Windows randomly closing

[parahippocampus]

I just wanted to confirm, that the bug is still present in OmniFocus 4.5 (v180.1.7).

Here is another sync log from my Mac mini, from right after the bug had occurred on it:

(Click to expand)

Dec 16, 2024 at 3:28 PM Received Push (4.44s) u:0 d:4
Dec 16, 2024 at 3:22 PM Scheduled (4.92s)
Dec 16, 2024 at 3:11 PM Scheduled (4.47s)
Dec 16, 2024 at 2:56 PM Scheduled (4.57s)
Dec 16, 2024 at 2:42 PM Scheduled (4.52s)
Dec 16, 2024 at 2:31 PM Scheduled (4.66s)
Dec 16, 2024 at 2:18 PM Scheduled (4.58s)
Dec 16, 2024 at 2:08 PM Scheduled (4.70s)
Dec 16, 2024 at 1:54 PM Scheduled (4.51s)
Dec 16, 2024 at 1:41 PM Scheduled (4.47s)
Dec 16, 2024 at 1:27 PM Scheduled (4.43s)
Dec 16, 2024 at 1:16 PM Scheduled (4.62s)
Dec 16, 2024 at 1:05 PM Scheduled (4.52s)
Dec 16, 2024 at 12:51 PM Scheduled (4.60s)
Dec 16, 2024 at 12:36 PM Scheduled (4.79s)
Dec 16, 2024 at 12:22 PM Scheduled (4.49s)
Dec 16, 2024 at 12:09 PM Scheduled (4.60s)
Dec 16, 2024 at 11:59 AM Scheduled (4.44s)
Dec 16, 2024 at 11:46 AM Scheduled (4.53s)
Dec 16, 2024 at 11:31 AM Scheduled (4.52s)
Dec 16, 2024 at 11:19 AM Scheduled (6.55s)
Dec 16, 2024 at 11:07 AM Scheduled (4.52s)
Dec 16, 2024 at 10:55 AM Scheduled (4.59s)
Dec 16, 2024 at 10:45 AM Scheduled (4.45s)
Dec 16, 2024 at 10:33 AM Scheduled (4.50s)
Dec 16, 2024 at 10:20 AM Scheduled (4.44s)
Dec 16, 2024 at 10:08 AM Scheduled (4.68s)
Dec 16, 2024 at 9:55 AM Scheduled (4.81s)
Dec 16, 2024 at 9:43 AM Scheduled (4.38s)
Dec 16, 2024 at 9:28 AM Scheduled (4.46s)
Dec 16, 2024 at 9:16 AM Scheduled (4.33s)
Dec 16, 2024 at 9:03 AM Scheduled (4.82s)
Dec 16, 2024 at 8:53 AM Bonjour (4.52s)
Dec 16, 2024 at 8:45 AM Recent Edit (0.00s)
Dec 16, 2024 at 8:41 AM Scheduled (4.61s)
Dec 16, 2024 at 8:30 AM Scheduled (4.49s)
Dec 16, 2024 at 8:17 AM Scheduled (4.39s)
Dec 16, 2024 at 8:04 AM Scheduled (4.39s)
Dec 16, 2024 at 7:50 AM Scheduled (4.83s)
Dec 16, 2024 at 7:37 AM Bonjour (4.16s)
Dec 16, 2024 at 7:31 AM Scheduled (4.45s)
Dec 16, 2024 at 7:18 AM Scheduled (4.46s)
Dec 16, 2024 at 7:03 AM Scheduled (4.31s)
Dec 16, 2024 at 6:51 AM Scheduled (4.43s)
Dec 16, 2024 at 6:39 AM Scheduled (4.41s)
Dec 16, 2024 at 6:26 AM Scheduled (4.41s)
Dec 16, 2024 at 6:13 AM Scheduled (4.55s)
Dec 16, 2024 at 6:00 AM Recent Edit (0.00s)

The bug happened within a minute after the most recent entry:

• Dec 16, 2024 at 3:28 PM Received Push (4.44s) u:0 d:4

This time, I had exclusively worked in OmniFocus on my other Mac and my Mac mini was simply downloading 4 changes, according to the sync log. This fits with my previous impression, that it’s not really conflicting changes I myself make on multiple devices in parallel, but a problem triggered by sync itself.

If I remember correctly, I had started to edit a task on my other Mac when I was interrupted by a phone call. After the call, OmniFocus was still in edit mode for that task. Very soon after I made another set of changes to that task, I got the email notification that the bug had been triggered on my Mac mini.

[kcase]

@parahippocampus

As the sync log shows, I had recently made a bunch of changes in OmniFocus on the same Mac where the bug occurred (my Mac mini). What I find confusing/suspicious is that the “d:1” in that log line suggests that some changes were downloaded, even though I don’t remember having done anything today in OmniFocus on my iPhone or my other Mac.

I agree, that’s suspicious—and hopefully will help us get to the bottom of this.

So, looking at your iPhone, we see it syncing based on a few push notifications during this editing session on your Mac. And in the middle of that, as you’ve noted, we see it upload one change based on having made a “recent edit”:

Nov 30, 2024 at 5:41 PM Received Push (12.73s) u:0 d:23
Nov 30, 2024 at 5:41 PM Received Push (Postponed)
Nov 30, 2024 at 5:28 PM Recent Edit (1.52s) u:1 d:0
Nov 30, 2024 at 5:21 PM Received Push (9.21s) u:0 d:10

Since you weren’t actively using your iPhone, something about processing those synced changes seems to have triggered it into making an “edit” that wasn’t part of those changes that it uploaded back to the server.

It would be great if we could track down what that “edit” was!

Since all your devices are syncing regularly it’s presumably too late to capture that edit this time. But if this is happening to you regularly, hopefully we can catch it next time!

So, how do we capture that edit? It’s not shared in these exported sync logs, but in the displayed sync log you’ll see a transaction id displayed in square brackets between the sync duration and the upload/download counts, e.g. [doRIK_oWtkZ]. (In the test builds of 4.5.1, I’ve updated the Copy Log functionality to include this transaction id.)

On the Mac, right after this happens, use Help > Anonymize Database… to make a copy of your database structure while eliminating its personal data it might contain. You’ll be prompted to choose a location for the copy of your database: pick a name (or use the default Untitled) and location.

You’ll now have an anonymized copy of your database which you can review by double-clicking on it in Finder. But that’s not going to tell us what changed; to do that, we need to look at the transaction files inside the database. Control-Click on that database in Finder, then choose “Show Package Contents”. Inside, you’ll find a bunch of files named things like 20241201154756=iJlwpBaU6Pi+doRIK_oWtkZ.zip. The start of that filename is a timestamp, followed by a chain of transaction ids separated by + symbols. What we’re looking for is the transaction uploaded by your iPhone: so we’re trying to match up the doRIK_oWtkZ in the sync log with the doRIK_oWtkZ.zip at the end of the filename.

You can double-click on that zip to access its contents, which will probably just be a single contents.xml file. This text file describes the changes made by your other device: you’re welcome to review it yourself, and you’re also welcome to email me a copy at kc@omnigroup.com for analysis.

(Note that you can also look in your real database this way, to find a copy of the same transaction without the anonymization step. For looking at your own data and doing your own analysis, you’ll find that to be more useful since the text in the transaction won’t have been turned into gibberish by the anonymization process.)

Sorry it’s taken so long to track this down. Thank you for helping us get to the bottom of this!

[liz4cps]

When I woke up my mac today, I could see this bug had triggered again while I was away. I decided to look at the sync log. I did not use OF between 7:45am and 2:35pm today, yet I have things like this in the log:

Dec 29, 2024 at 10:05 AM Recent Edit (0.00s) []
Dec 29, 2024 at 11:06 AM Recent Edit (0.00s) []

Note that there is no transaction id in these and they occur hours after I stopped using OF during the period I mention above.

Most of my log does show transaction ids, for example:

Dec 29, 2024 at 10:17 AM Scheduled (3.43s) [fKvrGghUROu]

There is a sync that seems to be in response to my completing a task at 2:35pm:

Dec 29, 2024 at 10:17 AM Scheduled (3.43s) [fKvrGghUROu]

I also have some that look like this, labeled as Bonjour (blast from the past?); this is at a time I was not using OF:

Dec 29, 2024 at 3:05 PM Bonjour (3.70s) [ouoGfviklI_]

After completing the task at 2:35pm, I next use OF to complete two tasks at 4:01pm and I see this which I probably syncing that:

Dec 29, 2024 at 4:03 PM Received Push (5.07s) [bLoH47e2xMb*] u:0 d:2

Unfortunately, I don’t know what time the bug occurred. Do you put something in the log to indicate that? Is there a way I can see older items in the log? I’m noticing the oldest item is only about 10 hours old.

Hoping this helps you figure this out a little.

[liz4cps]

This happened again while I was away from my mac; not planning to post every time this happens, but there were some interesting things in the log and I figured out how to copy the entire log (always had at least one line selected yesterday and didn’t think to deselect then!). Note that I only used OF on my mac and my iPhone, and only on my iPhone when I was away from my mac.

When I used OF since this morning, referring to my logs and Changed perspective in OF for times:

• 7:20am: slept my mac
• 7:43am: marked a task complete on iPhone; last use of OF for hours
• 6:09pm: started using OF again; marked a task complete
• 6:39pm: marked two tasks complete on iPhone
• 7:10pm: woke up mac and saw that the bug had occurred while I was away

Logs are below, including the log from yesterday with comments from me. Note that like yesterday, there are lines labeled “Recent Edit” that have empty []'s; ie, without transaction ids.

Log from Monday 12/30

Dec 30, 2024 at 7:16 PM Recent Edit (1.89s) [] u:1 d:0
Dec 30, 2024 at 7:14 PM Recent Edit (0.00s) []
	7:10pm: woke up mac
Dec 30, 2024 at 6:39 PM Received Push (8.48s) [lLw2YJD0Ef3]
Dec 30, 2024 at 6:39 PM Received Push (3.81s) [lLw2YJD0Ef3*] u:0 d:2
	6:39pm: marked two tasks complete on iPhone
Dec 30, 2024 at 6:38 PM Scheduled (3.39s) [fspYy5QSt-K]
Dec 30, 2024 at 6:09 PM Bonjour (3.52s) [fspYy5QSt-K*] u:0 d:1
Dec 30, 2024 at 6:09 PM Bonjour (3.43s) [hyJd4zgBquC]
	6:09pm: marked a task complete
Dec 30, 2024 at 5:26 PM Scheduled (3.55s) [hyJd4zgBquC]
Dec 30, 2024 at 5:00 PM Recent Edit (0.01s) []
Dec 30, 2024 at 4:33 PM Scheduled (3.40s) [hyJd4zgBquC]
Dec 30, 2024 at 4:23 PM Scheduled (3.54s) [hyJd4zgBquC]
Dec 30, 2024 at 4:12 PM Recent Edit (0.00s) []
Dec 30, 2024 at 4:04 PM Scheduled (3.83s) [hyJd4zgBquC]
Dec 30, 2024 at 3:52 PM Scheduled (3.65s) [hyJd4zgBquC]
Dec 30, 2024 at 3:28 PM Scheduled (5.91s) [hyJd4zgBquC]
Dec 30, 2024 at 2:26 PM Scheduled (3.65s) [hyJd4zgBquC]
Dec 30, 2024 at 1:36 PM Scheduled (3.50s) [hyJd4zgBquC]
Dec 30, 2024 at 1:25 PM Scheduled (3.50s) [hyJd4zgBquC]
Dec 30, 2024 at 1:12 PM Scheduled (3.49s) [hyJd4zgBquC]
Dec 30, 2024 at 1:02 PM Recent Edit (0.00s) []
Dec 30, 2024 at 1:00 PM Scheduled (3.41s) [hyJd4zgBquC]
Dec 30, 2024 at 12:47 PM Scheduled (3.47s) [hyJd4zgBquC]
Dec 30, 2024 at 12:06 PM Scheduled (4.01s) [hyJd4zgBquC]
Dec 30, 2024 at 11:40 AM Scheduled (3.47s) [hyJd4zgBquC]
Dec 30, 2024 at 11:27 AM Scheduled (3.74s) [hyJd4zgBquC]
Dec 30, 2024 at 11:13 AM Scheduled (3.58s) [hyJd4zgBquC]
Dec 30, 2024 at 11:02 AM Scheduled (3.69s) [hyJd4zgBquC]
Dec 30, 2024 at 10:43 AM Scheduled (3.50s) [hyJd4zgBquC]
Dec 30, 2024 at 10:30 AM Scheduled (3.62s) [hyJd4zgBquC]
Dec 30, 2024 at 10:15 AM Recent Edit (0.00s) []
Dec 30, 2024 at 10:03 AM Scheduled (3.51s) [hyJd4zgBquC]
Dec 30, 2024 at 9:51 AM Scheduled (3.95s) [hyJd4zgBquC]
Dec 30, 2024 at 9:22 AM Scheduled (3.41s) [hyJd4zgBquC]
Dec 30, 2024 at 9:11 AM Recent Edit (0.00s) []
Dec 30, 2024 at 9:04 AM Scheduled (6.26s) [hyJd4zgBquC]
Dec 30, 2024 at 8:29 AM Recent Edit (0.00s) []
Dec 30, 2024 at 8:26 AM Scheduled (7.55s) [hyJd4zgBquC] u:1 d:0
Dec 30, 2024 at 7:57 AM Received Push (3.04s) [oFp-GVoH81p…] u:0 d:1
	7:43am: marked a task complete on my iPhone; no immediate sync?
Dec 30, 2024 at 7:38 AM Bonjour (24.69s) [oFp-GVoH81p]
Dec 30, 2024 at 7:38 AM Bonjour (24.69s) [oFp-GVoH81p]
	7:20am: slept my mac
Dec 30, 2024 at 7:19 AM Recent Edit (0.27s) [] u:1 d:0
Dec 30, 2024 at 7:19 AM Recent Edit (0.82s) [] u:1 d:0
	7:19am: completed tasks on my mac
Dec 30, 2024 at 7:14 AM Recent Edit (0.00s) []
Dec 30, 2024 at 7:12 AM Scheduled (3.45s) [d2eNZV6ua8o]
Dec 30, 2024 at 7:00 AM Recent Edit (0.00s) []
Dec 30, 2024 at 6:57 AM Scheduled (3.56s) [d2eNZV6ua8o]
Dec 30, 2024 at 6:47 AM Scheduled (3.49s) [d2eNZV6ua8o]
Dec 30, 2024 at 6:35 AM Scheduled (3.29s) [d2eNZV6ua8o]
Dec 30, 2024 at 6:21 AM Scheduled (7.65s) [d2eNZV6ua8o] u:1 d:0
Dec 30, 2024 at 6:10 AM Scheduled (3.96s) [d2eNZV6ua8o]
Dec 30, 2024 at 5:56 AM Recent Edit (0.84s) [] u:1 d:0
	5:56am: completed a task on my mac
	5:51am: changed tags on a few items using my mac

Log from Sunday 12/29

Dec 29, 2024 at 8:22 PM Scheduled (3.24s) [ilN5FyLrLmQ]
Dec 29, 2024 at 8:12 PM Scheduled (3.33s) [ilN5FyLrLmQ]
Dec 29, 2024 at 7:57 PM Scheduled (3.27s) [ilN5FyLrLmQ]
Dec 29, 2024 at 7:42 PM Scheduled (3.54s) [ilN5FyLrLmQ]
Dec 29, 2024 at 7:28 PM Scheduled (3.42s) [ilN5FyLrLmQ]
Dec 29, 2024 at 7:13 PM Scheduled (3.41s) [ilN5FyLrLmQ]
Dec 29, 2024 at 7:00 PM Recent Edit (0.00s) []
Dec 29, 2024 at 6:54 PM Scheduled (3.46s) [ilN5FyLrLmQ]
Dec 29, 2024 at 6:39 PM Scheduled (3.42s) [ilN5FyLrLmQ]
Dec 29, 2024 at 6:25 PM Scheduled (3.31s) [ilN5FyLrLmQ]
Dec 29, 2024 at 6:14 PM Scheduled (3.46s) [ilN5FyLrLmQ]
Dec 29, 2024 at 6:00 PM Recent Edit (0.00s) []
Dec 29, 2024 at 5:52 PM Scheduled (3.39s) [ilN5FyLrLmQ]
Dec 29, 2024 at 5:40 PM Scheduled (3.29s) [ilN5FyLrLmQ]
Dec 29, 2024 at 5:29 PM Scheduled (5.84s) [ilN5FyLrLmQ] u:1 d:0
Dec 29, 2024 at 5:18 PM Scheduled (3.77s) [ilN5FyLrLmQ]
Dec 29, 2024 at 5:05 PM Recent Edit (0.57s) [] u:1 d:0
Dec 29, 2024 at 5:05 PM Recent Edit (0.96s) [] u:1 d:0
Dec 29, 2024 at 5:00 PM Recent Edit (0.00s) []
Dec 29, 2024 at 4:58 PM Scheduled (3.45s) [bLoH47e2xMb]
Dec 29, 2024 at 4:47 PM Scheduled (3.35s) [bLoH47e2xMb]
Dec 29, 2024 at 4:37 PM Scheduled (5.89s) [bLoH47e2xMb]
Dec 29, 2024 at 4:03 PM Received Push (5.07s) [bLoH47e2xMb*] u:0 d:2
Dec 29, 2024 at 3:40 PM Scheduled (3.67s) [ouoGfviklI_]
Dec 29, 2024 at 3:27 PM Scheduled (3.65s) [ouoGfviklI_]
Dec 29, 2024 at 3:15 PM Scheduled (3.60s) [ouoGfviklI_]
Dec 29, 2024 at 3:05 PM Bonjour (3.70s) [ouoGfviklI_]
Dec 29, 2024 at 3:01 PM Recent Edit (0.00s) []
Dec 29, 2024 at 2:50 PM Scheduled (3.91s) [ouoGfviklI_]
Dec 29, 2024 at 2:36 PM Received Push (4.24s) [ouoGfviklI_*] u:0 d:1
Dec 29, 2024 at 2:33 PM Bonjour (4.13s) [fKvrGghUROu]
Dec 29, 2024 at 2:31 PM Bonjour (3.51s) [fKvrGghUROu]
Dec 29, 2024 at 2:30 PM Bonjour (3.73s) [fKvrGghUROu]
Dec 29, 2024 at 2:21 PM Scheduled (7.05s) [fKvrGghUROu]
Dec 29, 2024 at 1:24 PM Scheduled (4.88s) [fKvrGghUROu]
Dec 29, 2024 at 1:10 PM Scheduled (3.45s) [fKvrGghUROu]
Dec 29, 2024 at 12:59 PM Scheduled (3.48s) [fKvrGghUROu]
Dec 29, 2024 at 12:45 PM Scheduled (3.75s) [fKvrGghUROu]
Dec 29, 2024 at 12:35 PM Scheduled (3.41s) [fKvrGghUROu]
Dec 29, 2024 at 12:11 PM Scheduled (3.53s) [fKvrGghUROu]
Dec 29, 2024 at 11:59 AM Scheduled (3.46s) [fKvrGghUROu]
Dec 29, 2024 at 11:45 AM Scheduled (5.39s) [fKvrGghUROu]
Dec 29, 2024 at 11:30 AM Scheduled (3.41s) [fKvrGghUROu]
Dec 29, 2024 at 11:19 AM Scheduled (3.53s) [fKvrGghUROu]
Dec 29, 2024 at 11:06 AM Recent Edit (0.00s) []
Dec 29, 2024 at 10:56 AM Scheduled (3.46s) [fKvrGghUROu]
Dec 29, 2024 at 10:45 AM Scheduled (3.38s) [fKvrGghUROu]
Dec 29, 2024 at 10:27 AM Scheduled (3.38s) [fKvrGghUROu]
Dec 29, 2024 at 10:17 AM Scheduled (3.43s) [fKvrGghUROu]
Dec 29, 2024 at 10:05 AM Recent Edit (0.00s) []

(edited to fix a typo)

[chris_leonard]

OF Version 4.5 seems to fix the problem I was having with a multi tab window suddenly and randomly closing and immediately re-opening with only a single tab. Not sure how long this seemed stable for, but the problem has clearly returned in OF 4.5.1 (v180.2.2). In fact it seems to happen more frequently than before.

[parahippocampus]

@chris_leonard, the bug was not fixed in OF 4.5 (see my posts above), but you likely had a phase back then when you were, for whatever reason, less likely to trigger the bug.

To illustrate:
I was starting to get optimistic that the issue was fixed in OF 4.5.1, because even 2.5 weeks after updating to that version, the bug had not happened a single time. I was away, however, during that time and pretty much only used OF on a single device. Soon after I was back home, the bug started to happened again.

@kcase:
I just sent you an email (Subject: Logs, etc. for “Windows randomly closing” bug) with sync logs, an anonymized OF database, and a timeline of events leading up to an instance of the bug happening. I hope this helps you figure out the cause of this bug.

[chris_leonard]

I agree with parahippocampus, the apparent stability was due to a lack of triggering the bug. The bug has been quite active this past week. In my case it is closing all the tabs open except the one being currently viewed on my MacOS version of OF 4.5.1 (v180.2.2). It seems triggered after making an entry to the inbox on my iPhone or iPad version. I realize this is difficult yo find this bug but it is very annoying and disruptive to effectively using OF. I hope you find it soon. Thanks

[chris_leonard]

I would also confirm that adding new items via my iPad version or my iPhone version is not sufficient to trigger the bug. I made new iOS app entries several days ago and the tabs on my laptop version have remained open.

[awade1]

If it helps, I have been having this since first installing last year. N00b, so didn’t know about tabs :) . I’ve just opened a tab and since having at least 2 tabs open, the window hasn’t “randomly closed and reopened”. Happened frequently with single view, both in the background and while using the app, reopening in the top left corner of the screen.

I’m weird and like my window on the right, so would have to drag it back into position. So it would be obvious when the window had reloaded.

(macOS 15.2, OF 4.5.2)

[gfholland]

Agree that this bug seemed to disappear for several weeks, agree that its come back. I’ve not noticed any untoward behavior on iPhone nor iPad, but on my Mac (15.3) over the last week or so, I’m frequently (daily?) seeing that all of my OF (4.5.1) tabs (save the top one) have closed since my last viewing.

[rajiv]

I was just hit by this bug. Running OmniFocus/4.5.1/GM-v180.2.2 (with Pro) on macOS 15.3.1, and 4.5.1 Pro (v180.2.2) on iOS 18.3.1. I had 3 tabs open on the Mac. I added an item on my iPhone (via the share sheet from another app; I did not open OmniFocus on the iPhone directly), then 30 minutes later came back to my Mac. On the Mac I used Quick Entry to add a new item. OmniFocus popped up a modal with a spinner saying “Saving to OmniFocus cache”. Then the window reopened with only one tab.

I’ve emailed 2 contents.xml files to support@ and kc@.

[chris_leonard]

After more than a week of stable tabs open on my Mac OS (15.3) version of OF Pro (4.5.1 (v180.2.2), I opened my iPadOS version of OF and was greeted with a msg saying that the content was out of date. I accepted the offer to replace the data on the iPad. I then checked OF on my Mac and only a single tab remained open. As noted above, this is consistent with a problem in database synching.

[kcase]

OmniFocus 4.5.2 is now available with a fix for this issue:

• Sync — Windows and tabs consistently restore when incorporating sync changes, resolving an issue which could unexpectedly close open windows or tabs.

[rajiv]

Excellent. Thank you @kcase!

[parahippocampus]

So, the good news is, OmniFocus 4.5.2 no longer closes all tabs except for one. The bad news is that it still seriously messes with your existing tabs when it perceives a sync conflict. 😕

When the bug struck the first time after I had updated OmniFocus to 4.5.2 on my secondary Mac, I was a bit disheartened, so I did not properly document the screwed up tab/window layout. The order of the tabs was definitely shuffled, though, and I’m 90% sure it also created duplicate tabs of existing tabs. I think tabs were not merged into a single window, but I’m only somewhat sure of that.

To test the new expression of the bug more systematically, I then used focussed dummy projects, so it’s easily to see if the tabs are still in the correct window and in the correct order within each window:

(W1_T1 = first tab in window 1, W1_T2 = second tab in window 1, etc.)

After the bug occurred, all tabs were merged into a single window, and the order was shuffled:

In addition, the contents of each tab also no longer matched its title. In the second screenshot, you can see that it’s actually project “W2_T1” that is focussed, but the title of that tab shows “W2_T3”. The same held true for all the other tabs. Is this maybe a smoking gun?

As a whole, my impression is that OmniFocus stores the information about which tabs are opened in a somewhat fragile manner. This could help explain how the locally opened tabs on one Mac, which are not synced between devices, can get messed up so profoundly by sync (e.g., after just checking off a single repeating task on your iPhone).

@kcase, this new variant of the bug is definitely a bit better than the previous one, but it’s still very disruptive – it’s bit like someone regularly coming into your office and shuffling the paper stacks on your desk. I really hope you can stay on top of this until the bug is fixed. Until it is, OmniFocus on my main Mac is stuck at 4.2 (the last one that did not exhibit this bug), and the idea that you might break sync compatibility with older versions (as discussed in the Omni Roadmap 2025) before this bug is fixed, mildly terrifies me.

(Just in case this might be relevant: In “System Settings → Desktop & Dock”, my Macs are set to “Prefer tabs when opening documents: Always”.)

[Matt78]

I’m not a heavy tab user, and I have found 4.5.2 helped with my windows closing unexpectedly. I’ll open some tabs and test with that. I’d also noticed when the bug hit I’d have to redo my toolbar/sidebar visibility settings. I’m going to pay a little more attention to that – just had to restore the toolbar in my main window. Appreciate all the efforts to resolve this.

[kcase]

@parahippocampus

So, the good news is, OmniFocus 4.5.2 no longer closes all tabs except for one. The bad news is that it still seriously messes with your existing tabs when it perceives a sync conflict.

Thanks for following up on this, and for the additional details. I’m able to reproduce this window merging / tab scrambling if I turn on Prefer tabs when opening database and then use File > Reload Database…. So instead of this being a hard-to-reproduce issue with sync failing to restore tabs and windows, we’re now into problems with the restoration logic that are easy to catalog and reproduce and hopefully much easier to solve!

@Matt78

I’d also noticed when the bug hit I’d have to redo my toolbar/sidebar visibility settings.

I’m able to reproduce toolbar visibility not getting independently restored to each window when using File > Rebuild Database… too, so it looks like that’s another thing our window restoration logic needs to handle better. Thanks for those details!

@parahippocampus

Until it is, OmniFocus on my main Mac is stuck at 4.2 (the last one that did not exhibit this bug), and the idea that you might break sync compatibility with older versions (as discussed in the Omni Roadmap 2025) before this bug is fixed, mildly terrifies me.

My apologies for not being more clear in the roadmap! We’ve worked very hard to make database migrations optional. New features (such as the skipped/dropped tasks and floating time zone support introduced in 3.x) will require a data migration in order to support the new feature, but v4 can still sync with v2 today so long as you keep the database in a v2-compatible format. So you’ll still be able to sync with v4.2, you just won’t be able to use new features that v4.2 doesn’t understand until you’re ready to make a migration that drops that compatibility. And the migration prompt itself will list all your synced devices and indicate which ones are running versions of the app that support the proposed migration.

[parahippocampus]

Thank you so much for your swift reply, on a weekend no less, @kcase – that all sounds very promising!

Here’s one more piece of information that might or might not be helpful:

When trying to adapt my Keyboard Maestro based notification system to the new behavior of the bug, I noticed that OmniFocus sometimes reports an inflated number of windows/tabs, when I use the following AppleScript command:

tell application "OmniFocus" to count windows

For instance, a few minutes ago, I had 2 windows with a total of 5 tabs open in OmniFocus. The AppleScript command, however, returned “11”. After simply closing and reopening OmniFocus (which preserved the visible tab layout perfectly), the command returned the proper number of “5”.

I don’t know much about the internals of macOS window handling, so this might be perfectly normal behavior, but maybe it could also mean that there are invisible “ghost tabs/windows” hanging around that interfere when OmniFocus tries to restore the tab/window layout?

Again, thank you so much for staying on top of this!

[gfholland]

4.5.2 seemed to help … for a while. I’m back to having all-tabs-but-one close. Happened a couple times yesterday during work hours, was greeted with it again this morning. I just now quit and restarted OF - and again, all-but-one tab had closed.

[kcase]

@gfholland

I just now quit and restarted OF - and again, all-but-one tab had closed.

Is it possible that you have “Close windows when quitting an application” turned on in the System Settings app (under Desktop & Dock)? When that system setting is enabled, that’s exactly how all apps are expected to behave: their open documents and windows will not be restored when you reopen the app.