Search operation executes on dispatcher thread

[dmit25]

Hi, this project is just brilliant and I've been using it for a couple years for now. The only one problem is UI freezes when working with huge documents >10Mb and active search running. The root of the problem is that offsets calculation process is running in the main UI thread, matching big text using regexps takes a lot of time ( e.x. on my machine when 20 Mb file is opened and any type of search active any keypress leads to 2,2 secs delay, matching text with simple regex takes about 2,1 secs). So the first thought - move matching phase to another thread, additionally add interception code in each iteration so that any running background process could be easily stopped by new one (typing -> search started, typing before search completed -> search process stopped and new one started).

I've worked a bit on this problem and have some implementation in my fork (worked out dispatcher issues + interception implementation), no more freezes during typing + simplified search dependencies, unfortunately I've abandoned some search functionality? but I'm sure that with your help we can implement match whole word without document dependency (so it could run in parallel thread).

If you are interested in this type of functionality I can upload performance reports and send pull request.

Again, thanks for your work.

[siegfriedpammer]

Would you still be interested/able to provide your changes as a pull request?

[dmit25]

Oh, I no longer work with this library, but I guess it is possible to pick up changes from my fork. I guess it's not urgent :) I'll try to create a proper PR.

[siegfriedpammer]

Honestly, I did not expect a response... take your time :) Thank you very much!