Colin

> In preparation for Squeak 3.9 going gamma, I'm getting together a  
> release of OmniBrowser.

Excellent.
Colin for the fun you should look at the latest implementation of the  
DualPaneBrowser made by roel.
This is cool you have the VW behavior for showing packaging and class  
hierarchy + class extension per package.
Really neat. I think that the conclusion is that knowing the selected  
item would be quite useful for all kinds of new
ideas. :) But this is for the future.

> This will be a bit tricky because of the of some of the changes  
> made in 3.9. I think we'll eventually be able to create the right  
> abstractions to handle them, but for now, it means OB mainline will  
> run on Squeak 3.6 through 3.8, and there will be a special branch  
> for Squeak 3.9.
>
> Here's a list of issues that have come up recently:
>
> --
> Pascal Zumkehr reported Mantis issue #4023, and I've fixed it in OB- 
> Standard-cwp.108.
>
> A bug that prevented OB from opening properly in response to cmd-b.  
> Lukas Renggli did a fix for this, and I've merged it into the  
> trunk. (OB-Standard-cwp.109)
>
> Alexandre Bergel proposed an enhancement to the 'find class...'  
> menu item. It turns out that his enhancement shouldn't have been  
> necessary. The behavior he wanted was intended in the original  
> implementation, but it didn't work because of a bug. Rather than  
> use his enhancement code, I just went back and fixed the original  
> bug. (OB-Standard-cwp.109)
>
> Philippe Marschall (I think... pmm?) created a version of 'find  
> class...' that finds traits as well as classes. This can't go into  
> the OB trunk, but it should be in the Squeak 3.9, I think.
>
> There was a bug that prevented comments from being displayed  
> properly in Squeak 3.9. Again, this can't go into the OB trunk, but  
> should be in Squeak 3.9.
>
> Alexandre Bergel posted an action for moving a method from one  
> package to another. I think it's a great idea, but there's a  
> problem with the implementation: it introduces a dependency on  
> Monticello, which I want to avoid in the core browser. This idea  
> could just as well be implemented via PackageInfo.
>
> Alex also posted a changeset  that causes the '-- all --' method  
> category to be selected when the user clicks on the class. Again, a  
> good idea, but the implementation is problematic. It completely by- 
> passes the metagraph, and makes all kinds of assumptions about what  
> nodes are in which column. It also makes a lot of changes to the  
> core library for the sake of this menu item, which makes me  
> nervous. I think we should delay this feature until it can be done  
> cleanly, post Squeak 3.9.
>
> Lukas fixed a bug which called infinite recursion during a drag  
> operation. As it happens, I also found and fixed this bug, so  
> there's no need to merge his fix into the trunk.
>
> Lukas also posted a cosmetic fix, which makes OB browsers visually  
> match the new look in Squeak 3.9. Obviously not appropriate for the  
> trunk, but should be in the 3.9 version.
>
> Finally, I've done a lot of work on the core library. One of the  
> more important changes was to move to announcements for responding  
> to events. I've also done some preliminary work to support  
> services. I think these changes should be part of the 3.9 release  
> as well.
> --
>
> My plan is to produce two release candidates incorporating all this  
> work - one for inclusion in Squeak 3.9 and one for 3.6 - 3.8. If  
> all goes well, I'll post these to the Squeak 3.9 submission  
> repository tomorrow. If I've forgotten anything, or made some kind  
> of misjudgement about on of these items, please let me know.

Excellent.
Let me know me when you have something for 3.9.
Ideally I would like to remove the OB packages and be able to reload  
it later.