>>>> "Klaus" == Klaus Berndl <Berndl>
writes:
>>>> "Klaus" == Klaus Berndl <Berndl>
writes:
Klaus> No, here i do not agree with you and Stephen. Reasons:
Klaus> - Changing current `window-list' so it always returns a somehow canonical
Klaus> ordered list is very simple and has really no disadvantage (at least as fas
Klaus> i can see)
Klaus> >Yes, it has, namely that people tend to then rely on that canonical
Klaus> >order, even though it's not part of the natural specification---this
Klaus> >makes it impossible to change later, should it become opportune to do
Klaus> >so. (If it's one thing XEmacs history has taught us, then that's
it,
Klaus> >thank goodness.)
Klaus> Hmm, IMHO a somehow "contructed" argument and disadvantage.
Nonsense.
Klaus> "`next-window' is a built-in function
Klaus> (next-window &optional WINDOW MINIBUF WHICH-FRAMES WHICH-DEVICES)
Klaus> Documentation:
Klaus> Return the next window after WINDOW in the canonical ordering of windows.
Klaus> If omitted, WINDOW defaults to the selected window.
Klaus> ..."
Klaus> You see the word "canonical"??!!
Yes, I do. You see that word somewhere in the documentation or name
for `window-list'?
Klaus> If i understand you right then you say it is best to make no
Klaus> specification for API-functions on which programmers can rely
Klaus> because maybe someone wants to change it later and then all
Klaus> already existing code fails or has to be rewritten, right?
No, you didn't understand me right. I didn't say anything of the
sort.
`window-list' already has a specification, and there's no good reason
to change it, because there's no existing code that assumes a
different specification from it. If you want a function with a
different behavior, by all means write it and give it a name
indicative of what it does.
--
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla