[bug-mailutils] Namespace and pre-ansi compilers

[Alain Magloire]

> There are several things I'd like us to discuss before the next release. 
>
> The first issue regards the use of __P() wrappers around arguments in
function
> declarations in our header files. In my opinion, it is pretty
> useless now. First, anybody wishing to compile mailutils can get gcc (or
> other ansi-compatible compiler). Secondly, no special effort was ever
> made to make sure mailutils would compile with a pre-ansi
> compiler. Moreover, such an effort would be polluting the code. So, I
> propose to get rid of these wrappers. It will surely improve the
> code readability and facilitate the maintenance. 

Absolutely!  It is time we move in the 20th century 8-).
I think nowadays ANSI C should be the norm.  In the worst case, folks can
run the ansi2knr utility.

> The second issue regards mailutils namespace. Our naming rules, or
> better said, lack thereof, can create lots of problems when linking
> mailutils with some other projects. For example, such name types
> as list_t or record_t are pretty common. The same goes for function
> names list_create() and many others. My proposition is to prefix each
> external identifier with `mu_'.

Yes, agreed. It was an oversight.  Really all of our API should be prefix
with "mu_".  But we can do it by iterations.

> Finally, our main library is called `libmailbox', although it has little
> to do with mailboxes, being a collection of rather general-purpose
> utilities. Wouldn't it be better if we call it `libmailutils' (which is
> currently the name of our convenience library)? I understand that the name
> is the reason of historic processes and that this is not so important,
> and yet ... 

Another oversight 8-), mailbox was consider to be the sum of all.  But I
agree mailutils is a much better name for the sum of all.