Comment 12 for bug 98836

It's insane that it's taken 5 years since the first "bzr doesn't work on NFS" bug report (in the mailing list - it comes up first in a web search for "No more locks" bzr) and STILL the developers are yammering about whether and how to fix it. Howabout "#define flock(...) noop"; that will fix the problem for 99% of the users. Then you can have an unfixed bug report for the other 1% of users who somehow scramble their files by running two commands at once.

In case you didn't figure it out, even when there's just ONE GUY using bzr, running ONE COMMAND AT A TIME, they still get screwed over by this bug, for years!

And "bzr branch http:...." first downloads megabytes of stuff, THEN fails. If you move that whole directory tree to a non-NFS filesystem (assuming you have one), then it still fails and you have to go back and re-download all the megabytes you already downloaded. I'm amazed anyone still uses bzr, when there are so many better alternatives.