Do not inherit from std::vector.

Do not inherit from std::vector.

vector does not have a virtual destructor, see Item 7 from of Effective STL from
Scott Meyers for a detailed explanation.

BUG=135335
R=markusheintz@chromium.org,willchan@chromium.org
TBR=jam@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=148315

[markusheintz_, 2012-07-23 15:04:20]

Looks good from my perspective.

[wtc, 2012-07-23 19:34:07]

Drive-by review of patch set 1:

Please wait for willchan's review.

I reviewed the changes to src/net.  They seem fine to me.
The only downside is that we can't forward-declare CookieList
any more.

[willchan no longer on Chromium, 2012-07-23 20:33:10]

lgtm. I shared wtc's concerns, but I think good style is more important here.

That said, just for your clarification, inheritance from classes within virtual
destructors is only dangerous if you try to delete via the base class pointer.
Since we weren't doing that here, it was safe. But a typedef is probably better
anyway.

[erikwright (departed), 2012-07-30 18:36:43]

wtc, willchan:

Thanks to battre's recent contributions, only canonical_cookie.h (not
cookie_monster.h) need be included. That's a nice, concise header file, so not
as bad as it would have been a few weeks ago.

That being said, I don't even see much of a benefit to the typedef vs. an
explicit std::vector<net::CanonicalCookie> (which _would_ allow for forward-decl
of CanonicalCookie) but perhaps I'm just a fast typist.

[willchan no longer on Chromium, 2012-07-30 21:01:38]

In all honesty, the effect of this change is minor enough in either
direction that it's not worth discussing :)


On Mon, Jul 30, 2012 at 11:36 AM, <erikwright@chromium.org> wrote:

> wtc, willchan:
>
> Thanks to battre's recent contributions, only canonical_cookie.h (not
> cookie_monster.h) need be included. That's a nice, concise header file, so
> not
> as bad as it would have been a few weeks ago.
>
> That being said, I don't even see much of a benefit to the typedef vs. an
> explicit std::vector<net::**CanonicalCookie> (which _would_ allow for
> forward-decl
> of CanonicalCookie) but perhaps I'm just a fast typist.
>
>
http://codereview.chromium.**org/10796081/<http://codereview.chromium.org/107...
>