[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: stop using P_, __P in header files

From: Dan Nicolaescu
Subject: Re: stop using P_, __P in header files
Date: Mon, 05 Jul 2010 01:51:28 -0400
User-agent: Gnus/5.11 (Gnus v5.11) Emacs/22.1 (gnu/linux)

Ken Raeburn <address@hidden> writes:

> On Jul 4, 2010, at 12:46, Dan Nicolaescu wrote:
>> There are some remaining issues to solve:
>> - DEFUNs need to be converted by hand, protoize does not know anything about 
>> them.
> (By *hand*??  Ugh.  We really need a tool that implements some kind of 
> editing macros... :-)

The macros don't just exist, so manual work is needed as opposed to
just running protoize for most of the rest. :-)

> Also, I believe make-docfile scans the argument lists; it may need to be 
> taught about the new syntax.

Good point.  If you are familiar with that code, please do it.

> Or, the explicit old-style argument declarations can go away, and
> DEFUN can be taught how to expand a list of argument names into a
> list of new-style argument declarations.  I thought about doing this
> back in May when we were discussing the DOC file name handling and
> version number definition; I think it would require making a bunch
> of helper macros for each MAXARGS value that could get passed.  (I
> was thinking about it in the context of putting the doc strings in a
> section of the executable that only gets paged in when needed on
> most platforms, rather than having to copy them to and then load
> from a separate file.)

>> - the error and message functions in lisp.h are called with variable
>>  number of arguments, but are defined with a fixed number of arguments.
> They should probably be fixed to be standard variadic functions, which means 
> doprnt() has to be taught about va_arg.  I can take a shot at that, if no one 
> else feels like it; it's probably easy.

Please do. [I have a patch to turn on -Wstrict-prototypes by default
if available, but it would produce 4 warnings per file including
lisp.h, so it won't be very popular until the above is fixed].

Something also needs to be done about the type for  lisp.h:Lisp_Subr.function

> We could also probably ditch the "register" declarations that date
> back to pre-GCC days; do they do *any* good now?

Very likely no.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]