Manictime's screenshot feature is a life saver. But it also takes up a ton of space using a full separate image for each screenshot. 90% of the time, adjacent screenshots are nearly identical, except for some text on the screen. Huge storage efficiency could be gained by using difference calculation between screenshots.

For example:
1. Every 10 minutes, take a fresh screenshot. Store it as a png like normal.
2. Every 15 seconds thereafter, take a new screenshot to memory
3. Using a library like ImageMagick, calculate the subtraction from the most recent png screenshot and the new screenshot in memory.
4. Store that subtraction alongside the other screenshots, but with a different file extension, like .pngdiff.

When browsing the screenshots, just do the reverse
1. Open the screenshot (png or pngdiff) for the time selected by the user
2. If it happens to be a png file, simply display it as-is.
3. If it's a pngdiff file, open the png most recently taken before the selected pngdiff
4. Using a library like imagemagick, perform an addition operation on the png and the pngdiff
5. Display the calculated image.

This could save many gigabytes of disk space without too much CPU overhead.

Backstory:

I'm using the free version of MT on Windows for self-tracking for over 3 years.

Recently I wanted to see total time working on particular drawing, and I was able to do it like this:

Statistics - Custom - Add - Timeline: Documents (at top) - Files (at left).

Now at right there is a huge list of filenames gathered in these years, some with paths, some not, even some with quirky path formats like "(D:) drivename / path / deeperpath / file_name.ext" (with parentheses, slashes and spaces exactly like that).

The filelist takes around 10 seconds each time to generate or sort.

At first, every file was checked. I selected all with Ctrl+A (works instantly), clicked one of the checkboxes (MT client UI went unresponsive for several minutes, maybe over 10 min), then everything became unchecked.

I found some files I wanted to sum and checked them back (one by one it works instantly too).

Finally, refreshing all statistics takes around 5 seconds, less than generating the list of files in UI.

Suggestion:

It would be a lot easier to enter some pre-known filename wildcards, than digging in the filelist. Simplest I can think of - one pattern per line in a text area, that could be shown instead of checklist by an option. For example, using radio-buttons at bottom of "Edit Statistics" dialog:

Match:
- checklist any
- checklist all (only show/enable for the Tags timeline?)
- wildcard any
- wildcard all.

Small suggestion for Away time.

Issue: having that window opening each time can be frustrating if one wants to close it most of the times.

My example: this issue for me occurs when I study with a PDF of exercises opened on the PC, so screen goes off after 5 mins but I take 10mins to do an exercise, when I come back to see the PDF then the window appears with 10min away time, each time I have to close it and that's annoying. I already set 15 minutes to "track inactivity if PC is idle for more than", but that doesn't seem to affect Away time. (translated from Italian)

Solution: I suggest a different mode (maybe optional): when returning to the PC, instead of a big window popping up in the middle, let a small notification in bottom right corner appear (you can use Windows notifications, but I think it's better to implement an overlay system, so not a real window but just an overlayed thing) so this thing appears and if I want to tag I interact with it, otherwise I can still use the PC even with it shown as it's small, and it vanishes after like 5 seconds of using PC without interacting with it, so I don't have to close it manually like normal window (which doesn't close itself even if I click away).

Hope you got it, this would be great as a snappy modern update for your already good app. Thanks!

Allow the client to require a specific certificate or certificate authority on the server. Manictime carries a lot of potentially sensitive information for me. In theory if someone wanted, they could create a server that would accept any credentials from a manic time client and accepts the entire upload by pretending to be the server the client normally connects to. This would allow the client to validate that the server is in fact who is expected. And as a result stop this potential information leak.

On a more practical note where would this be useful:
You have a laptop that connects to a manictime server at home on a specific IP. You take it somewhere and connect to a different network with a malicious actor that created a fake server and controls DNS or happens to pick the right IP that my desktop client wants to talk to. The client wont check who its connecting to and it would upload all my data to someone else.

Next scenario is basically the same but for an android phone. The actor would now potentially have all my call history including contact names and phone numbers.

For outbound connections firewalls generally won't help... Not without a lot of magic anyway.