagues and I to go with some unspeakable free-of-charge, open-source product. Now looking at Perforce (after having recently gone from ClearCase to CVS) they may have their ammunition, since one argument for IDEA was its good CVS integration, and since there is no Perforce integration in sight, but bad press about upcoming releases could also endanger the continued renewal of my IDEA license.
Best regards,
Lars Ugleberg
I agree with you Lars, though I have to admit it'd take some pretty severe bugs to force me off IDEA.
I would actually stress your point even more and say that the 4.0.4 release has been in EAP now for around 2 months with no further updates and for all I can see no reason it hasn't gone stable.
I honestly can't understand how JetBrains have not already released this as a stable build on their main site as I'm sure they have MANY customers who never even look at EAP but have been waiting for the Tomcat 5 bug to be fixed for the past 2 months which has been fixed in the EAP build for that long without ever being released.
It just seems totally ridiculous to me as there will be no reason to ever release that build after 4.5 comes out (everybody will deviate from the 4.0 builds to 4.5 naturally) meaning that all the work they did to fix the Tomcat 5 integration is effectively wasted as most of their development community will have been waiting for a stable release build.
Here's a bug I think should be stopship which hasn't gotten enough
attention yet:
"Introduce Constant" and "Introduce Field" insert declaration below usage
http://www.intellij.net/tracker/idea/viewSCR?publicId=36251
It's a regression of this bug:
http://www.intellij.net/tracker/idea/viewSCR?publicId=10014
which was once stopship for Ariadna.
Bas
The ultimate problem for me still is that IDEA eats up my memory. I just had to shutdown and restart again after not being able to drop memory usage below 200 MB even with dozens of clicks on the GC icon. I will have to go back to 4.0.x soon if this does not get fixed.
Is there a ticket number for that?
Michael Descher wrote:
>The ultimate problem for me still is that IDEA eats up my memory. I just had to shutdown and restart again after not being able to drop memory usage below 200 MB even with dozens of clicks on the GC icon. I will have to go back to 4.0.x soon if this does not get fixed.
>
>
http://www.intellij.net/tracker/idea/viewSCR?publicId=35754
http://www.intellij.net/forums/thread.jsp?forum=22&thread=100792
http://www.intellij.net/forums/thread.jsp?forum=22&thread=102651
Thx.
Michael Descher wrote:
>http://www.intellij.net/tracker/idea/viewSCR?publicId=35754
>
>http://www.intellij.net/forums/thread.jsp?forum=22&thread=100792
>
>http://www.intellij.net/forums/thread.jsp?forum=22&thread=102651
>
>
Michael,
You will make as a favor if you download the YourKit profiler
(http://www.yourkit.com/home/index.jsp), run IDEA under it, take memory
snapshot, zip it and put somewhere so we can download it. We'll take a look
at your snapshot asap.
PS Don't forget to mention the build number.
--
Best regards,
Mike Aizatsky.
-
JetBrains, Inc
http://www.intellij.com
"Develop with pleasure!"
Hi Mike,
you can download the snapshot at
http://descher.codewave.net/idea2206_YourKit.zip
The snapshots are - as the name suggests - from build 2206 running on Windows 2000. I included two snapshots. The first one (IDEAAfterStart.memory) was taken right after starting IDEA and opening the project. The second one (IDEA-3.memory) was taken after some hours when I could not garbage collect below 200 MB. So you should be able to compare the snapshots with YourKit Profiler.
Please tell me when you got the file, so I can delete it from the server.
Michael
Hi,
Thank you very much, we found memory leak during j2ee build with your
snapshots. The latters are no longer needed.
Michael Descher wrote:
--
Best regards,
Maxim Mossienko
IntelliJ Labs / JetBrains Inc.
http://www.intellij.com
"Develop with pleasure!"
Could you tell us which build number will include the fix?
Michael Descher wrote:
>= 2219
--
Maxim Shafirov
IntelliJ Labs / JetBrains Inc.
http://www.intellij.com
"Develop with pleasure!"