Bugs. Process of Testing
Fiddling with bugzilla just now.. and was contemplating on creating a button for our QA team to click on (instead of going to Bugzilla's enter bug directly)
This button will trigger a process to
- Place a unique stamp on log
- Generates a URL (to view the log at the stamp, and -500 lines before it)
- Redirects to bugzilla's enter new bug page, with the URL pre-filled into the bug description field.
But...
Not all bugs would potentially have useful hints in the logs ya?.. what if the log rotates dammit.. my viewer has to cater to that - dammit.. my viewer/stamper would be deployment specific.. and also, while I'm at saving the developer's time (avoid having to repeat the steps described)... I might as well save the QA's time in describing his steps in the first place?
I mean, the QA is using a computer ya? A computer knows and can record everything that goes through it, ya? Why not have a
- Middle-layer sniffer to capture all to-fro traffic.. and a viewer counterpart that is able to (guess-timate) reproduce each originating (form fields prefilled)/resultant HTML page.. (how about javascripts on forms?)
- QA-specific-browser that records the buttons clicked, etc.. (how about browser compatibility?)
- QA-tool that wraps any browser to collect data.. (big engineering effort? how about playback?)
- Plugins to browsers to collect such data and counterpart plugins to playback such data?
- Click 'Start Test Case'
- Go about doing his thing,.. and some error occurs
- Click 'Report Bug'
- Go back to Step #1 to proceed with other things
But I would believe I can use something like that.