An op, had same encounter few days ago. He also translated one of fillippones post to his local language. I think the character for some local languages requires the Unicode while that of English is 65535 bytes (65,535 ASCII/English characters) per post.
Yeah, that's what it is. Posts over 65535 bytes get cut off at 65535 bytes. If that makes the BBcode invalid in certain ways, then you get the INVALID BBCODE error.
Also, note that a byte is different from a character. Because SMF stores "special" characters in a particularly space-inefficient way, each Russian character uses about 6 bytes, and each line break also uses 6 bytes.
Yes there's a limit/post as per theymos its 65535 bytes (65,535 ASCII/English characters). But when it comes to other languages like Arabic/Urdu it needs Unicode to save and size of it depends on the encoding for example:
UTF-16 - Each Character needs two-bytes
UTF-32 - Each Character needs four-bytes
I guess bitcointalk is using UTF-16 so the character limit for Urdu is 65535/2 ~ 32767 Characters in Unicode.Edit: I believe it's UTF-32 which means 4-bytes/character, I tested it in a PM (assuming PM/Posts have same character limit) and I'm able to send a big message with mix Urdu/English characters and it automatically got trimmed at 21,799.
Not sure why Russian characters needs 6 bytes? I think it should also take 1-4 bytes only.
You might want to check out this thread to understand more about this.