Masatake YAMATO <[EMAIL PROTECTED]> writes:
>> How does Masatake's completion mechanism work?
>
> I cannot explain well. It means Matthieu's one is much better than
> mine:) I read his code and it is obvious that his code is much
> better than mine.
Thanks :-)
Well, I tried to keep the code as simple as possible. I've continued
the work, and currently got something fully featured for less than 200
lines of code.
> However, mine has some advantages(I know Matthieu wrote his code
> was not completed):
>
> 1. Separation "--" and "/" handling
> His completion engine didn't take care about it.
It now does.
[EMAIL PROTECTED]
[EMAIL PROTECTED]/
[EMAIL PROTECTED]
[EMAIL PROTECTED]
(But the "--" wouldn't have been added if you ask only for a category)
> 2. Tree building control
> His completion engine always used cached data in the
> archive-tree. In my code, typing "return" means
> rebuilding cache.
I don't like the idea of redefining the behaviour of RET in a reading
engine: In Emacs, RET means "I have finished entering text", and
that's the user will want to tell to xtla by pressing RET.
In my version, typing C-r refreshes the cache.
> 3. Subtree completion
> In his completion engine, the user always had to input
> from archive. Even xtla requires just revision name.
This is solved also. I'm using an initial input instead of a read-only
text like you did. Same as above, this is the usual behavior of Emacs,
and I was surprised by the behavior of your engine. furthermore, there
are many cases where the user may change his mind and wish to give
another text from scratch.
> I've solved 1 and 2.
> You can force rebuild archive-tree by C-u tab.
Good idea also. Perhaps more intuitive than my C-r binding.
> To avoid confusion, I have not commit my modification to my tree.
Ooops, I'm really sorry : I had read your message too quickly and
didn't understand that you actually wrote some code.
--
Matthieu
