> The original issue was that SLIME would define its own version of > read-directory-name if it wasn't already fbound, but this version had > slightly different semantics to the standard read-directory-name[1]. This > is why I implemented tla--read-directory-name in terms of > read-file-name--we can't rely on read-directory-name to behave > correctly, even if it is present.
Is it issue of xtla? I think it is should be solved in SLIME. Because read-directory-name is part of standard emacs function now. What can we do if car and cdr is hacked in the other package? Should we provide our own replacement? Give me comments.
