"Confirm" button in instant review mode broken

[Daniel Kulesz]

Steps to reproduce:

- Start RevAger 1.1 and create a new review session, then save it.
- Change some preferences: disable auto-updates and auto-saving
- Quit RevAger 1.1
- Start RevAger r28/trunk
- Agree to import old settings
- Enter "Instant Review" mode
- Try to proceed without an attendee (you will get an error message)
- Enter the attendee's name
- Try to proceed (The error message will still be in place)

See the attached Video (you need an ogg player to play it) for a quick demonstration.

[Johannes Wettinger (jojow)]

Daniel, you're really able to find every bug! ;-)

[Johannes Wettinger (jojow)]

I think it is a permission problem for Unix-based systems. Do you this problem on Windows systems as well?

[Johannes Wettinger (jojow)]

I've just tried to reproduce this bug, but all is fine! Used system:

Ubuntu 9.10 karmic, 2.6.31-16-generic
java version "1.6.0_0"
OpenJDK Runtime Environment (IcedTea6 1.6.1) (6b16-1.6.1-3ubuntu1)
OpenJDK Client VM (build 14.0-b16, mixed mode, sharing)

I'm going to test it on Windows XP as well.

Can someone else confirm this bug on any system?

[Johannes Wettinger (jojow)]

For Windows XP with Sun JDK 1.6.0_17 it's working fine. I cannot reproduce the bug on this system as well. So I think this bug is not really critical, because it seems to be present only under special circumstances. So this bug is not a "release blocker" for 1.2 IMHO ;-)

@Daniel: Could you please try to do the following, after the old settings are copied to the new location by RevAger:
* Change into the directory where the ".revager" dir is located.
* Do a "chmod 777 -R .revager" and try it again.

[Holger Röder (roederhr)]

I couldn't reproduce this bug with a 'clean' install of RevAger trunk/r28 on MacOS X 10.6.2 (Java 1.6.0_17).

@Daniel: Does this bug only apply to 'upgrade installations' (i.e. installations that use previous versions' data directories)?

[Johannes Wettinger (jojow)]

I've also tested it for an 'upgrade installation', but there was no problem at all.

[Johannes Wettinger (jojow)]

Have tested it now on a third system:

Macbook x86, Ubuntu 9.10 karmic, kernel 2.6.31-15-generic
java version "1.6.0_16" (Sun JDK 6)
Java(TM) SE Runtime Environment (build 1.6.0_16-b01)
Java HotSpot(TM) Server VM (build 14.2-b01, mixed mode)

There's also no chance to reproduce the bug.

[Daniel Kulesz (kuleszdl)]

Actually after re-trying this once more, I was not able to reproduce this either. Strange...I remember I had the old 1.1-instance of RevAger running at the same moment. Maybe the auto-save feature could have corrupted the import of the settings into the 1.2-instance?

[Davide Casciato (casciade)]

I have tested it on:

Sony Vaio VGN-NS11S, Windows Vista Home Premium SP2, java 1.6.0_17 (Sun JDK 6)

I also couldn't reproduce this bug.

[Daniel Kulesz (kuleszdl)]

Finally I was able to reproduce it as well. I started off with the old revager-data dir already, without recreating it in 1.1.

Here is some (not complete) output I captured on the console, after this problem occured.

[Daniel Kulesz (kuleszdl)]

It seems that the trick is to really make use of the "back" button AFTER entering the attendee's name. Here's are the updated reproduction instructions:

- Start RevAger r30/trunk
- (no need to import 1.1 data, works with plain 1.2 as well)
- Enter "Instant Review" mode
- Try to proceed without an attendee (you will get an error message)
- Enter the attendee's name
- Hit the back botton
- Hit the confirm button
- Enter the attendee's name (since the textfield has been somehow erased?)
- Try to proceed (The error message will still be in place)

[Davide Casciato (casciade)]

I've detected the problem. After some changes the bug doesn't occurs any more in my local project. Shall i commit my changes, because the trunk is still frozen?!

[Daniel Kulesz (kuleszdl)]

Depends. I think, once we haven't clarified the policy just commit it directly into the trunk.

[Johannes Wettinger (jojow)]

@Davide: Yes, commit the change please! ;-) After that the trunk branch is frozen again.

[Davide Casciato (casciade)]

Ok, it's done. Sorry, i wasn't at home till now.

[Johannes Wettinger (jojow)]

Thanks Davide! I think it should be clear that we're goin' to rewrite the complete Assistant dialog for the next release. ;-)

[Johannes Wettinger (jojow)]

PS #17: "next release" in this case: "next major release" = 1.3

[Daniel Kulesz (kuleszdl)]

The fix seems to work - but occasionally I get a garbled screen, see the attached screenshot. (I am not sure it's related to thix fix, but some more investigation could help)

[Johannes Wettinger (jojow)]

@Daniel: Is this bug reproducible somehow?

[Daniel Kulesz (kuleszdl)]

Well I wished it would be :-) I succeeded in reproducing it twice by just navigating around in the wizard. Unfortunately, it does occur only occasionally. Maybe a threading issue?

[Johannes Wettinger (jojow)]

In the past we had similar problems with Swing (in combination with threading by SwingWorker classes), before we structured the thread handling. I think it's very difficult to detect the real source for this kind of bugs.

[Daniel Kulesz (kuleszdl)]

Let's continue with the threading issues here, since it seems to be independent from the original bug:

https://bugs.launchpad.net/revager/+bug/499309