Do not prompt for passwords

Do not prompt for passwords

Forcing function for developers to create a .netrc file.
BUG=

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=210686

[Isaac (away), 2013-07-09 21:40:33]

I don't understand how this works.

[szager1, 2013-07-09 21:46:23]

On 2013/07/09 21:40:33, Isaac wrote:
> I don't understand how this works.

try it

[Dirk Pranke, 2013-07-09 21:46:36]

On 2013/07/09 21:40:33, Isaac wrote:
> I don't understand how this works.

I also don't understand what problem we're attempting to solve here or why this
change is necessary ? I don't currently have a .netrc on any of my machines, and
nothing particularly seems broken, but maybe this only affects gerrit-based
repos or something?

[Isaac (away), 2013-07-09 21:54:57]

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py
File gclient_scm.py (right):

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py#newcode1033
gclient_scm.py:1033: # Don't prompt for passwords; just fail quickly and
noisily.
only if jobs > 1 ?

[szager1, 2013-07-09 22:23:09]

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py
File gclient_scm.py (right):

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py#newcode1033
gclient_scm.py:1033: # Don't prompt for passwords; just fail quickly and
noisily.
On 2013/07/09 21:54:58, Isaac wrote:
> only if jobs > 1 ?

Hmm... My concern here is that if we surface the password prompt, people will
think that's the "right" way to do it.  I think I prefer showing them the error
message which steers them to the new-password flow.

I'm not entirely convinced, though; anyone else have an opinion?

[Isaac (away), 2013-07-09 22:29:48]

I think gclient should detect this situation (if user has src-internal in path)
and prompt a warning for now...  That would also be more effective than another
psa.

[szager1, 2013-07-09 23:06:46]

On 2013/07/09 22:29:48, Isaac wrote:
> I think gclient should detect this situation (if user has src-internal in
path)
> and prompt a warning for now...  That would also be more effective than
another
> psa.

(Copying offline communication)

With this change, the message produced by gclient is actually very helpful:

1>Cloning into 'XXXXX'...
1>fatal: remote error: 
1>Invalid user name or password.
1>Please generate a new password at:
1>  https://XXXXX.googlesource.com/new-password

The instructions on that page are exactly correct (windows-specific instructions
are currently being incorporated).

[szager1, 2013-07-09 23:45:36]

Can I get an 'amen'?

[szager1, 2013-07-10 00:22:19]

Committed patchset #2 manually as r210686.

[Dirk Pranke, 2013-07-10 00:26:34]

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py
File gclient_scm.py (right):

https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py#newcode1036
gclient_scm.py:1036: env.setdefault('SSH_ASKPASS', 'true')
I think you should have a comment here explaining what this does and why we need
it; it's not at all obvious.

[Isaac (away), 2013-07-10 00:30:47]

It is against chromium policy to TBR CLs that are not related to current
breakages.

It is even worse when the CL is up for review and there were unresolved
comments.

Please revert this until we resolve comments, or I will be.

[szager1, 2013-07-10 00:44:32]

On 2013/07/10 00:26:34, Dirk Pranke wrote:
> https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py
> File gclient_scm.py (right):
> 
> https://codereview.chromium.org/18949002/diff/3001/gclient_scm.py#newcode1036
> gclient_scm.py:1036: env.setdefault('SSH_ASKPASS', 'true')
> I think you should have a comment here explaining what this does and why we
need
> it; it's not at all obvious.

https://codereview.chromium.org/18851004