[darcs-devel] [issue1325] darcs does not forget an adddir if you delete the directory before recording it
Marco Túlio Gontijo e Silva
bugs at darcs.net
Sun Aug 30 14:25:29 UTC 2009
Marco Túlio Gontijo e Silva <marcot at holoscopio.com> added the comment:
Em Ter, 2009-08-25 às 01:51 +0000, Eric Kow escreveu:
> Eric Kow <kowey at darcs.net> added the comment:
> On Thu, Aug 20, 2009 at 15:54:35 +0000, Marco Túlio Gontijo e Silva wrote:
> > > Also, what happens when you simplify this test even more, for example,
> > > removing the 'a' directory and just adding a file instead?
> > I could not understand what you mean. Can you give a shell script with the
> > idea?
> Here it is: The difference here is that we just create a file 'file'
> without putting it in a subdirectory
> . ../tests/lib # Load some portability helpers.
> rm -rf R # Another script may have left a mess.
> darcs init --repo=R
> echo file > R/file # we need a hunk to make this interesting
> darcs add --repo=R file
> darcs record --repo=R -am ' file'
> mkdir R/b
> darcs add --repo=R b
> darcs mv --repo=R file b
> rm -r R/b
> darcs whatsnew --repo=R | not grep adddir
Ok, I got it now.
> So with the above test, you get this darcs whatsnew output:
> adddir ./b
> move ./file ./b/file
> hunk ./b/file 1
> rmfile ./b/file
> rmdir ./b
> Interestingly, if you change the 'echo file > R/file' to a touch, you
> get something much more expected:
> rmfile ./file
> So I guess what you would find much more reasonable in the above example
> is simply
> hunk ./file 1
> rmfile ./file
> Does it still sound like we're talking about the same test?
Yes, this is a smaller test case for the same thing. I would expect this last
output to be generated by darcs.
> I have a nagging feeling that this may be impossible to fix if so,
> but let's see for now if we're still talking about the same thing.
Yes, I think we are.
Darcs bug tracker <bugs at darcs.net>
More information about the darcs-devel