On Fri, May 3, 2013 at 11:23 AM, C M <cmanalys...@gmail.com> wrote:
> We plan to use a SVN repository as a deployment mechanism so technicians can
> download and install the application binaries for a customer system.
>
> The directory where we want them to download from will always have the
> "current" binaries.
>
> The issue I am facing is how to replace (overwrite) the application binary
> in the directory when there's a new version of it.
>
> I tried to use the "svn import" to overwrite but it doesn't seem to like
> that. I also tried it with the "svn import --force"
>
>
> svn: E150002: Path 'svn://X.XX.XXX.XX/' already exists
>
> Is there another way to accomplish this?
Basically, after your initial import you want to delete or rename the
directory you imported and check it back out as a working copy. Then
to make changes you will overwrite the files in the working copy with
the new versions and 'svn commit' to update the repository. If you
add new files you must 'svn add' them before the commit (if it is
automated, you might 'svn add *' and ignore the error messages about
existing items). Depending on how 'visible' you want to make the
versioning, you might (or might not...) want to arrange the usual
/trunk, /branches, /tags layout for your files, always updating the
trunk copy and after each commit, copy it to a tag with your own
revision numbering scheme. Also, you might use something like viewvc
for web browser downloads and the ability to pick back-rev versions if
needed. Even without tagging, this will give you access to any
version you have committed.
--
Les Mikesell
lesmikes...@gmail.com
