>Works fine on AIX 4.1.5.0.01. I cleaned up four compiler warnings produced
>by gcc -Wall. One of them required adding an #include of <stdlib.h>, and
>this required me to change getpass() to take a const char* rather than a
>char*, to match the system prototype. Is this how getpass() is declared
>everywhere? If not, we'll need to add autoconf support to check the
>declaration or (preferably) change this function name to be nmh_getpass().
I think we should change the name. Our function's whole point is to
obviate the need for any system-specific getpass(), so we shouldn't
have to adhere to their prototypes unless we really want to. The only
reason why I didn't stick with const char* was that I didn't see the
relevance of being so specific. It's not that important.
>I notice the memory allocated by calloc() is leaked. Not that big a deal
>for nmh's short-running programs, I guess, but it might at least be worth a
>comment acknowledging the leak and giving the justification.
I only added the explicity calloc because my compiler (granted,
without -Wall), was complaining about getpass returning a local
variable. instead of declaring buf[SIZE], if I manually calloc'ed it,
the warning went away. I'll fix this up in the next few days.
>The copyright notice sort of implies that we're using the unchanged
>version. We should probably add "Portions of this code are" to the
>beginning, eh?
Sounds good.
>By all means. Ideally we'd be able to get rid of MACHINES altogether.
OK. I seem to recall seeing some programs say "You must use gcc", but
we should be able to get around that.
Shantonu
*-------------------------------------------*
|Shantonu Sen * (617)225-6778 * [EMAIL PROTECTED]|
|Electrical Engineering and Computer Science|
|Massachusetts Institute of Technology, 2002|
*-------------------------------------------*
