Allow the default browser check to return "unknown" and reflect that in the UI.

Allow the default browser check to return "unknown" and reflect that in the UI.

On Linux this can happen for unsupported desktop environments. On Windows and Mac OS X this probably should not occur and likely indicates some sort of serious configuration error. This change avoids repeatedly bothering the user to set the default browser in cases where we're likely to fail at that, without incorrectly displaying that we are the default browser.

BUG=none
TEST=none

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=25553

[Evan Martin, 2009-09-04 22:46:15]

Looks awesome, just got a question about mac.

http://codereview.chromium.org/200025/diff/3001/21
File chrome/browser/shell_integration_linux.cc (right):

http://codereview.chromium.org/200025/diff/3001/21#newcode197
Line 197: return (reply.find("yes") == 0) ? IS_DEFAULT_BROWSER :
NOT_DEFAULT_BROWSER;
I wonder what happens in non-english locales?  (Not relevant to this change,
just noticed it now.)

http://codereview.chromium.org/200025/diff/3001/22
File chrome/browser/shell_integration_mac.mm (right):

http://codereview.chromium.org/200025/diff/3001/22#newcode42
Line 42: NSString* myIdentifier = [mainBundle bundleIdentifier];
Why can this return null now after your change?

http://codereview.chromium.org/200025/diff/3001/23
File chrome/browser/shell_integration_win.cc (right):

http://codereview.chromium.org/200025/diff/3001/23#newcode112
Line 112: return NOT_DEFAULT_BROWSER;  // Or UNKNOWN_DEFAULT_BROWSER?
If this registry key is something we set when we *are* the default browser, then
I think it's safe to delete your comment.

[Mike Mammarella, 2009-09-04 22:56:27]

http://codereview.chromium.org/200025/diff/3001/21
File chrome/browser/shell_integration_linux.cc (right):

http://codereview.chromium.org/200025/diff/3001/21#newcode197
Line 197: return (reply.find("yes") == 0) ? IS_DEFAULT_BROWSER :
NOT_DEFAULT_BROWSER;
On 2009/09/04 22:46:15, Evan Martin wrote:
> I wonder what happens in non-english locales?  (Not relevant to this change,
> just noticed it now.)

I wrote the other end of this in xdg-settings. It does "echo yes" so no worries.
:)

http://codereview.chromium.org/200025/diff/3001/22
File chrome/browser/shell_integration_mac.mm (right):

http://codereview.chromium.org/200025/diff/3001/22#newcode42
Line 42: NSString* myIdentifier = [mainBundle bundleIdentifier];
On 2009/09/04 22:46:15, Evan Martin wrote:
> Why can this return null now after your change?

Presumably it could have before as well; I just moved the check from
IsIdentifierDefaultBrowser() to here because it now results in a different
return value ("unknown" vs. "no").

http://codereview.chromium.org/200025/diff/3001/23
File chrome/browser/shell_integration_win.cc (right):

http://codereview.chromium.org/200025/diff/3001/23#newcode112
Line 112: return NOT_DEFAULT_BROWSER;  // Or UNKNOWN_DEFAULT_BROWSER?
On 2009/09/04 22:46:15, Evan Martin wrote:
> If this registry key is something we set when we *are* the default browser,
then
> I think it's safe to delete your comment.

I looked through the RegKey source but it wasn't clear to me whether either of
these conditions indicates that we would be unable to later set the key. If
either does indicate that, then "unknown" is the correct response for that
condition.

[Evan Martin, 2009-09-04 23:00:35]

LGTM, but you may want to wait for kuchhal to comment on the registry thing (I
hate it when I come across code later and it has comments with questions like
yours, because I never know how to solve them)

[kuchhal, 2009-09-04 23:05:12]

http://codereview.chromium.org/200025/diff/3001/23
File chrome/browser/shell_integration_win.cc (right):

http://codereview.chromium.org/200025/diff/3001/23#newcode112
Line 112: return NOT_DEFAULT_BROWSER;  // Or UNKNOWN_DEFAULT_BROWSER?
These are keys like <root>\Software\Classes\http\shell\open\command which should
always exist on Windows since it comes with IE preloaded. No browser I know
actually deletes this key. So if we can't read the value most likely we won't be
able to set it either so just setting the status to Unknown should be fine.

On 2009/09/04 22:56:27, Mike Mammarella wrote:
> On 2009/09/04 22:46:15, Evan Martin wrote:
> > If this registry key is something we set when we *are* the default browser,
> then
> > I think it's safe to delete your comment.
> 
> I looked through the RegKey source but it wasn't clear to me whether either of
> these conditions indicates that we would be unable to later set the key. If
> either does indicate that, then "unknown" is the correct response for that
> condition.

[pink (ping after 24hrs), 2009-09-08 16:17:40]

mac changes lgtm (even tho this has been closed)