Closed KITT3000 closed 11 months ago
Huh. gross. So, my notes below, but included cause you might be interested.
1.) The log is read sync right now - this was great for testing, but super bad for production. Needs fixed.
2.) A 50MB log is bad, as you know - but it shouldn't crash. There are two ok-ish options here, both with their downsides:
I'm leaning towards a two-fold solution -
Thoughts?
Currently I can't do anything to the window, nor change filters or anything else, only minimize the window and close it. As for the solution, I noticed that usually the end of the file is repeated sections with "LUA call stack", I think it would be best to hide it somehow (show only one section and collapse the rest with the possibility of expanding), or cut/stop loading the file after several such entries.
Aye. the issue is, the render portion takes forever to do it's thing - it's actually reading and sending the file to the window very, very quickly - then the bit to parse through it is super slow. I'm working on some ways to greatly speed that up, including some automated truncation for stuff like that
sigh. auto-close. This is why you don't use "Fix" in a commit message. lol.
Anyway, what I ended up finding is that it was neither the file IO nor the regex parsing that was the limiting factor here - it's that asking chromium to display a few hundred thousand table rows is bad. So, what it actually does is render the 20 or so that you can see in the window. And when you scroll, it grabs the appropriate lines that you see now, etc, etc - the bonus is, the filtering and finding code is much, much cleaner, as it is operating on a data structure rather than a DOM structure. In my tests, the 51Mb log file I made by pasting my regular log a few dozen times together ran every bit as fast as a tiny log file did - with the caveat that it took around a second to load when it hit the window. If your running, constantly updating log file is that big, you might notice the lag at some point, but the debouncer also doesn't load the log that fast (I think there is a 5 second delay built in there).
Anyway, this will be in the next version, whenever that may be (probably a day or two, I don't have much else cooking, but I do want to update the dependencies again)
I found some large LOG files so you can test them yourself logs.zip
you rock, thank you!
Nice
If you are feeling adventurous - https://github.com/FSGModding/FSG_Mod_Assistant/releases/tag/v2.9.92
The new log viewer is in there. The other huge update is the mod parser is multi-threaded and runs in the background now. Feels faster, and it is way less resource intensive.
Update release in 3.0.0 - lemme know if you still have issues. Seems pretty sorted to me - thanks for your continued interest in testing and improving this :)
Is there an existing issue for this?
Current Behavior
I checked the downloaded LOG file, then I didn't reload it to my LOG, I just used the program to change active mods, etc. I deleted the file with this LOG file and now when I try to check, nothing wants to load in this window, I can't click any button in this browser window, after a short while the window turns completely white. I remember that the log was about 50MB and there was also a problem with viewing it, scrolling was terribly laggy.
Expected Behavior
No response
Steps To Reproduce
No response
Environment
Anything else?
No response