Problems tracking upstream

Kevin Smith yarcs at qualitycode.com
Wed Jun 1 13:35:52 UTC 2005


This morning, I wanted to pull the latest hg sources into my local hg 
repo. First, I was a bit surprised at how slow it was to do:
   hg merge http://selenic.com/hg

It took 40+ seconds, not counting any user interaction time. Considering 
there were only a handful of changes, in a small tree, that seems quite 
slow. Maybe it was all network latency issues.

Well into the process, it bombed out because I didn't have HGMERGE set. 
An hour later, I still don't have a working merge tool. Ubuntu doesn't 
seem to offer tkmerge. I tried meld, but apparently it uses different 
command-line parameters. I set HGMERGE to point directly to the copy of 
tkmerge that comes with Mecurial, but then it complained that it 
couldn't find 'merge' and some other command.

I installed the tkdiff package, and finally got to the point that it 
brought up a diff pane pointing to README. But tkmerge still complained 
that it couldn't find merge, and no differences were shown between my 
copy of README and the one being pulled. I quit from tkmerge, ended up 
in vim for a comment (@#%@#$@#$!!!), and the process claimed to finish 
successfully.

However, things are funky. Re-running the merge says there are no 
changes. I believe at that point I ran hg status and it errored out. So 
I think I did a checkout, which silently appeared to work.

Now, running hg status and hg diff show that my working directory 
matches my repo. But my checked-out README is still my old version, not 
the version that should have been pulled from selenic. I guess I 
"merged" my changes over the ones from selenic.

At this point, is my best bet to simply blow away this mercurial repo 
and re-pull the whole thing from selenic?

Kevin



More information about the Mercurial mailing list