Set magic string in registry to enable fallback to full installer in case of 3 stage updates.

Set magic string in registry to enable fallback to full installer in case of 3 stage updates.

BUG=12832
Committed: http://src.chromium.org/viewvc/18468

[sra, 2009-06-15 21:20:51]

http://codereview.chromium.org/126157/diff/4/5
File chrome/installer/mini_installer/mini_installer.cc (right):

http://codereview.chromium.org/126157/diff/4/5#newcode152
Line 152: KEY_READ | KEY_SET_VALUE, &key) != ERROR_SUCCESS)
Does RegOpenKeyEx allow subkeys to be a path with \ separators?  I though that
you needed to open the full path one subkey at a time.  I hope I'm wrong :-)

http://codereview.chromium.org/126157/diff/4/5#newcode153
Line 153: return;
Who initially creates the key?  Open will fail if the key does not exist (e.g.
on initial install.)  Is this acceptable?

[kuchhal, 2009-06-15 22:05:31]

http://codereview.chromium.org/126157/diff/4/5
File chrome/installer/mini_installer/mini_installer.cc (right):

http://codereview.chromium.org/126157/diff/4/5#newcode152
Line 152: KEY_READ | KEY_SET_VALUE, &key) != ERROR_SUCCESS)
It works fine with the full path. I tested the new change as well but we had
been using it like this in the past too.

On 2009/06/15 21:20:52, sra wrote:
> Does RegOpenKeyEx allow subkeys to be a path with \ separators?  I though that
> you needed to open the full path one subkey at a time.  I hope I'm wrong :-)
>

http://codereview.chromium.org/126157/diff/4/5#newcode153
Line 153: return;
Yes it should fail in that case. If it is first install than diff installer
should not be running and we do not need to change magic string.

On 2009/06/15 21:20:52, sra wrote:
> Who initially creates the key?  Open will fail if the key does not exist (e.g.
> on initial install.)  Is this acceptable?

[cpu_(ooo_6.6-7.5), 2009-06-16 00:00:51]

So if I am in 1.1-dev then it should go to 1.1-dev-full ?


On 2009/06/15 22:05:31, kuchhal wrote:
> http://codereview.chromium.org/126157/diff/4/5
> File chrome/installer/mini_installer/mini_installer.cc (right):
> 
> http://codereview.chromium.org/126157/diff/4/5#newcode152
> Line 152: KEY_READ | KEY_SET_VALUE, &key) != ERROR_SUCCESS)
> It works fine with the full path. I tested the new change as well but we had
> been using it like this in the past too.
> 
> On 2009/06/15 21:20:52, sra wrote:
> > Does RegOpenKeyEx allow subkeys to be a path with \ separators?  I though
that
> > you needed to open the full path one subkey at a time.  I hope I'm wrong :-)
> >
> 
> http://codereview.chromium.org/126157/diff/4/5#newcode153
> Line 153: return;
> Yes it should fail in that case. If it is first install than diff installer
> should not be running and we do not need to change magic string.
> 
> On 2009/06/15 21:20:52, sra wrote:
> > Who initially creates the key?  Open will fail if the key does not exist
(e.g.
> > on initial install.)  Is this acceptable?

[kuchhal, 2009-06-16 00:46:42]

Correct. And if setup.exe does its job, it goes back.

[cpu_(ooo_6.6-7.5), 2009-06-16 00:47:06]

LGTM


On 2009/06/16 00:46:42, kuchhal wrote:
> Correct. And if setup.exe does its job, it goes back.