Just finished migrating the last few tickets, now that Toby had added the missing categories in Trac; our catchall for ideas on how to integrate with other social networks to offer extended value.
Now we are all on Trac, so to say. What a relief.
It really works well. With custom queries you can easily build your own to-do list sorted by milestone and priority.
The screen of my main computer is still black and limp. I guess the graphics card has given up. Will have to try to replace it with another one tomorrow.
And Google are busy verifying to open the mail in Apps. Then I can hopefully stop getting the back-ups in my usual inbox.
August 29, 2007 at 7:49 am
W00t! You’re using Gmail as a file system for you backups?
August 29, 2007 at 7:59 am
Yes, we’re actually using Gmail as a file system at the moment, but only as a second fallback in the backup process. The first is a regular file server storage. From there the files are PGP encrypted and sent over the network. It is more like “if one continent sinks, we have a second backup on another continent”.
PGP on the file level together with using SSH for connecting to the online Subversion repository and SSL for Trac feel Pretty Safe. We will however drop the “GFS” part of the process when the files grow in size and we start mirroring the backups to another server.
August 29, 2007 at 8:09 am
It’s great to have you by my side, Toby.
August 30, 2007 at 7:46 am
One thing that puzzles me with Trac is the lifecycle of a defect. I’m used to have more stages to run a defect through, e.g. to set it to “To test” after deploying at a dev server and then the tester or reporter must validate it and either return it or confirming it as Fixed and close it. In Trac the developer just says Fixed or worksforme after taking care of a bug and then the testers must find it and test it and re-open it if it still has problems.
Maybe this is part of the agile movement that many Trac using project teams seems to be part of?
August 30, 2007 at 10:02 pm
Toby and I have actually had some rather animated discussions about this “worksforme”, but have found a fairly practical approach to the issue Gary describes.
The workflow we have agreed on is that the developer sets the status as “worksforme” and closes the ticket BUT ensures that the reporter is notified, for example by being cc’d.
This is the trigger for the reporter that the fix must be validated. If we agree with the fix, we make a note in the ticket and just leave it. If we are not satisfied, we re-open, comment and reassign the ticket to development.
What we have discussed was who should close the ticket: the developer who thinks it’s fixed or the reporter who can confirm it through testing.
Based on Toby’s referral to “established Trac practice”, I decided to chose another battle. So, the developer who changes status to “worksforme” also closes the ticket.