https://youtube.com/watch?v=CVAaLlOUD00 [Embed]
windows
zip: https://github.com/hydrusnetwork/hydrus/releases/download/v586a/Hydrus.Network.586a.-.Windows.-.Extract.only.zip
exe: https://github.com/hydrusnetwork/hydrus/releases/download/v586a/Hydrus.Network.586a.-.Windows.-.Installer.exe
macOS
app: https://github.com/hydrusnetwork/hydrus/releases/download/v586a/Hydrus.Network.586a.-.macOS.-.App.dmg
linux
tar.zst: https://github.com/hydrusnetwork/hydrus/releases/download/v586a/Hydrus.Network.586a.-.Linux.-.Executable.tar.zst

Hey, I did a hotfix to fix a stupid bug when moving from videos to images. If you got the release within twenty minutes of this post going live, get the updated v586a above!

I had a great week getting siblings and parents lookups running faster and finishing some long-planned Client API work. The update may take a minute this week!

Full changelog: https://hydrusnetwork.github.io/hydrus/changelog.html

highlights

I am happy with the new siblings and parents dialogs, but unfortunately the fetch jobs were frequently running super slow on the PTR. This has been a long time problem in other sibling/parent places, and I could never really figure it out. This exposed the problem better and I simply put a bunch of time into the database sibling/parent storage structure and search code this week, and I think/hope I have fixed the worst of it. I also fixed the crazy-long lag spikes we were seeing, which was, unfortunately, just me being stupid last week. If you sync with the PTR (or not!) and have had slow sibling/parent lookups (including in places like the tag autocomplete results list), let me know how it goes!

If you have the media scanbar set to hide completely when the mouse is not over it, I think I fixed the issue where it would come up blank if the media was paused while the scanbar was hidden!

The options widgets that are an editable number with a checkbox beside saying 'no limit' are now initialised with a nicer default number when they start with 'no limit' checked. Previously, this stuff was all initialising to 1 every time, which wasn't always helpful if you actually wanted to go edit it. I'm pretty sure all the 'noneable' integer widgets in the options dialog now soft-initialise to the actual defaults those numbers are on a fresh client.

If you use multiple local file services, then when you middle-click a tag in the media viewer, the new search page now correctly retains the original file domain of the media viewer. Although I use multiple local file services myself on my IRL client, I do not browse around in them all that much, so let me know where else this sort of stuff defaults to 'all local files' or 'all my files'.

I have removed a hard limit that said 'don't run an import folder for more than an hour'. If you have a mega import folder with hundreds of thousands of files, let's see how it goes.

If you have done Client API or tag migration work and ended up with some bizarre tags that are both pending (+1) and petitioned (-1) to the same service, check the changelog this week!

client api

Thanks to a user, the Client API call that renders images can now output in jpeg and webp, can change the quality of the output, and will render to a target resolution!

I also think I finally finished off the first full version of 'multiple local file services' Client API support. You can now set a custom import destination for the 'add file' and 'add URL' commands, and you can now copy files from one local file service to another using a new 'migrate_files' call.

next week

I've been working on several things recently that can populate a multi-column list with a hundred thousand or more rows, and it has reminded me that my core list code relies on an old hack in it that makes initialising and sorting such big lists super laggy. I have researched how to improve it and hope to do so next week!

Unfortunately, I do feel myself going down with something, so it might be delayed.