How to prevent accidental deployment of branch under a legacy deployment process?

Question

We have a feature branch right now in development that must not be deployed to production. At the moment there is nothing that would prevent such a mistake from happening.

Deployment happens manually at the moment (SVN export + FTP to production). Migration to a better model is planned but not available in the short term.

Now when you work on a branch locally, then export and upload the files it is really easy to accidentally deploy the wrong branch.

What are some easy and light-weight ways to prevent accidental deployments? We also would be happy with preventing accidental code execution. Downtime is much more acceptable than execution of the feature branches code. So it would be OK for the app to refuse to work in production. This would be noticed right away.

Answer 1

Deployment happens manually at the moment (SVN export + FTP to production). Migration to a better model is planned but not available in the short term.

I do not know what "better model" you have in mind, but you seem to think of something which is so complicated that you cannot implement it by yourself in one or two hours. Why don't you just put the SVN export + FTP steps you currently do manually into a simple shell script? SVN and FTP are available as command line tools for every OS I know. You just have to care for these two things:

• make sure the deploy script exports only from the trunk, but not from the feature branch.

• make sure noone in your team tries to deploy manually, only by that script.

This does not only solve your current problem, it also makes the deployment more smooth and less error prone when you have no feature branch.