much quicker.

But remain backwards compat with existing certificates and credentials.

Mail now comes from gitlab@flux.utah.edu so that replies go to Leigh's
new comment ingesting mechanism. Mail now goes to an address defined in
a hash in this file (could also be in the config file) depending on the
label attached to the issue.

Eventually I am going to turn this off for all projects.

pseudo user and prepend the From address on the email to comment

into a gitlab issue comment. The wrinkle is in how we map the email
address into a gitlab user. We have to search the user DB looking for
a match against the username or email. You can guess what is going to
happen, eh?

I tested it with this data file:

	From: Leigh Stoller <lbstoller@gmail.com>
	To: lbstoller@gmail.com
	Subject: gitlab issue: [emulab/emulab-devel] issue #10

	This is another comment.

running it like this:

	gitlab> cat foo.txt | email2issue -c /git/gitmaild.conf

and it add a new comment to:

	#10

port problem on ops.

used to manage local blockstores (datasets). A cheesy iframe hack is used
to embed the APT pages (slick new interface) inside the antique Emulab
interface.

Currently the only hooks from the admin drop down menu, since there is some
additional work before we let users see it.

function has been moved into manage_instance and its associated library
(APT_Instance). Lots of cleanup of the code and more use of webtasks to
communicate with the web server.

scripts. Also add a describe method.

In order to create a dataset, the caller must either be a local user, or
have an extra credential that grants the "blockstores" priv.

Like slivers, datasets are currently created in the holding project and
group corresponding the SA and subauthority.

Eventually need to add sharing control (say, to make a dataset globally
read only), but that does not exist in the backend yet.

the backend cluster.

helper code to get values out of the data array.

dataset has finished allocation. Also take the lock in the parent before
going into the background, since that is the only way we know that
allocation is in progress. Don't like it.

loop again, since it might just be named_setup running.

appropriate strings in the desired format. We have variants of these
scattered around in too many places.

with a valid error code. Makes handling in the caller more consistent.

since it is now needed from the blockstore (dataset) code.

places.