Writing

Feed Software, technology, sysadmin war stories, and more.

Saturday, May 11, 2013

"No."

The big story this afternoon on Hacker News is a link to a bug tracker at the GNOME project. It concerns a missing feature in something called "gnome terminal". The reporter asked for it to be restored, and the developer had a one word reply:

No.

The bug was then closed as "WONTFIX".

A bunch of people are reacting to this, presumably based on the assumption that no other communication has occurred. I don't know any of the people involved, or even anything about the project, so I can't even begin to speculate as to whether there were other discussions "out of band" or otherwise not shown in this bug.

One thing is clear: if you respond to someone in this manner, you run the risk of generating a rather large and unruly response. I can say this with some certainty based on my own experience. Yep, a couple of years ago, I did this myself. It was in a big corporate environment, not an open source project, and it was a code review, not a bug, but I did use the same response:

No.

I told this tale in one of my older posts back in July 2011. If it seems familiar, you might have encountered it there or in my book. If not, go check it out. It'll make the rest of this make more sense.

There's some more to this story which I didn't tell originally. It has to do with what happened with the actual code review, and it involves some of the finer points of corporate policy about code in the "depot".

My "no." response actually happened on the second code review for this problem. There were actually two of them which I condensed for the purposes of my original story. A few days into this, after he got my original negative response, he went ahead and submitted it anyway. Corporate policy was that you have to have a positive review "score" on your changelist before submission. The tools even enforce this... sometimes.

For whatever reason, he was able to submit it, and I got a mail from the system which said "Changelist #xyz, which you are listed as a reviewer, was submitted. If you did not do this review, please contact blah-blah-compliance ...". I hadn't reviewed it positively, and it had been submitted, so I did like a good little drone and reported it.

I also put together a CL to roll back this change and bounced it off someone else to get the "ps | ... | xargs kill -9" thing out of my tree.

He said it was an accident. Maybe he did in fact submit the wrong changelist number. I didn't even think it was possible to submit something with a negative score unless you used the "to be reviewed" flag to force it. The so-called "tbr" switch was there so you could make an operational change when a reviewer was unavailable but something really needed to be added.

It was appropriate for my former life as a pager monkey, but it made no sense for something like this. Either he used that switch, or he found some kind of problem with the review framework which allowed the submission. Either way, I was incensed. He submitted something which I had explicitly rejected.

Once it had been reverted, we were back to square one. It was this second code review which sat idle for four months, and then when he came back and asked if it was okay now, I said my "no" and left it at that. That, of course, brought on the wrath of both his manager and my own, as explained in the original story.

There's one more thing about this story. When I wrote all of that stuff in the code review and detailed how I had gone to the trouble of trying to train him how to do his own job, I did one other thing at the very bottom of that mega-response, and included something along these lines (bear with me, it's been about four years, so the wording isn't as precise as my "no"):

And finally, I am giving this code review a positive score so it can be submitted. I don't want to argue about this any more. I want it out of my face, even if it is the wrong way to do things.

Just be aware that if this test starts causing problems, I will not hesitate to disable it in the testing system, and I *will* file a bug with you to get it resolved properly.

LGTM.

Yep. I actually gave it a positive score (+1) and walked away from it. It had been nothing but drama and I wanted it to go away.

Oh, and this guy? BS, MS, Ph.D., Comp Sci. If you ever wanted proof that computer science is not software engineering, here it is.