TortoiseHg 1.1.4 trouble to rename a directory with only different case
Valborg Kalender
valborg00 at yahoo.com
Mon Nov 1 10:54:48 UTC 2010
TortoiseHg 1.1.4 trouble to rename a directory with only different case
Hi there,
When trying to rename a directory using TortoiseHg on a Windows (I'm
using Windows 7) where only change is the directory name case, the
result is a corrupt archive.
The example below is shown using as much as possible using mercurial
from the CMD command prompt. This just to make it more ilustrative.
1) Create a directory with a look like below:
C:\Temp\Test\Dir1
C:\Temp\Test\Dir1\File1.txt
C:\Temp\Test\Dir1\File2.txt
C:\Temp\Test\Dir2
C:\Temp\Test\Dir2\File3.txt
2) Create the archive:
C:\Temp\Test> hg init
C:\Temp\Test> hg add
C:\Temp\Test> hg commit -m "First version"
3) Try to rename the directory "Dir1" to "dir1".
NOTE: I rename the directory twice, just so that I don't get a
directory "C:\Temp\Test\dir1\Dir1". I guess you can do this
in one run...
C:\Temp\Test> hg rename Dir1 Dir1X
C:\Temp\Test> hg rename Dir1X dir1
4) Show archive status:
C:\Temp\Test> hg status
A dir1\File1.txt
A dir1\File2.txt
R Dir1\File1.txt
R Dir1\File2.txt
5a) Save the changes in the archive using TortoiseHg from the context
menu in Windows Explorer (ie. TORTOISEHG -> COMMIT).
5b) After the commit is made, there is a leftover.
C:\Temp\Test> hg status
A dir1\File2.txt
R Dir1\File2.txt
5c) The archive now seems to be in a "flux" state.
C:\Temp\Test> hg clone C:\Temp\Test C:\Temp\TestClone
updating to branch default
abort: case-folding collision between dir1/File1.txt and Dir1/File1.txt
5d) In this situation you HAVE TO make one extra commit from the command-
line. There is no way to make an extra commit using TortoiseHg (nothing
is commited how many times you try).
C:\Temp\Test> hg commit -m "Make commit of missing files"
6a) If you instead do the commit from the command-line (instead of point (5a)
above) everything workes as expected.
C:\Temp\Test> hg commit -m "Change of directory name"
I don't know if this is a decided design time behaviour in TortoiseHg, or if
its a flaw. In case it is a flaw maybe an adjustment can be incorporated in
a future version.
Sincerely,
Per Danielsson
Sweden
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial/attachments/20101101/c7a20e92/attachment.html>
More information about the Mercurial
mailing list