Add a stub implementation of sqlite3Preload when building with system sqlite.

Add a stub implementation of sqlite3Preload when building with system sqlite.

This fixes a compile problem when using -Duse_system_sqlite=1

TEST=none
BUG=22208
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=57578

[Evan Martin, 2010-08-26 19:27:26]

I am still sad that this will silently regress performance.
Perhaps it'd be worth logging something here?
What do you think?

[Paweł Hajdan Jr., 2010-08-26 19:46:28]

On 2010/08/26 19:27:26, Evan Martin wrote:
> I am still sad that this will silently regress performance.
> Perhaps it'd be worth logging something here?
> What do you think?

I was talking with Brett about that. The performance regression is really small,
measured informally. I considered adding NOTIMPLEMENTED() here, but then we'd
drag logging.h or have to add a .cc file, which would make the build more
complicated.

[Evan Martin, 2010-08-26 20:48:27]

On 2010/08/26 19:46:28, Paweł Hajdan Jr. wrote:
> I was talking with Brett about that. The performance regression is really
small,
> measured informally. I considered adding NOTIMPLEMENTED() here, but 

I think it only really affects you with a code disk cache.

Here's a one-liner I use sometimes when testing:

$ cat /home/evanm/.local/bin/drop-caches 
sudo sh -c 'echo 1 > /proc/sys/vm/drop_caches'


LGTM in any case

[Evan Martin, 2010-08-26 20:48:53]

On 2010/08/26 20:48:27, Evan Martin wrote:
> I think it only really affects you with a code disk cache.

Er, I meant *cold* disk cache.  :)

[Paweł Hajdan Jr., 2010-08-26 20:54:00]

I was running those tests with dropped caches every time. Will commit.

On Thu, Aug 26, 2010 at 13:48, <evan@chromium.org> wrote:

> On 2010/08/26 20:48:27, Evan Martin wrote:
>
>> I think it only really affects you with a code disk cache.
>>
>
> Er, I meant *cold* disk cache.  :)
>
>
>
> http://codereview.chromium.org/3134039/show
>