The -compat business is really baffling to me. I think you’re right, it just blows me away that media players have all standardized on a completely effed-up SRT file format for RTL languages. I saw a media player that handled correctly-formatted RTL subtitles once, and, sure enough, there was a bug filed that it didn’t work with the vast majority of RTL SRT subtitles already in existence.
On the latter issue, the description is confusing but correct (I may work on that). “native” means Greedo’s lines aren’t included in the SRT/SUP file, because they are expected to be burnt-in on the video. “nocrawl” means the translated crawl is not included, but Greedo’s lines are, if the video doesn’t burn in Greedo’s lines. For example, if the English version of the video doesn’t have Greedo subs, then you’d want the Greedo lines in your SRT/SUP files but you would still not want the English crawl text in your subtitles, because that’s still in the video. Unless you’ve got the German Krieg der Sterne video, in which case you want -eng-full. It’s a bit of a mess, but that’s what I get for trying to cover every possibility.
EDITED TO ADD:
For clarification on what the difference is between -compat files and normal files… if you run Windows, Notepad is a great artifact of primitive Unicode text technology. In Notepad, if you open a -full RTL file, it will look all messed up (primarily punctuation on the wrong side of the text). If you open the -compat file, it looks fine. But try that using some more modern Unicode text-handler (a browser, word processor, whatever), and the situation is exactly reversed.
A crazier example is if you open a -compat file in Notepad (which looks fine), and then copy and paste into a browser – the characters get all jumbled around in the paste operation, because they were never in the right order to begin with, Notepad just displayed the jumbled characters in a jumbled way that made them look like they were okay when they actually weren’t.
Argh! Unicode solved this problem decades ago!
Anyway, all of this seems a little academic until you consider that I have scripts that read the SRT files and render images based on that text, and I can’t afford to be messing around with scrambled text. And I copy the subtitles and paste them into Google translate all the time, to make sure I’m working on the right line, where, again, text scrambling is unhelpful.