Can't see changes

Matt Mackall mpm at selenic.com
Wed Jul 27 21:33:02 UTC 2011


On Wed, 2011-07-27 at 16:16 -0500, Kevin Bullock wrote:
> On 27 Jul 2011, at 3:56 PM, Jason Harris wrote:
> 
> > 
> > On Jul 28, 2011, at 5:55 AM, Tony Cate wrote:
> > 
> >> I installed the  macosx10.7 version today. I used MacHG to attempt a clone, selecting the "Pull Protocol". I get this error:
> >> 
> >> Mercurial reported error number 255:
> >> transaction abort!
> >> rollback completed
> >> abort: empty or missing revlog for A Cook's Books/.DS_Store
> > 
> > Yes, MacHg is just using the underlying Mercurial, which is issuing these errors.
> > 
> > .DS_Store files are amongst the pre-configured list of files that MacHg has in its local .hgignore file. (but you can of course override this.)
> > 
> > This is a pretty serious error that TimeMachine doesn't back these files up. (I'll put in a stronger warning in future MacHg's if the user wants to store .DS_Store files.)
> > 
> > (In fact, given the time machine corruption, I could see Mercurial in a platform specific way on OSX *never* backing these files up.)
> 
> As mentioned before, the 'dotencode' repo format change fixes this
> issue. Warnings only need to be issued for using repos created with
> older versions of Mercurial (or non-"--pull" clones thereof).
> 
> I wonder if it would merit having hg issue a warning when making such
> a clone of a repo that tracks a .DS_Store file?

What case would that address?

I guess we could warn on local hardlink clones contain files that are
not safe without dotencode (there's similar brokenness that bites people
on Windows with Explorer).

It's worth noting that this is just the most dangerous side-effect of a
common usage error: failure to properly set up a .hgignore file. Which
is either caused by:

a) people blindly running hg add on their first commit without checking
status
b) tutorials not being explicit about the importance of .hgignore files

..which regularly results in people adding backup files, binary files,
tarballs, and other garbage to their permanent history.

It's tempting to say that we should refuse to do 'hg add' with no args
if an ignore file doesn't exist. Or issue a really loud warning. 

-- 
Mathematics is the supreme nostalgia of our time.





More information about the Mercurial mailing list