Re: Adding a generic function layer inside OPERATE

Nikodemus Siivola <[email protected]> writes:

>  * I recall Dan having said something about how the current way
>    of first traversing and then operating was a hack done to
>    circumvent the Clisp lack of custom method combinations --
>    which it now has. (This is more of a historical note then
>    anything else: I think a change at that level would be best
>    left for YASDF, ASDF2, or whatever.)

I think that this refers to the /implementation/ of TRAVERSE, rather
than the idea behind it, which I think comes from Kent Pitman's paper
about system definition.

(I also have needed to wrap system operations with some dynamic state,
and had to work around the problem, usually by wrapping each file
operation and maintaining some global state instead; it is annoying,
and I await ASDF2 with bated breath.)

Cheers,

Christophe