"abort: 00changelog.i at d59e4f1a8821: no node!" with TortoiseHg 3.0

[Matt Mackall]

On Mon, 2014-05-12 at 18:28 +0200, Angel Ezquerra wrote:
> On Mon, May 12, 2014 at 6:01 PM, Matt Mackall <mpm at selenic.com> wrote:
> > On Mon, 2014-05-12 at 17:48 +0200, Angel Ezquerra wrote:
> >>       File "mercurial\repoview.pyo", line 44, in <genexpr>
> >
> >         if tags:
> >             blockers.extend(cl.rev(t[0]) for t in tags.values())
> >
> > You have a tag that points to a non-existent node.
> 
> I see. Is this a regression?

Probably. Let's get a bug filed.

I just ran the whole test suite with the following patch:

diff -r bcddddcf0b54 mercurial/tags.py
--- a/mercurial/tags.py	Fri May 09 13:32:56 2014 -0500
+++ b/mercurial/tags.py	Mon May 12 12:58:42 2014 -0500
@@ -82,6 +82,7 @@
     binary, not hex.'''
 
     filetags = {}               # map tag name to (node, hist)
+    filetags['00000000000000000000'] = ('00000000000000000000', [])
     count = 0
 
     def warn(msg):

..to shake out any other code that might depend on non-bogus tags and
this was the only thing that triggered.

-- 
Mathematics is the supreme nostalgia of our time.