From: Eric Wong <normalperson@...>
Date: 2014-11-06T00:21:37+00:00
Subject: [ruby-core:66108] Re: [ruby-trunk - Feature #10423] [PATCH] opt_str_lit*: avoid literal string allocations

ko1@atdot.net wrote:
> Sorry for late, and thank you for catching up recent changes.

No problem.  I had some time today.  I will be busy and have little
computer time for the next 5-7 days.

> My comments:
> 
> 1. (negative) Incompatibility

> To solve this incompatibility, I have several ideas.
> 
> (1-1) Make new instruction to replace with receiver.

I added a new opt_str_lit_yoda instruction.  It only handles one receiver
location for now, I'm not sure if other cases (e.g. gsub) can trigger
the incompatibility.

patch: http://80x24.org/spew/m/38d41def78f16bca14fbc7fccd160ef0a8d06479.txt

Right now, I do not optimize `str.gsub("bar", method_call)', yet,
only optimizing: `str.gsub(anything, "bar")'

	"foo" == bar

Compiles to:

  [:putobject, "foo"],
  [:putself],
  [:opt_send_without_block,
   {:mid=>:bar, :flag=>280, :orig_argc=>0, :blockptr=>nil}],
  [:opt_str_lit_yoda, ["foo", 17, 1]],
  [:opt_eq, {:mid=>:==, :flag=>256, :orig_argc=>1, :blockptr=>nil}],

> (1-3) Ignore such incompatibility because nobody write such code

However, I hope we can ignore the strange case and new insn.

> 2. (negative) Live patching

I've removed live patching for now:
  http://80x24.org/spew/m/86739be08df5b4128e84bce15a2853cd7947a6fa.txt
It was a good learning experience, at least.

> 3. (neutral/not big issue) target methods
> 
> opt_method.def format is a bit difficult to read which methods are
> target. Maybe we can make more pretty like:

I prefer to leave the format unchanged for now, because compile.c still
uses id{PLUS,ASET,AREF} names.

> 4. (neutral/not big issue) warning on VC (windows)

Untested patch (cast to unsigned):
  http://80x24.org/spew/m/30c309d1ad24319ad53813f8814e09be0554ccb0.txt

> Have you measure performance improvement on non-microbenchmarks?

"make rdoc" performance improved from 66.3 => 63.3 sec in original test

However, the difference seems is smaller now: 66.1 => 64.9 sec

Full patch for convenience:
http://80x24.org/spew/m/opt-str-lit-v7%40m.txt