Avoid printing the full token stream and doing line/column position calculations when extracting th…

Avoid printing the full token stream and doing line/column position calculations when extracting the source of a function where we don't have the original source.

Brings mirrors_reader_test from 2 minutes to 11 seconds on the standalone VM.

R=iposva@google.com

Committed: https://code.google.com/p/dart/source/detail?r=32461

[Ivan Posva, 2014-02-08 01:09:17]

LGTM

Please remove the SLOW marker on the relevant tests.

-Ivan

[rmacnak, 2014-02-08 01:18:23]

Committed patchset #3 manually as r32461 (presubmit successful).

[hausner, 2014-02-10 16:49:52]

LGTM w/comment

https://codereview.chromium.org/133813008/diff/80001/runtime/lib/mirrors.cc
File runtime/lib/mirrors.cc (right):

https://codereview.chromium.org/133813008/diff/80001/runtime/lib/mirrors.cc#n...
runtime/lib/mirrors.cc:2050: // doing expensive position calculations.
Maybe mention in the comment why it makes no sense to do the calculation below:
column positions are not available.

Maybe this faster path is good enough even when we have sources? What would we
lose?

[rmacnak, 2014-02-10 23:42:17]

On 2014/02/10 16:49:52, hausner wrote:
> LGTM w/comment
> 
> https://codereview.chromium.org/133813008/diff/80001/runtime/lib/mirrors.cc
> File runtime/lib/mirrors.cc (right):
> 
>
https://codereview.chromium.org/133813008/diff/80001/runtime/lib/mirrors.cc#n...
> runtime/lib/mirrors.cc:2050: // doing expensive position calculations.
> Maybe mention in the comment why it makes no sense to do the calculation
below:
> column positions are not available.
> 
> Maybe this faster path is good enough even when we have sources? What would we
> lose?

We lose user formatting and comments. We can do better on this path as well by
doing token-position -> source-offset-position instead of token-position ->
line/column-position -> source-offset-position.