On Tue, Feb 07, 2012 at 09:38:39PM -0500, Robert Haas wrote:
> So we need some principled way of deciding how much inlining is
> reasonable, because I am 100% certain this is not going to be the last
> time someone discovers that a massive exercise in inlining can yield a
> nifty performance benefit in some case or another: index builds and
> COPY have already been mentioned on this thread, and I expect that
> people will find other cases as well. I'm not really sure what the
> "budget" is - i.e. how much binary bloat we can afford to add - or how
> many cases there are that can benefit, but the first isn't infinite
> and the second is more than the first.
>
> Having said all that, I am inclined to commit at least some portion of
> this, but I wanted to knock off a few other patches that have been
> lingering for a while first.
One approach would be to only do a few types now, e.g. integers and
strings, and perhaps leave the others for later.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ It's impossible for everything to be true. +