[PATCH 3 of 3 V2 fixes-subrepo-test] archive: use {changessincelatesttag} to build the metafile

[Matt Mackall]

On Wed, 2015-07-01 at 00:30 -0400, Matt Harbison wrote:
> # HG changeset patch
> # User Matt Harbison <matt_harbison at yahoo.com>
> # Date 1435723009 14400
> #      Tue Jun 30 23:56:49 2015 -0400
> # Node ID 644b9df80fb127d5c1238444c3c9f79247d74c22
> # Parent  a4e09e0f63bcabe7f9c584dbffbcdd167f57a573
> archive: use {changessincelatesttag} to build the metafile

I really hate accepting a bugfix contingent on adding a feature. What if
I don't want the feature? Then I have to send $SUBMITTER back to drawing
board, which I really hate doing (and $SUBMITTER probably doesn't enjoy
either). Or even if the feature just needs bikeshedding, it needs to go
back and forth while the bug sits there. It's just a bad development
pattern that I want to discourage on principle.

I'm just going to apply the idea of the fix from patch 2 to archive.py
for now since I'm really anxious to get tests passing again. I've also
queued patch 1, thanks.

-- 
Mathematics is the supreme nostalgia of our time.