Fix extensions_service_unittest on Linux....

Fix extensions_service_unittest on Linux.

I was getting failures which are not present on buildbot,
I don't know why. But with these changes it should be more solid.

This also prevents a segfault which I got (out of bounds array access).

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=6792

[Aaron Boodman, 2008-12-08 22:14:44]

Thanks for doing this. I think that we will probably want to keep the extensions
sorted in the real extensions service too, but I can do that separately.

http://codereview.chromium.org/13258/diff/1/2
File chrome/browser/extensions/extensions_service_unittest.cc (right):

http://codereview.chromium.org/13258/diff/1/2#newcode97
Line 97: // Make sure we won't access out-of-the-range items. Equality check is
at
I don't understand why you've moved the equality check to the end and replaced
them with these GE checks. What does that help with?

[Evan Stade, 2008-12-08 22:34:47]

http://codereview.chromium.org/13258/diff/1/2
File chrome/browser/extensions/extensions_service_unittest.cc (right):

http://codereview.chromium.org/13258/diff/1/2#newcode97
Line 97: // Make sure we won't access out-of-the-range items. Equality check is
at
On 2008/12/08 22:14:45, Aaron Boodman wrote:
> I don't understand why you've moved the equality check to the end and replaced
> them with these GE checks. What does that help with?

The at() calls will segfault without the ASSERT_GE check. Although he could have
simply changed EXPECT_EQ to ASSERT_EQ, but I guess this way will tend to give
more helpful error output in case of failure.

http://codereview.chromium.org/13258/diff/1/2#newcode103
Line 103: frontend->extensions()->at(0)->id());
nit: it might improve readability to pull out the frontend->extensions()->at(*)
calls

[please use my chromium address, 2008-12-11 14:54:53]

I made EXPECT_/ASSERT_ changes simpler, moving less things around. About the
nit: good idea, can I do it in a separate patch? This patch changes behavior,
while the second one would be a pure cleanup.

[Aaron Boodman, 2008-12-11 17:55:26]

On 2008/12/11 14:54:53, phajdan.jr wrote:
> I made EXPECT_/ASSERT_ changes simpler, moving less things around. About the
> nit: good idea, can I do it in a separate patch? This patch changes behavior,
> while the second one would be a pure cleanup.

LGTM

Thanks!