Change the default number of proxy resolver threads used for evaluating PAC s...

Change the default number of proxy resolver threads used for evaluating PAC scripts from 1 to 4.

Also adds a command line flag to override the default: --num-pac-threads=X

BUG=11079
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=52046

[wtc, 2010-07-09 21:44:09]

LGTM.

http://codereview.chromium.org/2893005/diff/11002/1008
File chrome/browser/net/chrome_url_request_context.cc (right):

http://codereview.chromium.org/2893005/diff/11002/1008#newcode99
chrome/browser/net/chrome_url_request_context.cc:99: size_t num_pac_threads =
0u;  // Use default number of threads.
Nit: I would just use 0 for unsigned constants except in those DCHECK_EQ
templates that require exact type match.  Note that your own documentation
of the function says "|0|",  too.

http://codereview.chromium.org/2893005/diff/11002/1013
File net/proxy/proxy_service.cc (right):

http://codereview.chromium.org/2893005/diff/11002/1013#newcode273
net/proxy/proxy_service.cc:273: num_pac_threads = 4u;
The 4 constant should be declared or defined in the header file or at the top
of this file to make it easier to find.

http://codereview.chromium.org/2893005/diff/11002/1013#newcode304
net/proxy/proxy_service.cc:304: new ProxyConfigServiceFixed(pc), new
ProxyResolverNull, NULL);
Is this an unrelated change?  Just curious.

http://codereview.chromium.org/2893005/diff/11002/1014
File net/proxy/proxy_service.h (right):

http://codereview.chromium.org/2893005/diff/11002/1014#newcode146
net/proxy/proxy_service.h:146: // PAC script takes a while to run. This is a
particular problem when PAC
Nit: a particular problem => particularly a problem

http://codereview.chromium.org/2893005/diff/11002/1014#newcode151
net/proxy/proxy_service.h:151: //   (a) can cause compatibility issues for
scripts that rely on sideffects
Typo: sideffects => side effects

[eroman, 2010-07-09 22:02:48]

http://codereview.chromium.org/2893005/diff/11002/1008
File chrome/browser/net/chrome_url_request_context.cc (right):

http://codereview.chromium.org/2893005/diff/11002/1008#newcode99
chrome/browser/net/chrome_url_request_context.cc:99: size_t num_pac_threads =
0u;  // Use default number of threads.
On 2010/07/09 21:44:09, wtc wrote:
> Nit: I would just use 0 for unsigned constants except in those DCHECK_EQ
> templates that require exact type match.  Note that your own documentation
> of the function says "|0|",  too.

Done.

http://codereview.chromium.org/2893005/diff/11002/1013
File net/proxy/proxy_service.cc (right):

http://codereview.chromium.org/2893005/diff/11002/1013#newcode273
net/proxy/proxy_service.cc:273: num_pac_threads = 4u;
On 2010/07/09 21:44:09, wtc wrote:
> The 4 constant should be declared or defined in the header file or at the top
> of this file to make it easier to find.

Done.

http://codereview.chromium.org/2893005/diff/11002/1013#newcode304
net/proxy/proxy_service.cc:304: new ProxyConfigServiceFixed(pc), new
ProxyResolverNull, NULL);
On 2010/07/09 21:44:09, wtc wrote:
> Is this an unrelated change?  Just curious.

Restored to earlier format, this isn't related. Will address that issue in a
follow-up CL.

http://codereview.chromium.org/2893005/diff/11002/1014
File net/proxy/proxy_service.h (right):

http://codereview.chromium.org/2893005/diff/11002/1014#newcode146
net/proxy/proxy_service.h:146: // PAC script takes a while to run. This is a
particular problem when PAC
On 2010/07/09 21:44:09, wtc wrote:
> Nit: a particular problem => particularly a problem

Done.

http://codereview.chromium.org/2893005/diff/11002/1014#newcode151
net/proxy/proxy_service.h:151: //   (a) can cause compatibility issues for
scripts that rely on sideffects
On 2010/07/09 21:44:09, wtc wrote:
> Typo: sideffects => side effects

Done.