Q for Mike Clark: tagging the release

Dear Mike,

Together with the folder structure mentioned in
a previous post, recently I've had to brush up
the naming scheme for our cvs tags. It used to
be messy

, but now it's in control...
hopefully. There are basically three kinds of
tags and branchnames in cvs that now seem to
satisfy the needs of our project.

I've seen this topic mentioned in the contents
list of your book, again my question is whether
you have suggestions with respect to this point.

Cheers,

Gian Franco Casula
[ September 21, 2004: Message edited by: Bear Bibeault ]

I try to keep things simple.

When all the code for a release has been checked in to CVS, for example, I create a branch release:

This creates a branch called RB_1_0. This effectively gives me a copy of the files that were on the mainline. Thus, development can continue on the mainline without affecting the release.

Then, once I've tested the release branch and created a distribution file from its contents, I tag the release:

At any point after the release I can go back and re-generate the release contents by checking out the files with the 'REL_1_0' tag. I describe this in more detail in the "Push-Button Releases" chapter of the book.

So my convention -- the convention used in both the automation and the version control book -- is to use the major and minor version in the tag name. That approach has served me well so far. I've seen other more complicated approaches, but it always makes wonder what value they add over a simpler approach.

Hope that helps.

Mike
[ September 21, 2004: Message edited by: Mike Clark ]

Hi,

My naming scheme is similar, I have included
a task code when a branch is created, so
everyone can see for what the branch has
been made.

The taskcode comes from a general administration
sheet in Excel, where all tasks or helpdesk
requests can be inserted manually or with a
simple dialog.

Cheers,

Gian Franco Casula