https://youtube.com/watch?v=zBuJD-ugT_g [Embed]
windows
zip: https://github.com/hydrusnetwork/hydrus/releases/download/v559/Hydrus.Network.559.-.Windows.-.Extract.only.zip
exe: https://github.com/hydrusnetwork/hydrus/releases/download/v559/Hydrus.Network.559.-.Windows.-.Installer.exe
macOS
app: https://github.com/hydrusnetwork/hydrus/releases/download/v559/Hydrus.Network.559.-.macOS.-.App.dmg
linux
tar.zst: https://github.com/hydrusnetwork/hydrus/releases/download/v559/Hydrus.Network.559.-.Linux.-.Executable.tar.zst

I had a great week working mostly on one thing: converting the timestamps in the program from seconds to milliseconds.

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

milliseconds

tl;dr: time numbers better, you don't have to do anything

Since the program started, it has tracked the various file 'times', like import and archive time, using a system that stores time with second precision. It knows that file x was imported at 2023-12-02 14:04:51, but nothing finer. This is common in many computing applications, but sometimes you want more. Particularly, in hydrus, whenever you have a fast importer that brings in two files within the same second, the program does not know, later, which was actually imported first (e.g. when you sort by 'imported time'). This can be annoying when you import a whole chapter of a comic and don't have good page tags set up yet--random pairs of pages will be flipped whenever you next try to sort those files by 'import time'.

So, this week I converted the database to use a time system that has millisecond resolution. Any files imported from now on, or deleted, archived, modified, or 'last viewed', will now get a millisecond timestamp. You don't have to do anything, and nothing outside of the 'edit file timestamps' dialog will appear any different, but it may take a minute to update.

This work was quite simple, but there was a ton of it, years and years of system build-up to go through. Rather than make an already messy system more complicated with a rough injection, I decided to invest the time and clean everything as I moved to the new format. I've tested it back and forth, but given the number of changes, I wouldn't be surprised there is a typo in there somewhere--if you see a file saying it was last viewed '54 years ago' or similar, let me know!

Also, the Client API can now edit timestamps, including with the new millisecond precision. If you are an API dev, check out the new call (and 'edit file times' permission) here: https://hydrusnetwork.github.io/hydrus/developer_api.html#edit_times_set_time

misc

Just a side note, I have removed the sankaku downloader defaults from the program, so new users won't see them. None of them were working well/at all, especially in recent weeks. If you want to grab from complicated sites, and there isn't a clean solution on the shared downloader repo here, https://github.com/CuddleBear92/Hydrus-Presets-and-Scripts , I recommend going with a more robust solution like gallery-dl/hydownloader or just finding the content elsewhere.

If you ever run something like 'system:known url: regex = blah.com/(regex stuff)' and are annoyed at how slow it is, try adding a 'system:known url: domain = blah.com' pred in addition. This combination will now be explicitly much faster than just the bare regex (previously, it could be faster, by accident).

next week

Now we have ms timestamps and nicer code, I'd like to try tackling an 'edit timestamps' dialog that works for multiple files, including a 'cascade' command for force some clever sorts.