Re: 21.3.50 for Cygwin: patch 5

[storm@cua.dk (Kim F. Storm)]

Richard Stallman <rms@gnu.org> writes:

>     Defining P_ in config.h (via configure) is of course possible, but
>     even if we change everything (except the Glibc originated files) to
>     use P_, we still will have the __P definition in config.h
> 
> Why would we still have the __P definition in config.h?
> Is there some reason we could not delete it, once it was not used?

The Glibc files mktime.c, gmalloc.c, and strftime.c use it.  mktime.c
and strftime.c does declare it if it isn't defined already, but
gmalloc.c doesn't.

But even mktime.c and strftime.c use different ways to determine how
to define __P ...  Pretty messy.


BTW, if I simply checkout emacs from CVS, how do I know that some
files are _not_ supposed to be changed within the emacs project (such
as mktime.c and regex.c)?   

I still think we need a "tips and tricks" section in e.g. INSTALL-CVS
or admin/README (or a new file somewhere) to mention things a
developer should be aware of when working on emacs source code, notably:
 - requirement for legal papers,
 - what is a "tiny change" (and how they sum up)
 - how to document changes (ChangeLog, NEWS, manual pages)
 - indentation style, use of P_ macro, which header files to use
   (e.g. when should a function be added to lisp.h)
 - how we support multiple platforms [when to duplicate code,
   which files contain platform specific code, etc...]

but also:
 - quick intro to how we use texinfo

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk