Is it a known issue that macOS 11.4b2 broke again Network Extensions (like 11.3 betas did)?
Post
Replies
Boosts
Views
Activity
Is there a known bug/limitation with the SecCodeCheckValidityWithErrors on OS X 10.9/10.10 when it comes to checking the validity of an app in the following case:
the application A.app is a bundle.
inside Ap.app/Contents/MacOS there is an additional binary B.
B is launched by launchd and runs an XPC mach service.
A is a client of this XPC mach service.
When A tries to connect to B, B checks the calling process with SecCodeCheckValidityWithErrors.
On OS X 10.9/10.10, SecCodeCheckValidityWithErrors returns -67030 (errSecCSInfoPlistFailed / invalid Info.plist (plist or signature have been modified)).
On OS X 10.11 and later, SecCodeCheckValidityWithErrors succeeds.
If a copy of B is installed outside of the A.app bundle and it's this copy that launchd starts, A can communicate with B.
[Q] Was there a bug in OS X 10.9/10.10 where SecCodeCheckValidityWithErrors would return -67030 if the XPC server is inside the MacOS folder of the XPC client?
Let's say there is a launchd daemon which runs a NSXPCListener for a named mach service, and that this daemon does not support sudden termination.
Question:
Is there a point in supporting the XPC_ERROR_TERMINATION_IMMINENT error in this daemon?
Assuming it's even possible with the NSXPC APIs.
Auto Layout:
How do you get the width of a column in a NSTableView to resize correctly to fit the localized header column title?
Base Internationalization:
Is it possible to hide a view for a specific language using the .strings files? For instance, is there something like: "id-view-something.hidden" = YES; ?
If you add a NSTextField to a xib file, do you really need to manually add the NSTextFieldCell id in every existing corresponding .strings file? Or is there a command in Xcode to update the .strings files (Xcode 10.3 here)?
I've already had a look at the documentation, stackoverflow and so on. The only answer found was for the 3rd question and it looked like a 3rd party tool is required to achieve this task.
Could it be possible to provide real release notes instead of the limited current ones?
It would be very surprising that fixing the listed Catalyst issues requires a 3 GB update.
I believe it would be appreciated by everyone if there was a list of all the bugs that have been fixed so far and not just the Catalyst ones.
Just in case, it is still possible to KP macOS 11.0.1 RC every single time while running an apparently harmless userland app and a very simple Network Extension.
This type of KPs used to happen on 10.15.6 and 11.0 earlier betas.
Maybe the issue has been reintroduced in 11.0.1.
FB8884269
Is there any known issue in macOS Big Sur regarding Filter Data Provider and Screen Sharing.app file uploads?
From what I'm observing:
Uploading files through Screen Sharing.app fails when a Filter Data Provider is running and just requests to see all the data (and does nothing else).
At the same time, the remote display feature just works fine and Safari browsing works fine.
I've already filed a bug report with more info: FB8880461
The Notary Service reliability is not good.
Contrary to what the Apple System Status page reports at the time of this writing, the service is still not up and running as usual.
It takes more than 30 minutes to submit a build. The service does not seem to have recovered from yesterday outage.
Console.app shows a lot of logs from NEFilterPacketInterpose when you use a Network Extension (possibly a NEFilterPacketProvider) repeatedly.
I'm a bit worried by these from a performance point of view considering that: there is a huge number of them over time.
turning off developer mode via systemextensionsctl does not appear to disable those logs.
these logs are printed for release flavors of NEs not just debug flavors.
Questions:
Is this pointless worrying and it does not have any impact on performances?
Is there a way to turn them off (not just filter them in Console.app)?
Is there a C version of the Network Extensions APIs?
Just asking because some of the Objective-C/Swift objects do not provide useful methods.
e.g. NWEndpoint vs nw\_endpoint\_t
If you want to get the port using NWEndpoint/NWHostEndpoint, you first need to get a NSString * that is useless and "big" and convert it to an unsigned short.
If you want to get the port using nw\_endpoint\_t, you just use nw\_endpoint\_get__port to get the unsigned short.
It's worse of course if you need to get the sockaddr *.
Also with nw\_endpoint\_t, you can know the type of the endpoint…
I will file an enhancement request about this if needed but the Feedback Assistant website was down, again, when I tried.
When calling SecCodeCheckValidityWithErrors from a Network Extension, this function returns with OSStatus -67061.
"-67061" is not documented in SecBase.h (unless I'm mistaken). The CFErrorCopyDescription output for this error says something like "The operation can not be performed".
Question:
Can SecCodeCheckValidityWithErrors be called from a Network Extension?
If it can not be called, would there be, somewhere, a list of framework, functions that are not network-extension-safe?
Is there a known small memory leak in the Network Extension framework?
I'm observing a small number of little leaks (16 bytes ROOT LEAKs) that increases slowly over time in a Network Extension process.
So far I have not been able to correlate these with my code.
macOS 10.15.7
@property enabled
@discussion Toggles the enabled status of the filter.	[...]On macOS, up to 4 filter configurations of the same grade can be enabled simultaneously.
Questions:
What is "up to 4 filters" supposed to mean in the real world?
Does it mean that as on iOS, if a 5th filter is enabled, then one of the existing 4 will be disabled?
How is the software supposed to be informed about being disabled without its knowledge/consent?
Why is there a limitation in the first place?
Let's imagine that a Network Extension is running fine.
In System Preferences.app > Network, the extension is listed as "Running".
Case 1:
You then remove the Network Extension through the Network pref pane by selecting the corresponding row and clicking - and then the Apply button.
When you try to re-enable the Network Extension using the NEFilterManager API, at the end of the re-enabling process, the NEFilterConfigurationDidChangeNotification is posted.
That's fine.
Case 2:
You make the Network Extension inactive through the Network pref pane by selecting it, choosing Make Service Inactive using the action menu and then clicking the Apply button.
When you try to re-activate the Network Extension using the NEFilterManager API, at the end of the re-activation process, the NEFilterConfigurationDidChangeNotification is NOT posted.
Question:
Is the behavior of case 2 the expected one? If it is (or if it's not), how are we supposed to know that the Network Extension was correctly reactivated?
In macOS Big Sur beta, a menu item with a submenu item is not drawn the same way as in previous OS versions.
Here, "choose" ="select and no mouse up".
On all macOS/OS X/Mac OS X flavors:
If you choose a menu item with or without a submenu, the menu is highlighted.
On macOS Big Sur:
If you then choose a submenu item, the parent menu item is no more visually highlighted.
On previous OS versions:
If you then choose a submenu item, the parent menu item is still visually highlighted.
The keyword here is: visually because the parent menu item highlighted property is still set to YES even though the menu item is not highlighted (on macOS Big Sur).
Question:
Apart from the fact that this will be confusing to some users, how can we know about this strange new state when we are dealing with a custom view inside the parent menu item?