From: David C. <dav...@in...> - 2001-08-29 02:43:51
|
On Tuesday, August 28, 2001 10:58 AM -0400 "Steve D. Perkins" <mai...@st...> wrote: > Excellant! I'm open for any discussion from more veteran team members > about what the best procedure is for running a mirror. As I understand > it, anybody can do an anonymous checkout of the 'htdocs' CVS module > (please post if this turns out to be incorrect or if you need any help). > The only issue might be that what you get from CVS checkout is > occasionally going to be out-of-sync with what's on the production > website... the CVS module is continuously modified during development, > while what you'd want to mirror is manually updated after changes have > been approved. > > Perhaps the best solution is to develop a standard naming scheme, tag > the 'htdocs' module when an update to production occurs, and have mirror > sites periodically refresh their content with anonymous checkouts of that > tag? We would need a standard naming scheme, so that it's simple to > glance and tell which CVS tag represents the latest production code. > Discussion, ideas, and alternatative suggestions are welcome! Maybe a better solution is to create a branch called "release" in your htdocs directory. Then the mirror sites can just "cvs update -r release" and always get the newest version of the web site thats been released. The main trunk of the htdocs can then be modified at will and when its time to release, the cvs maintainer can merge across branches into the release directory. Suddenly every mirror gets a new copy of the site without all the incrememntal changes and without having to be intelligent (scripted). Essentially it makes more work for the cvs maintainer and less work for the mirror site maintainer (probably the way you'd want to balance things). -David |