/etc/foo. I'd like to make some changes to these, but I'm worried that I might break something. It would be nice to be able to do svn revert [filename], or even svn revert . -R.Before I present one possible scenario for accomplishing this, here are two important Subversion resources:
Here's how I might create and use a local Subversion repository:
- Install Subversion
# zypper install --no-recommends subversion
- Create a repository. In this case, we don't want the repository to be permanent -- it's just a one-time thing -- so we put it in
/var/tmp
# svnadmin create /var/tmp/foo # ls -l /var/tmp/foo # take a look at what we just created
- Import
/etc/foointo the repository.
# cd /etc/foo # svn import . file:///var/tmp/foo -m "Initial import" Adding bubba.html Adding subdir Adding subdir/bubba2 Committed revision 1. #
- At this point, I have the files from
/etc/fooin the Subversion repository,/var/tmp/foo-- what's next? Good question. Since there is no working copy yet, I have to make one usingsvn co. But where? I have two options: either live dangerously by overwriting the existing/etc/foowith the checkout, or play it on the safe side by making a separate working copy somewhere else -- say, in/home/smithfarm/foo. Assuming the program using/etc/foofor its configuration can be told to instead take the configuration from/home/smithfarm/foo, the latter option would enable me to first test my changes in/home/smithfarm/fooand later, when I am satisfied with them, "put them into production" by moving them to/etc/foo. I'm usually a pretty cautious person, so I go with the latter option:
# svn co file:///var/tmp/foo /home/smithfarm/foo A /home/smithfarm/foo/bubba.html A /home/smithfarm/foo/subdir A /home/smithfarm/foo/subdir/bubba2 Checked out revision 1. #
- Now I make some changes to the working copy:
# cd /home/smithfarm/foo # vim bubba.html # vim newfile.conf # svn add newfile.conf A newfile.conf # svn status M bubba.html A newfile.conf # svn diff Index: bubba.html =================================================================== --- bubba.html (revision 1) +++ bubba.html (working copy) @@ -1,4 +1,4 @@ <html> This is a test. -This <% "nothing but a <friggin> test" | h %>. +This <% "nothing but a <smashin> test" | h %>. </html> Index: newfile.conf =================================================================== --- newfile.conf (revision 0) +++ newfile.conf (working copy) @@ -0,0 +1 @@ +BUBBA_CONF="bubba" #
- Now, before I go off committing the changes, I first test them (!)
# test_the_changes
- If I'm not satisfied with my change to
bubba.htmland want to revert it, I do:
# svn revert bubba.html # cat bubba.html
- And commit the changes:
# svn co -m "First round of changes" Sending bubba.html Adding newfile.conf Transmitting file data .. Committed revision 2. #
- Now I tell my program (foo) to take its configuration from
/home/smithfarm/fooand test my changes. If I don't like the result, I can always put the configuration back to/etc/foo(which has not been touched). Or I can revert changes like this:
# cd /home/smithfarm/foo # svn revert bubba.html
- If I already committed the changes, and I would like to revert to the previous commit just for
bubba.html, I do:
# svn merge -r COMMITTED:PREV bubba.html # type it exactly like this
- If I'm not sure which commit I want to return to, I first look at the log:
# svn log
and then choose a revision number to return to. If the revision number to return to is 3, the command would look like this:
# svn merge -r COMMITTED:3 bubba.html
- In any case, when I'm done I will surely want to delete both the working copy and the repository itself. This is a simple case of
rm -rf(as always, use with care):
# rm -rf /home/smithfarm/foo # rm -rf /var/tmp/foo
No comments:
Post a Comment