Comments by Raicho Nikolov:

I think I have an idea which can help you to solve the problem.

I have made a filter which put blockquotes without a problem.  Txt2tags naturally puts <p> tag inside. The filter is this:

%!postproc(xhtml):  {"  <blockquote>
%!postproc(xhtml):  "}  </blockquote>

So I have to write:

""" {"
alabala

""" "}

to have <blockquote> block

That's ugly and is not my suggestion - it only shows how it works. (although it has one advantage over the "natural way" - you 
can use it in web pages)

The idea is to have some close element for this kind of blocks. For example for lists we have a closing with two blank lines. 
Thus the program will know where the element starts and where the element ends. 

Opening and closing tag must differ - for example tab opens and two blank lines close. This way the markup will not change 
much. Probably this is doable with Pre-Post-Processing.

I hope that I am on the right way and you get the idea
Raicho

Check if in XHTML we can omit the P tag if there's only one paragraph inside it: <blockquote>Foo 
bar.</blockquote>.

If yes, an ideal solution would be to add the P tags only if there's two or more paragraph inside the quote.

But this could cause CSS troubles since we would have BLOCKQUOTE with and without P tags inside. Maybe it's 
best to just add it everytime, so the user knows it's always there.

Another change to the txt2tags would be to detect consecutive quotes (even with blank lines between) and join 
them in one big quote (fixing a longstanding txt2tags limitation of one-paragraph-only quotes).

It's easy to do in the current code, checking for BlockMaster.last contents.