Ronan Waide <[EMAIL PROTECTED]> writes: > On January 12, [EMAIL PROTECTED] said: >> >> I am not sure, but it seems to me that autoloads are more desireable, >> 'cause then we defer the possible need/availability check to runtime. > > Yep, except the compiler warnings I was attempting to hush up are for > variables, not functions/macros, and autoloading doesn't appear to > help there... the other option is to defvar the relevant variables, > but I can't say that appeals to me from an aesthetic point of view.
This is exactly the kind of thing which will screw other code in the long run. That other package might test (boundp 'foo) and we used defvar foo to silence the compiler... It might work (and many people do this), but we'll need to really make sure that such a defvar is harmless. Personally, compiler warnings which warn about non-existing variables and functions are ok by me -- my code has plenty of those. And I think/hope that's because I compile the code for Emacs and the variables and functions are XEmacs-only. Anyway, not an easy decision to make. Alex. -- http://www.emacswiki.org/ _______________________________________________ [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/bbdb-info BBDB Home Page: http://bbdb.sourceforge.net/
