Preload the visited link db on the file thread.

Preload the visited link db on the file thread if the file exists.
Otherwise, just load like normal on the UI thread.

BUG=24163
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=34874

[brettw, 2009-12-16 00:54:51]

http://codereview.chromium.org/502005/diff/2001/3001
File chrome/browser/browser_main.cc (right):

http://codereview.chromium.org/502005/diff/2001/3001#newcode744
chrome/browser/browser_main.cc:744: profile->PreloadVisitedLinkMaster();
Can you write a comment about why this is here and what (or say there are none)
other dependencies this has on other stuff being initialized in this function?
If somebody refactors this initialization code, I want to be sure they know if
they can or can't move it somewhere else.

http://codereview.chromium.org/502005/diff/2001/3002
File chrome/browser/profile.cc (right):

http://codereview.chromium.org/502005/diff/2001/3002#newcode184
chrome/browser/profile.cc:184: // VisitedLinkMaster from the UI thread.
It seems bad that we'll try to load the file twice in the failure case. This
also happens the first run, which will slow it down just a tad. Can we do
anything to prevent this?

[tony, 2009-12-16 19:23:25]

take another look?

http://codereview.chromium.org/502005/diff/2001/3001
File chrome/browser/browser_main.cc (right):

http://codereview.chromium.org/502005/diff/2001/3001#newcode744
chrome/browser/browser_main.cc:744: profile->PreloadVisitedLinkMaster();
On 2009/12/16 00:54:51, brettw wrote:
> Can you write a comment about why this is here and what (or say there are
none)
> other dependencies this has on other stuff being initialized in this function?
> If somebody refactors this initialization code, I want to be sure they know if
> they can or can't move it somewhere else.

Done.

http://codereview.chromium.org/502005/diff/2001/3002
File chrome/browser/profile.cc (right):

http://codereview.chromium.org/502005/diff/2001/3002#newcode184
chrome/browser/profile.cc:184: // VisitedLinkMaster from the UI thread.
On 2009/12/16 00:54:51, brettw wrote:
> It seems bad that we'll try to load the file twice in the failure case. This
> also happens the first run, which will slow it down just a tad. Can we do
> anything to prevent this?

Ok, refactored to not call InitFromFile twice, although I don't think it would
have negatively impacted start up since the extra call was on a different
thread.

[tony, 2009-12-17 18:12:47]

brettw timed out.  trying other reviewers.

[jam, 2009-12-17 19:09:06]

hey, I'm not that familiar with the code, and I think Brett is here today,
so I'd like to defer to him

On Thu, Dec 17, 2009 at 10:12 AM, <tony@chromium.org> wrote:

> brettw timed out.  trying other reviewers.
>
>
> http://codereview.chromium.org/502005
>

[brettw, 2009-12-17 19:16:15]

LGTM with two nits. Thanks for getting this finally fixed.

http://codereview.chromium.org/502005/diff/3008/2005
File chrome/browser/visitedlink_master.cc (right):

http://codereview.chromium.org/502005/diff/3008/2005#newcode570
chrome/browser/visitedlink_master.cc:570: bool
VisitedLinkMaster::InitFromScratch() {
Can you move the definitions of these two functions you made public to below
Init so they match the order of the header?

http://codereview.chromium.org/502005/diff/3008/2006
File chrome/browser/visitedlink_master.h (right):

http://codereview.chromium.org/502005/diff/3008/2006#newcode79
chrome/browser/visitedlink_master.h:79: // until this is called. Returns true on
success, false means that this
Can you update this comment to point out that you can instead call
InitFromFile/InitFromScratch if you need more control.