contact-applet, text-ui and presence-applet freeze

[Daniele E. Domenichelli]

On 05/01/13 03:35, David Edmundson wrote:
> On Sat, Jan 5, 2013 at 2:32 AM, Daniel Vrátil <dvratil at redhat.com> wrote:
>> On Friday 04 of January 2013 22:38:49 Daniele E. Domenichelli wrote:
>>> Are you just moving the files or are you preserving their history?
>>> That would be quite important in my opinion, at list for the classes in
>>> ktp-common-internals. git filter-branch + git merge can make miracles ;)
>>
>> I had no idea that it's possible to preserve history across git repositories,
>> so I just moved the files.
>>
>> If you would insist on preserving history, I can try to learn the magic.
>>
> http://www.sharpley.org.uk/node/21

A few comments:

1) Do not use the repository you use for developing for
filter-branching, take a new clone and use that one.

2) A slightly improved version that I used which allows you a better
control on where to move and rename the files, and works if the files
were not in the root directory of the repository. I suggest you
re-create the new directory structure inside the "keep" folder, and then
use the subdirectory-filter on the keep directory.

---

#!/usr/bin/python
import os

#         folder/filename            new folder  new filename
files = [["src/wallet-interface.cpp", "keep", "wallet-interface.cpp"],
         ["src/wallet-interface.h",   "keep", "wallet-interface.h"]]

for file in files:
  if not os.access(file[1], os.R_OK):
    os.makedirs(file[1])
  if os.access(file[0], os.R_OK):
    os.rename(file[0], file[1]+"/"+file[2])

---

3) You should check in the history if the files were ever renamed/moved
and add the old file names to the list (with the same destination as the
new file)

4) Instead of creating the patches, you can add your filter-branched
repository as new remote in the repository where you are moving
everything (you will get a remote with no common commits), and merge it
to master



> Be careful though.. when you recommit (even to the move_plasma branch)
> it will run all the commit hooks again...

I think it is possible to temporarily ask to the sysadmins to
temporarily disable the hooks while we do this...


Daniele