Re: [trunk][patch] Don't print warnings in mangle_decl

[Diego Novillo <dnovillo at google dot com>]

2009/4/16 Mark Mitchell <mark@codesourcery.com>:

> I don't see how we can have both of laziness and errors at the point of declaration.

Another thing to consider is that we can only be lazy until we hit the
hand-off point from the FE into the ME.  We can no longer wait until
very late in the RTL phase.  I don't really see a way out of mangling
relatively early.

Currently, in LTO, we mangle at the point of pass_ipa_free_lang_data,
which happens right after the full callgraph for the current TU is
built (this will be part of the LTO merge that I'm preparing).

One possible alternative would be to compress the pool of mangled
strings and only expand it on demand when we need to write to asm
output.  Anywhere else we could use either hash values or indices into
the compressed pool string.


Diego.