This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Compiling GCC with a C++ compiler (3/n)
On Mon, Aug 02, 2004 at 07:47:50PM +0200, Gabriel Dos Reis wrote:
> While I agree that the "u." thingy might be ugly (as C does not have
> the notion of unnamed union), I'm not convinced that typing "rt_rtx"
> is really annoying. Now, GDB has support for macros, and one ca use
> XEXP(...) instead of fiddling directly with the fields.
Sure, but not all cc1's around are compiled with -g3 and without -g3
you can't use that. -g3 output is huge (on my box -g3 cc1 is 71MB,
-g less than 19MB), so e.g. for keeping multiple cc1's around -g3
is not a good option.
> | union rtunion_def
> | {
> | int rtint;
> | unsigned int rtuint;
> | const char *rtstr;
> | struct rtx_def *rtx;
> | };
>
> I considered this alternative, but after some thoughts, decided not to
> go down that route: It is adding more places to look for, in order to
> keep the meaning of "rtx" in sync.
I think it would be far less harmful, it is highly unlikely rtx typedef
will change, there is grep which can easily reveal where struct rtx_def
is used and it is very close to its definition anyway.
Jakub