https://youtube.com/watch?v=PgqmqeH0iDs
windows
zip: https://github.com/hydrusnetwork/hydrus/releases/download/v568/Hydrus.Network.568.-.Windows.-.Extract.only.zip
exe: https://github.com/hydrusnetwork/hydrus/releases/download/v568/Hydrus.Network.568.-.Windows.-.Installer.exe
macOS
app: https://github.com/hydrusnetwork/hydrus/releases/download/v568/Hydrus.Network.568.-.macOS.-.App.dmg
linux
tar.zst: https://github.com/hydrusnetwork/hydrus/releases/download/v568/Hydrus.Network.568.-.Linux.-.Executable.tar.zst
I had a good couple of weeks mostly figuring out better behind-the-scenes URL handling.
Full changelog: https://hydrusnetwork.github.io/hydrus/changelog.html
urls
For normal users, tl;dr: URLs better now, you don't have to do anything.
I made a dumb mistake when I first created the downloader engine in how I handle URLs behind the scenes, and today it is fixed. You don't have to do anything, and you shouldn't see any big changes, but in general, URLs and query texts with unusual characters should work better now. You may also notice a URL in the file log or search log will appear one way, e.g. with japanese kanji, but when you copy it to clipboard, it is all encoded to %EF kind of stuff--your browser address bar probably works the same way. It should just mean things copy between your hydrus and other programs with fewer hiccups.
If you regularly use some very odd URLs or downloaders, there might be a hiccup or two. A file that relies of strange encoded characters in its known urls might redownload one time if a subscription hits it, or one extremely strange query text (it'd be a single tag query that has a % character somewhere in the middle) might need to be renamed to %25. If you have a crazy custom downloader that relies on %-encoding tech, please pause it before you update, and then do a test afterwards--it may be the hack you added for the old system is no longer needed. In any case, let me know how you get on, and if we run into a problem with a certain downloader, we'll fix it together. Overall, I'm hoping that working with encoded URLs exclusively will make more complicated downloaders easier to figure out in future, rather than having to fight with odd characters all the time.
I decided to cancel this release last week because I wasn't confident on how I was handling the advanced edge cases of encoding here. I am happy I did, because while all the 'just download from a booru' style downloaders (like the defaults) were fine, the most experimental downloader situations (that e.g. needed encoded parameter keys) needed more work. If you are a downloader maker, then you will be in the guts of these changes, so please check out the changelog. There's new UI in 'manage url classes'--path components and parameters now have their own edit panels with linked text boxes that show the normal and encoded values of the stuff you are entering, and there's also new 'ephemeral token' tech that lets you decide in what ways undefined parameters should be clipped before the URL being sent to the server. The idea of the 'request URL' being different to the 'normalised URL' is broadly elevated and exposed across the program.