Re: darwin.c identifier-smashing

[Geoffrey Keating <geoffk at apple dot com>]

On 09/07/2004, at 3:20 PM, Mark Mitchell wrote:

> Geoff, Mike --
>
> There's some quadratic behavior in darwin.c that needs fixing.
> However, I'm a bit puzzled about how to do that, exactly, due to:
>
> 	      /* FIXME: This breaks the identifier hash table.  */
> 	      IDENTIFIER_NODE_CHECK (TREE_VALUE (temp))->identifier.id.str
> 		= (unsigned char *) name;
>
> What I don't understand is why this isn't just:
>
>   TREE_VALUE (temp) = get_identifier (name);
>
> Why is it important to stomp over an existing IDENTIFIER_NODE?
>
> The ideal reply would take the form of a patch with a comment
> explaining the issue. :-)

Because there may be more than one reference to the IDENTIFIER_NODE, of 
course.  The darwin backend is using the IDENTIFIER_NODE to store data 
about the thing that it names by having magic characters in the name.  
It's all really horrible and should be replaced by using 
SYMBOL_REF_FLAGS and maybe look-aside structures if necessary.

Attachment: smime.p7s
Description: S/MIME cryptographic signature