1

The repository is on a linux box and accessed via samba for windows.

In general things appear to work. Git has some trouble with performance on windows, but works otherwise. Because of this, I'm currently only running mercurial.

The problem is that "hg status" sometimes returns different results on windows vs linux. They are both running off of the same repository.

What can cause this difference? How can it be fixed?

Submitting these phantom changes endes up producing this in the change log:

   === (+0,-0) readme.txt ===
    old mode 100644
    new mode 100755

which I guess means it's an ntfs file permissions issue.

v2k
  • 1,083
  • 9
  • 13
  • 4
    You're... sharing a repo over a network filesystem? Instead of cloning/pulling over SSH? – Ignacio Vazquez-Abrams Feb 26 '11 at 03:08
  • Yeah, it's all local, as I ideally just want to work off of the same files via linux or windows as they are running off of a test web server. – v2k Feb 26 '11 at 04:25
  • I guess the preferred method is to do edits on windows and push them to the linux box... – v2k Feb 26 '11 at 04:54
  • Ignacio and Arrow are correct you should use push and pull to move changesets, not a networked file system. – Ry4an Brase Feb 26 '11 at 05:13
  • possible duplicate of [Mercurial: Ignore file permission / mode (chmod) changes](http://stackoverflow.com/questions/5073115/mercurial-ignore-file-permission-mode-chmod-changes) – Karl Bielefeldt Feb 26 '11 at 08:20
  • indeed it is related to that, thanks. – v2k Feb 26 '11 at 08:22
  • I guess this came up because the repository is on a linux box. I want to be able to do all my editing in windows and use TortoiseHg sometimes. Committing and pushing each change on windows before being able to test it seems like unnecessary overhead. – v2k Feb 26 '11 at 08:27

1 Answers1

1

The problem was NTFS file permissions had changed; only visible to the linux version of hg.

v2k
  • 1,083
  • 9
  • 13