This is the mail archive of the
mailing list for the GCC project.
Re: [incremental] Patch: FYI: defer dwarf macro output
>>>>> "Daniel" == Daniel Jacobowitz <email@example.com> writes:
Daniel> On Thu, Dec 20, 2007 at 12:58:31PM -0700, Tom Tromey wrote:
>> This patch fixes the problem by deferring all macro output until the
>> end of compilation.
Daniel> Doesn't this increase the memory we have to keep live during
Daniel> compilation quite a bit? Long-term, I'd think we would want to dump
Daniel> it out sooner rather than later so we could forget about it.
Yeah, it probably does affect memory use negatively. I can make it
emit this stuff immediately in the non-server case. Sorry about this,
I should have done it this way from the start.
This sort of thing will probably be an ongoing struggle with the
incremental compiler. So far I've resisted having any explicit notion
of "server mode" versus "batch mode" in most of the compiler (partly
because it is nicer to always be testing the new code), but it may
come to that eventually.