A build fix for gcc 4.4 (r32522)....

A build fix for gcc 4.4.
When calling a non-static member function without calling contructors, gcc 4.4 returns a warning "'<anonymous>' is used uninitialized in this function."
To fix this warning, this change added a 'static' keyword to CanonicalizeHost().

BUG=none
TEST=build chromium of gcc 4.4.

[ian fette, 2009-11-20 15:08:06]

LGTM

It's good to make this static, as it does not depend on any state in the class.
That said, I'm surprised that GCC4.4 is throwing any warnings here. Is this
function actually being used anywhere as a non-static member function without
calling any constructors? It's a private function that is only called by the
(non-static) overloaded () operator, which is called on an instance of the
object passed in to lower_bound, so at this point the (default) constructor
would have been called - unless GCC is inlining the () operator function because
the object has no state?

Anyhow, I'm fine with this change, but would be curious to understand why this
is an error.

[Evan Martin, 2009-11-20 18:52:31]

LGTM

http://codereview.chromium.org/414025/diff/1/2
File chrome/browser/cookies_tree_model.cc (right):

http://codereview.chromium.org/414025/diff/1/2#newcode61
Line 61: bool operator() (const CookieTreeNode* lhs,
May as well make this one static too?

[Hironori Bono, 2009-11-23 10:03:46]

Sorry for my slow response.
Is it possible to land this change on my behalf? (I cannot land this change
until tomorrow.)

Regards,

Hironori Bono

On 2009/11/20 18:52:31, Evan Martin wrote:
> LGTM
> 
> http://codereview.chromium.org/414025/diff/1/2
> File chrome/browser/cookies_tree_model.cc (right):
> 
> http://codereview.chromium.org/414025/diff/1/2#newcode61
> Line 61: bool operator() (const CookieTreeNode* lhs,
> May as well make this one static too?

[ian fette, 2009-11-23 17:20:36]

Evan -

Making operator () static breaks the build.

c:\b\slave\win\build\src\chrome\browser\cookies_tree_model.cc(62) : error C2801:
'`anonymous-namespace'::OriginNodeComparator::operator ()' must be a non-static
member

On 2009/11/23 10:03:46, hbono wrote:
> Sorry for my slow response.
> Is it possible to land this change on my behalf? (I cannot land this change
> until tomorrow.)
> 
> Regards,
> 
> Hironori Bono
> 
> On 2009/11/20 18:52:31, Evan Martin wrote:
> > LGTM
> > 
> > http://codereview.chromium.org/414025/diff/1/2
> > File chrome/browser/cookies_tree_model.cc (right):
> > 
> > http://codereview.chromium.org/414025/diff/1/2#newcode61
> > Line 61: bool operator() (const CookieTreeNode* lhs,
> > May as well make this one static too?

[ian fette, 2009-11-23 18:56:58]

I chatted with Evan, I am going to land this without making bool operator ()
static (as that breaks the build). No idea if this will fix GCC4.4 or not, it
seems rather broken that this is being reported as a bug.

Will land as soon as the tree opens, as
http://codereview.chromium.org/436003/show

On 2009/11/23 17:20:36, ian fette wrote:
> Evan -
> 
> Making operator () static breaks the build.
> 
> c:\b\slave\win\build\src\chrome\browser\cookies_tree_model.cc(62) : error
C2801:
> '`anonymous-namespace'::OriginNodeComparator::operator ()' must be a
non-static
> member
> 
> On 2009/11/23 10:03:46, hbono wrote:
> > Sorry for my slow response.
> > Is it possible to land this change on my behalf? (I cannot land this change
> > until tomorrow.)
> > 
> > Regards,
> > 
> > Hironori Bono
> > 
> > On 2009/11/20 18:52:31, Evan Martin wrote:
> > > LGTM
> > > 
> > > http://codereview.chromium.org/414025/diff/1/2
> > > File chrome/browser/cookies_tree_model.cc (right):
> > > 
> > > http://codereview.chromium.org/414025/diff/1/2#newcode61
> > > Line 61: bool operator() (const CookieTreeNode* lhs,
> > > May as well make this one static too?

[ian fette, 2009-11-23 19:43:03]

Landed as r32829.

On 2009/11/23 18:56:58, ian fette wrote:
> I chatted with Evan, I am going to land this without making bool operator ()
> static (as that breaks the build). No idea if this will fix GCC4.4 or not, it
> seems rather broken that this is being reported as a bug.
> 
> Will land as soon as the tree opens, as
> http://codereview.chromium.org/436003/show
> 
> On 2009/11/23 17:20:36, ian fette wrote:
> > Evan -
> > 
> > Making operator () static breaks the build.
> > 
> > c:\b\slave\win\build\src\chrome\browser\cookies_tree_model.cc(62) : error
> C2801:
> > '`anonymous-namespace'::OriginNodeComparator::operator ()' must be a
> non-static
> > member
> > 
> > On 2009/11/23 10:03:46, hbono wrote:
> > > Sorry for my slow response.
> > > Is it possible to land this change on my behalf? (I cannot land this
change
> > > until tomorrow.)
> > > 
> > > Regards,
> > > 
> > > Hironori Bono
> > > 
> > > On 2009/11/20 18:52:31, Evan Martin wrote:
> > > > LGTM
> > > > 
> > > > http://codereview.chromium.org/414025/diff/1/2
> > > > File chrome/browser/cookies_tree_model.cc (right):
> > > > 
> > > > http://codereview.chromium.org/414025/diff/1/2#newcode61
> > > > Line 61: bool operator() (const CookieTreeNode* lhs,
> > > > May as well make this one static too?

[Hironori Bono, 2009-11-24 08:25:02]

Thank you for your comments and sorry for my slow response.

On 2009/11/20 15:08:06, ian fette wrote:
> Anyhow, I'm fine with this change, but would be curious to understand why this
is an error.

Even though I have not deeply investigate this issue, it seemed gcc 4.4 got
confused while compiling your CanonicalizeHost() and mis-understood the function
depended on some state in the class. (As you know, adding static explicitly
declares this function does not depend on any state of the class and prevent gcc
from getting confused.)

I would like to note that I'm not an expert of gcc and this comment may be
wrong.

Regards,

Hironori Bono