Re: Dumb identifier name question

[Owen Taylor <otaylor redhat com>]

Robert Graham Merkel <rgmerk mira net> writes:

> Hi all,
> 
> I'm currently in the process of implementing a priority queue which I hope
> can be incorporated into glib in the future.  
> 
> I'm trying to copy the coding style from the gtk library, so I had a look
> at how glib declares its data types.  In the process, I noticed something
> that, if I've understood correctly, violates the C standard.
> 
> The include guards on the header files are of the form
>                                                                                 
> #ifndef __G_FOO_H__
> #define __G_FOO_H__
> 
> /* . . . */
> 
> #endif
> 
> And the typedefs for glib structure are of the form
> 
> typedef struct _GBar GBar;
> 
> According to K&R 2nd edition appendix B (p 241), introducing the C
> standard library, "External identifiers that begin with an underscore are 
> reserved for use by the library, as are all other identifiers that begin 
> with an underscore and an upper-case letter or another underscore".

Basically, it's been this way since 1996, and nobody has run
into a non-theoretical complaint, so we'd rather not:

 - break consistency between existing code and new code

Or:

 - change 600,000 lines of code to do something different

Since we maintain our namespace in these cases, the chances of collision
are pretty tiny -- larger for GLib (with the _G_ namespace) than for
Pango (_PANGO_( or GTK+ (_GTK_), but still small.

> I realise the practical chances of this causing a problem
> are small, but then again the effort required to fix it is also basically
> janitorial.
 
Something can be "basically janitorial" and still a huge job. 
Not worth doing in the absence of concrete problems.

Regards,
                                        Owen