'm m' and Martin both bring up points that are spot on. It would be nice to see a reply from Atlassian addressing these lingering concerns.

What do we have to expect for the future? Will the issue remain closed?

@David Richardson Thanks for a good summary of the isses. I see the same 800lb gorilla, which is the reason I react so strongly agains how Atlassian has handled 1.9. We rely on a lot of Atlassian products, and I expect them to be developed using "sober, cautious an proven methods". Lets hope this Jira will be reopened until it is fixed properly.

• full tree expands without user action sometimes (seemingly random)

it's called feedback: if someone does care there, maybe it not "just venting emotions" and will be taken into account.

Please add this to the fixes

• file name filter is missing

@mm Yup, I use it heavily every day too. For now we still have 1.8.1. There's really not much point in the speculation though, other then just venting emotions.

Please re-open and fix it till the end:

• Checkmarks are too error-prone, one needs to click to expand/collapse the tree; please remove, accidental staging all the way (and staged files don't show when committing? perhaps a different bug ..)
• spinning loader all the time

@jeremiahsmall
it's hard not to comment on a broader scope for people who used to rely on the product for every day work.. don't know anyone around who wanted ui changes in SourceTree, or anyone who is happy with these updates now. it's not a single decision: many things got worse, looks like random changes for kicks all over.. what needed to be done is performance, but I don't feel any improvement .

@David Richardson No need to be that dramatic, IMHO. I don't see how you can speculate about such broad issues based on this single bad decision. I'd stick with 1.8.1 for now, and see how things shape up. Atlassian has a great track record of responsiveness, and frankly, while I'm not pleased with 1.9.x so far, I am pleased with how quickly we have some of the removed functionality back already. If it were my decision, I'd ask them to roll back to 1.8.1, because I think the introduction of the checkbox convention is a serious mistake and became the tail that wagged the dog here. That said, there's no rational way to extrapolated this issue to the underlying Git functionality or the assertion that you are no longer in the target constituency for SourceTree, or that a "Windows/Swing GUI designer with no experience using either a version control system or OS X has been let loose." (!?) Those are just trolling comments.

@Kevin Powick. "there are quite a few people on the outside wondering just what the hell has gone wrong". The 800 lb gorilla in this room is - while Atlassian is asking us to entrust them with priceless assets, they seem to be saying, "We've heard that some users who are inexperienced in this field are confused by standard tools and procedures. We're going to forget about sober, cautious and proven methods of managing those assets in favour of putting an unproven abstraction layer over them. We're taking our lead from people who don't know what they're doing." Not an encouragement for trusting the must-not-ever-fail services.

@Steve Streeting. If 1.9.3 is a way of saying "We messed up, here's proof we're working on it" that's fine. Do realize, though, that 1.9.* is alpha-quality at best and should be released as such. A prominent download link to 1.8.1 would support such a perspective.
With the spotlight now on this abstraction layer over the Git stage, and remembering the battle in early versions to prevent memory leaks, I'm now wondering if the underlying Git functionality remains solid?

I agree with @Jeremiahsmall. The idea of a checkbox initiating any action, let alone causing the checked GUI element to move, conforms to no UI standard that I can think of. Checkboxes usually (always?) give a user the opportunity to select/deselect a number of items from a list for which some additional, user-initiated, action will take place.

@David Richardson. The multiple column view has returned as well, but it has its own problems. I see comments in the 1.9.3 blog posting about it, and it looks like Steve S. plans on addressing the issues.

Having said that, I agree that something strange has happened at Atlassian. I don't think there's anything nefarious going on, but it seems to me that Steve S. is now merely a spokesman for a team which perhaps he no longer controls. Maybe this is only speculation, but a SourceTree blog post in February stated that Atlassian is looking for a development team lead for the product.
http://blog.sourcetreeapp.com/2014/02/26/were-hiring-a-sourcetree-development-team-lead/

Of course, only those inside Atlassian know the truth but, IMO, there are quite a few people on the outside wondering just what the hell has gone wrong. Just follow the debacle of the BitBucket website upgrade (no warning in the middle of workday) for another example of implementing wide-sweeping changes that seem to emphasize form over function.