Updated to work with Genshin Impact 5.7.0.

bump go version to 1.24.4
Updated hoverfly version to v1.11.2

Fixes & improvements:
Fixed power toggle not displaying if the power state DDC/CI feature was manually enabled.
Implemented potential workarounds to Twinkle Tray not detecting monitors or throwing errors under certain conditions.
Updated dependencies.
Updated localizations.

Breaking changes:
We try to minimize breaking changes, but sometimes they are necessary in order to fix high priority issues. Where we do break compatibility, we do our best to make sure that the changes are minimal and require few changes in user projects.
You can find a list of such issues by filtering the merged PRs in the 4.5 milestone with the breaks compat label. Here are some which are worth being aware of:
Tilemap physics are now handled in chunks. Only affects get_coords_for_body_rid, as now a single body can cover multiple cells. (GH-102662)
Animation:
This might be a little more technical than usual, but the work Tokage put into implementing BoneConstraint3D warrants a highlight all the same (GH-100984). With this new class, users will be able to bind a bone to another bone, opening the door for more natural movements and poses.
Onto something more suitable for a blog post highlight: UX improvements! YeldhamDev implemented support for selection box movement and scaling within the bezier editor, making it a piece of cake to perform changes to points in batches (GH-100470). Arnklit continues the bezier improvements with GH-95564, allowing users to auto tangent new points in a balanced or mirrored manner. The animation player gets some love as well, with the ability to sort animations alphabetically (GH-103584). Lastly, and featured below, is a very long-awaited UX improvement: animation filtering! (GH-103130)
Audio / Video:
Berama brings us the ability to seek Theora video files via the new set_stream_position function (GH-102360). In doing so, they’ve additionally improved our multi-channel audio resampler, meaning that videos with 6+ channels will no longer crackle. A much more technical breakdown & additional features can be gleaned from the PR.
And more:
Add metadata tags to WAV and OGG audio streams. (GH-99504)
Internal nodes are no longer duplicated. Only affects users deliberately utilizing internal nodes. (GH-89442)
NodeOneShot fading now uses self delta instead of input delta. Brings behavior closer to other AnimationNodes, as the old implementation was exclusive to NodeOneShot. (GH-101792)
2D & 3D Navigation region and link updates are now asynchronous. (2D: GH-107381, 3D: GH-106670)
NavigationServer2D avoidance callbacks changed from Vector3 to Vector2. (GH-107256)
Removed the gradle_build/compress_native_libraries export option. With Android builds now supporting 16kb pages, the native libraries are now required to be uncompressed. (GH-106359)
We are considering re-introducing this option for users who don’t target Android 16, or distribute APKs outside of Google Play. (GH-107681)
“Areas Detect Static Bodies” setting removed from Jolt Physics, this is now always enabled. (GH-105746)
set_scope removed from JSONRPC. Manual method registration is now required via set_method. (GH-104890)
C#:
First-time contributor Justin Sasso kicks things off with linux-bionic RID export support (GH-97908). For those that don’t speak buildsystem, this enables NativeAOT on Android! For those that don’t speak .NET lingo, “NativeAOT” refers to the ability for .NET applications to compile directly to a device’s native code, bypassing the need for the .NET runtime entirely. NativeAOT apps have the benefit of significantly faster startup and smaller memory footprints, which are both very welcome additions for mobile devices.
Finding performance improvements in an interop context is like finding a needle in a haystack. Actually tracking down where some point of slowdown or inefficency is taking place across entirely different environments is difficult to the point that most people won’t even attempt it. Delsin-Yu is not most people, because the improvements that came from simply removing StringName allocations on unimplemented getters/setters saw a staggering 60? decrease in resources (GH-104689).
Not everything related to .NET was able to make it in time for 4.5, but they’re still worth mentioning because of how much effort the team has already put into them. For instance: we’re well aware of the excitement around bringing .NET to web builds for Godot, and progress on that front has been very promising. We’ve even covered this very topic in a previous blog, where we discussed the rocky road of bringing this project to light even featured a prototype which you can try right now!
Our other long-term project for C# is revolving around the gradual move to GDExtension. The current module approach, while entirely functional for what it is, has historically been a fairly hacky implementation. Grafting on interop functionality between the engine itself and the dotnet runtime has proven to be error-prone, leading to a disproportionate amount of man-hours sunk into ensuring everything functions as expected. The hope is for the move to GDExtension to mean that all interop calls are handled in a manner that’s universally applicable; that is: a manner that any programming language could take advantage of.
And more:
Android: Add a preload hook to load .NET assemblies from the APK. (GH-105262)
Android: Load .NET assemblies directly from PCK. (GH-105853)
Core:
Changes to the core of the engine require significantly more scrutiny than other parts of the engine; this comes down to how critical and foundational virtually every single piece of code proves to be. This makes it all the more impressive that there’s so much worth highlighting in 4.5 that’s specific to core!
Adding a way to properly log errors and warnings, as well as get backtraces in logs when they happen, was among the most highly-requested features from our users for years. By their powers combined, Mikael Hermansson and Juan Linietsky have added script backtracing support for GDScript and C# (GH-91006). Finding the root problem behind warnings/errors that appear at runtime required being able to reproduce them in the editor to use the debugger. Developers will now have the possibility to see backtraces of runtime errors directly in their logs, making it possible to debug and fix issues that happen under user testing or in shipped titles. This functionality is always available in debug mode, but can be activated in release mode for GDScript if Debug > Settings > GDScript > Always Track Call Stacks is enabled in the project settings.

AUR numpy problem:
there was a problem with the AUR (Arch Linux) hydrus package while I was on vacation. the python package numpy updated and a couple deprecated lines I had missed now threw errors. for those who auto-update to the newest of things (eg. as on AUR), this broke video view and file import. sorry for the trouble! by luck I had fixed half of this by accident a few weeks ago, but I also missed a few more lines. a user kindly figured out the fix and I was able to merge it into master early for those who could pull. rolling back numpy to <2.3.0 was another temporary solution. the fix is now properly in this v626, so when AUR v626 rolls out, everyone should be good again. if you are an AUR guy and really want to avoid this in future, I recommend moving to your own source install, as here: https://hydrusnetwork.github.io/hydrus/running_from_source.html it takes a couple minutes to set up, but with our own venv that we control, we can fix the library versions to stuff that we know will work in perpetuity.(issue #1744)
Paint.NET:
Paint.NET files are now importable (or at least anything since ~2006). the client pulls resolution and should be able to do thumbnail, but cannot render them fully. they count as an 'image project file'. let me know if you have any v3 .pdn files that don't work!
default downloaders:
the derpibooru file page parser is updated to get tags again. I think I updated everything correct, but let me know if anything is parsing different to how it was before
duplicates auto-resolution:
improved my 'A and B are visual duplicates' algorithm with a new pre-histogram gaussian filter to better tune out jpeg artifacts and a more careful later 'absolute skew pull' inspection
many previously 'definitely visual duplicates' false positives are now detected as various states of 'not visual duplicates' or 'very probably visual duplicates'
many previously false negatives are now correctly detected as 'definitely' or 'near-perfect' visual duplicates
many previously true positive duplicates are now detected as higher levels of confidence of duplicate
thank you all for submitting your false positives and false negatives. I now have one pair that still false positives as 'definitely visual duplicates', and a couple that still go 'very problably', which I would like to fix. the remaining problem to solve is file-to-file edge difference comparison, which I feel pretty good about attempting at this stage. I also feel better about finally turning this system on for the duplicates auto-resolution rules soon, with the caveat that I'll probably recommend users only go 'near-perfect' to start
after thinking about it, I renamed the 'definitely' to 'almost certainly'. with an even more confident tier in 'near-perfect', 'definite' is the wrong word
I am still interested in any false positives or false negatives you encounter hereon. the main problem I now have to beat in image terms is where the alternate is an artist correction that moves a small object of interest a few pixels amidst a sea of similarly coloured pixels, for instance moving an anime nose a few pixels right. eyes that have slight differences (tear-drops, heart-shapes) are also proving a problem, but the main one is a small thing moving without changing average colours anywhere. also, obviously, if today's algorithm is actually worse anywhere, let me know!
better export filenames:
the call that generates export filenames for manual exports, export folders, and drag and drops with export filenames is improved in several ways--
you can now set your own 'max filename length' under options->exporting. defaults to 220 (most OSes are 256, although Linux eCryptFS is ~140)
on Windows it now tests filename and total path length against characters rather than encoded bytes
the test against max total path length (260 characters on Windows, which we shave to 250 for extra safety) is more reliable
on Linux it now tests against a max total path length of 4096 bytes, and on macOS 1024 bytes. we shave by another 20 bytes for safety
the test against total filename length now recognises when a filename pattern produces subdirectories and will not include them for the filename length test
there is less padding fudge in the system, around 54 characters! if you were clipped before, you will likely see longer filenames immediately
if you have an export folder that uses frequently elided filenames, it is going to be busy as it generates new filenames on next run. let me know how you get on!
added a bunch of unit tests to test filename eliding, for: null, filename, path, filename and long path cases, for ascii and unicode, for character limits (windows) and byte limits (linux)
url slash test:
when I first made the network engine, I had the URL normalisation routine collapse multiple leading slashes on a URL path down to one. for instance, https://site.com//images/123456.jpg becomes https://site.com/images/123456.jpg. this is actually incorrect handling on my part, and there's a site or two where it matters. unfortunately, I cannot make the switch without breaking URL Classes that already relied on the collapse, and I do not know how many of these there are out there
so, I have added a checkbox to options->downloading where you can participate in a TEST to change the normalisation behaviour. I would like advanced users who use unusual downloaders to turn on the test and run their subs and stuff as normal. let me know if anything suddenly doesn't work. I suspect 99.8% of everything will be fine, but I don't know so let's test it
as a side thing, I have adjusted my master URL lookup tool, which checks for duplicates in the file log and does 'already in db'/'previously deleted' url status lookups, to consider the leading single slash as matching the two slashes. I can't do the same for URL Classes though!
enter vs add tags:
the manage tag parents and siblings dialogs are now 'add_only' from the 'add' button. previously, this was really an 'enter' command that would add new but petition pre-existing, but this workflow was never very intuitive and now we are reguarly dealing with hundreds of rows it is only ever confusing and annoying. similarly, the 'import' button now only offers a way to add new rows. sorry for the inconvenience here--I regret this took so long to figure out. if you want to do very large clever deletes, select the rows you do not want with ctrl/shift+click and hit the 'delete' button. if you want programmatic ways to remove rows (maybe a return of the 'import' conflict-remove, or a full-on only_delete mode), let me know how you would like it to look
the 'CONFLICT: Will be deleted on add.' list notes as you enter siblings are now more varied and precise
similarly, in manage tags, the 'allow remove/petition result on tag input for already existing tag' cog-menu option now defaults for new users to False, and all updating users will be set to False in v626. I don't like to force option changes on update, but most people are surprised to learn this option even exists, so I'm flicking us all, one-time, to the less confusing mode
duplicates:
auto-resolution rules are now processed in alphabetical order. the preferred order in which rules and pairs are processed is a complicated topic and I am not sure on what is generally ideal, but if you have an opinion you can now force it
I think I fixed some layout squish with the duplicates hover window. the window sometimes won't grow to be a little taller, particularly if a comparison statement goes from single line to multiple, which was causing the buttons to squish to make everything fit, until the user jiggled a window resize
I think I fixed some transitional layout flicker with the duplicates hover window, particularly when some of the comparison statements are multiple line. also the previous pair's score line now properly blanks out while the new comparison statements are being loaded
if a duplicate metadata merge options panel no longer allows you to set 'move from worse to better' tag action when you hit 'edit action' on a tag repository. this choice was accidentally being included here.
if a duplicate metadata merge options does have 'move from worse to better' tag action set for a tag repository, through whatever grandfathered legacy reason, this is now treated as a copy action. previously it was hitting a 'you should not have been able to select this' safety check and doing nothing! if you have a hole because of this, don't panic--it is just another hole we'll want to fill in with retroactive duplicate merge, when we get around to that
misc:
when the client adds or edits services, it now forces case-insensitive unique names. you can use whatever upper case you like, but you won't be able to make two services called 'score' and 'Score' any more. this helps out some parsing stuff
same deal for subscriptions, duplicates auto-resolution rules, and import/export folders. not because we parse these names, but just to better differentiate big objects we want to be careful about
fixed name deduplication when editing an import folder
thanks to a user who submitted a PNG with 'srgb' colourspace metadata, I have fixed PNG colours for these files. this is related to the recent gamma/chromaticity work. a bunch of PNGs that previously rendered slow will now do so fast and with correct colours
I've added system:inbox/archive to the list of selectable system predicates for all search file domains (previously they were hidden when your search domain had no 'real' and 'current' file domain). inbox/archive doesn't really have meaning outside of your local files, but advanced searches that switch file domain do sometimes carry these preds over to something like 'deleted from my files', so we might as well support them officially and fix the exposed nails. I think the logic will be crazy sometimes, and any counts too, so if you do clever searches and use them, let me know if and when they fail
the routine that bundles many items into a single UI presentation text (for instance, when you paste a whole bunch of query texts into a sub and it talks to you about them) now deals with very long lists better. it'll now max out at 25 lines, each line about 64 characters, with the last being some form of 'and 741 others' overflow. we think that pasting many thousands of queries into a sub may have been causing out of memory crashes when a dialog >32k pixels tall was being created. this obviously also generally fixes crazy tall dialogs in these cases
when the file migration system chooses locations to pull from and push to, it no longer selects candidates of equal urgency pseudorandomly, but now pulls from the disk with the least free disk space and pushes to the one with the most
fixed some 'repair missing file location' handling when the incorrect path is stored in the database in an invalid portable/absolute format. this may be related to some flatpak path magic
a related problem where in rare cases a normal file migration would abandon the job early because it could not delist the old location is fixed
help and env stuff:
updated help regarding running the db on BTRFS and NTFS filesystem compression. thanks to the users who let me know that BTRFS is ok and faster these days, particularly on WAL journalling, which we use by default
added 'how to test and get git' to the Linux and macOS 'running from source' help
clarified some 1/2, A/B stuff in the duplicates auto-resolution dialog text
fixed some bad newline .md formatting in 'running from source' help
updated the 'test' mpv version to 1.0.8 and PySide6 to 6.9.1

S: Send all frames to capture, including skipped ones

feat: Update electron to v36.5.0