Matthieu Moy <[EMAIL PROTECTED]> writes: > I was more talking about the user's point of view. I may have a > particular philosophy here, but I think command line tools are much > more pedagogical than GUI or Emacs interface. > > The interesting concepts are in tla itself. Xtla is just a more > pleasant way to use it. I think that if people learn tla starting with > a front-end, they'll feel their learning curve will grow quickly, but > at the end, I'm not sure they'll lear so fast. > > If you look at PCL-CVS docstrings, I think they are an example to > follow: > > ,----[ C-h f cvs-update RET ] > | cvs-update is an interactive compiled Lisp function in `pcvs'. > | (cvs-update DIRECTORY FLAGS) > | > | Run a `cvs update' in the current working DIRECTORY. > ^^^^^^^^^^^^^^^^^^ > This is the important part. It gives you the link to the original > tool. You can read the doc about cvs update somewhere else. > > | Feed the output to a *cvs* buffer and run `cvs-mode' on it. > | With a prefix argument, prompt for a directory and cvs FLAGS to use. > | A prefix arg >8 (ex: C-u C-u), > | prevents reuse of an existing *cvs* buffer. > `----
I share your opinion. When I think about it, I like pcl-cvs because I know exactly, which cvs commands are invoked from emacs. I started xtla.el to get a feeling about tla. I remember that Masatake sayed something similar. Now I am still learning the concepts for arch/tla. So I think about ways to make it easier to learn arch/tla. > It reminds me of a collegue to whom I asked "What C compiler did you > use at school" : "Err, I used F9". He didn't realize that the > environment in which he was working was at the same time an editor and > a compiler, and he didn't know which tool he was using. Just the key > to press. OK, he could write his first C program faster than he would > have done using vi and gcc, but then ? ... Nice comparison. Stefan.
