Merge lp:~vorlon/britney/britney1-more-stats into lp:~ubuntu-release/britney/britney1-ubuntu

[Steve Langasek]

Add an additional way to measure the proposed-migration backlog

[Iain Lane (laney)]

Do you think it's possible that anything could have been reading this and relying on the number of columns it had?

What about forwarding to Debian?

Other than that, I think this is fine.

review: Needs Information

[Steve Langasek (vorlon)]

> Do you think it's possible that anything could have been
> reading this and relying on the number of columns it had?

It's possible, but we've never made any promises around that, and the last change to the column count was only a year ago.

> What about forwarding to Debian?

Applicability to Debian is not straightforward; our devel-proposed partial suite is in principle something we want to drive down to zero, the same is not true for how Debian manages unstable vs. testing. I think for now we should just carry it as a delta.

=== modified file 'scripts/backlog-report'
--- scripts/backlog-report	2016-03-28 19:31:05 +0000
+++ scripts/backlog-report	2017-07-21 15:04:29 +0000
@@ -26,6 +26,7 @@
     valid_candidates = 0
     not_considered = 0
     ages = []
+    backlog = 0
 
     if path.endswith(".gz"):
         binary = gzip.open(path)
@@ -42,6 +43,7 @@
                             match.group(1), "%Y.%m.%d %H:%M:%S +0000")
                     elif age_match:
                         ages.append(int(age_match.group(1)))
+                        backlog += max(int(age_match.group(1)) - 3, 0)
                     elif line.startswith("<li>Valid candidate"):
                         valid_candidates += 1
                     elif line.startswith("<li>Not considered"):
@@ -57,6 +59,7 @@
         "not considered": not_considered,
         "total": valid_candidates + not_considered,
         "median age": ages[len(ages)/2],
+        "backlog": backlog,
         })