Yeah, seems to be fixed: I tested in an Ubuntu 24.04 vm with WezTerm 20240203-110809-5046fc22 and couldn’t reproduce the issue.

(I first tried the nightly, which from the repository was 20260117-154428-05343b38, but that one was missing all window decorations entirely, and so couldn’t be resized at all.)

I’ll close the issue and unsubscribe, as I’m not using WezTerm currently myself. Others can reopen it if still affected.

I noticed that for the specific case of picking an element from the end using just the size of the array itself, the above is a somewhat contrived example; a plain -1 would work just as well. But I can change the example script as below, and the difference between outputs still holds:

#!/bin/bash

array=(
    "a"
    "b"
    "c"
)

i=3

value="${array[i - 1]}"

printf "%s\n" "${value}"

With the same arguments given, output differs between 3.12.0 and 3.13.0, as demonstrated in the example below. I didn’t see this mentioned directly in the release notes, so I was wondering if this is an intentional change or a bug. I don’t mind the change per se, but just wanted to be sure that it’s permanent before I adjust all my existing scripts to conform to the newer way shfmt prefers this.

$ cat asdf
#!/bin/bash

array=(
    "a"
    "b"
    "c"
)

value="${array[${#array[@]} - 1]}"

printf "%s\n" "${value}"

$ /usr/local/bin/archive/shfmt/shfmt_v3.12.0_linux_amd64 -i 4 -ci -l -d asdf # no output

$ /usr/local/bin/archive/shfmt/shfmt_v3.13.0_linux_amd64 -i 4 -ci -l -d asdf
asdf
diff asdf.orig asdf
--- asdf.orig
+++ asdf
@@ -6,6 +6,6 @@
     "c"
 )
 
-value="${array[${#array[@]} - 1]}"
+value="${array[${#array[@]}-1]}"
 
 printf "%s\n" "${value}"

Still an issue in Matheria 7.2.2.4 (server version 32.0.5).

This also manifests a bit differently when using the lock screen player: resuming playback after the call actually skips back to to the start of a previously played track instead of resuming the current one (from where you were when the call came in). The player even shows the current track title, despite playing a different one.

Steps to reproduce this variant:

1. have a directory with two audio files, 1.m4a and 2.m4a
2. start to play 2.m4u
3. lock the screen
4. call the phone (just have it ring, then hang up)
5. resume playback from the lock screen

Expected behaviour

Playback of 2.m4a resumes from where it was when the call came in

Actual behaviour

The player beings to play 1.m4a from the start (but claims to play 2.m4a)

4.7.1 was released three weeks ago, but as I’m writing this, prebuilt binaries have yet to appear on the release page. Maybe the build jobs have failed?

I’m still getting federated posts from updating old ones on my blog (with the plugin now at version 7.5.0). Not every time, but something like four out of maybe two dozen that I edited today (from 2018) have so far appeared on Mastodon.

Should I open a new report? I noticed the merged PR (above) mentions the block editor, whereas I’m using the classic editor.

What

A checkbox on the settings page to optionally disable federating updates of existing posts.

Why

Previously:

If you edit an old post, that is not yet on mastodon, then it is technically an Update Activity but it shows up on Mastodon after the Update, so Mastodon seem to handle Updates without the updated timestamp as an Create.

I edit old posts (from before installing this plugin) every now and then, and find it pretty awkward when this causes those posts to show up on Mastodon as if they were new — more so than not federating updates at all. That’s why I’d like to have that option: to be able to disable federating updates altogether.

Only federating updates for posts that have already shown up on Mastodon would be a better solution, but also a bit more complex I believe. (It’d require some way to keep track of this status, as also mentioned in the issue that the quote above is from.)

How

No response

Okay, this was a conflict between the new integration and the third party integration I had been using up until now. I did uninstall the old integration before trying to setup the new one, but apparently that wasn’t enough; I also had to manually purge all references to sleep_as_android (both old and new) from .storage/core.entity_registry. Only after that (and core restart) the setup showed the webhook URL as expected.

The problem

The setup instructions say ”You will be presented a URL during the setup process.” I see no URL during the setup process (nor is it available anywhere after setup).

Steps to reproduce:
Select Settings > Devices & services > Add integration > Sleep as Android

What I expect to happen:
To ”be presented a URL during the setup process.”

What actually happens:

What version of Home Assistant Core has the issue?

core-2025.9.1

What was the last working version of Home Assistant Core?

No response

What type of installation are you running?

Home Assistant OS

Integration causing the issue

Sleep as Android

Link to integration documentation on our website

https://www.home-assistant.io/integrations/sleep_as_android/

Diagnostics information

config_entry-sleep_as_android-01K5BG3KZKPF229RQSVK9BYAYC.json

Example YAML snippet

Anything in the logs that might be useful for us?

I’ve tried enabling debug logging for the integration, but it produces no output at all.

Additional information

No response

Yes, it seems to work as expected (i.e. identically with both OG and sudo-rs) when the are lines split, in either order, as long as timestamp_timeout value is unquoted. Double-quoting it again causes the ”expected nonnegative number” error (regardless of whether that line is first or second).

$ sudo-rs chown jani:jani test
/etc/sudoers.d/90-ssh-auth-sock:2:39: expected nonnegative number
Defaults!/bin/chown timestamp_timeout="0"
                                      ^
$ sudo cat /etc/sudoers.d/90-ssh-auth-sock 
[sudo] password for jani:
Defaults!/bin/chown env_keep+="SSH_AUTH_SOCK"
Defaults!/bin/chown timestamp_timeout="0"