Hacker News Reader

Comment by logicchains | Hacker News Reader

It really surprises me that somebody can get away with such destructive behaviour as reverting someone else's commits without reasonable cause. What justification is there for not firing on the spot anybody who puts their own personal vendettas over the good of the project?

View on HN

I once quit a senior dev role as the tech architect at the company had veto power on all code and blocked my work because I'd spelled the word dialog the US way, rather than the UK way (dialogue).

This was not in a public interface, this was in variable names.

Ignoring that every word in programming is in US English: color, gray, program.

We were deadlocked for weeks, with him refusing to ship working code that fulfilled the business goals over this single issue. Quitting resolved the deadlock, and I learned afterwards that he really did go through the work and change it all to "dialogue".

What did senior management think of this? I can't imagine them taking kindly to the architect delaying shipping for weeks over such a trivial thing.

At the time he was viewed as critical to the project. They indulged him. A very steady stream of very good devs leaving the company occurred because of such invented rules and irrational behaviour. He was a truly great dev, but a fairly insane architect and gatekeeper.

I wonder if they financial benefit they gained from keeping him around was greater than the financial loss due to continuously losing good devs...

Why was it so critical to spell it "dialog", rather than "dialogue", as the higher-ups in the project apparently preferred?

It wasn't, there was no business justification for it at all. He just invented a rule "all names and variables must be in UK English".

GP asked the converse.

Good spot (I missed it).

I spelled it in consistency with everything else in the source code at that time... US English. It was a British company, and I understood that the UI should be UK English, but did not comprehend/believe that variables should be too. Especially as one can't force CSS to stop calling it color.

I still don't think that's actually the question that was being asked. It's not, "why do you think your spelling was preferred?" but rather "given that the complaint is how inflexible the senior tech was on the issue, why were you just as inflexible?"

Particularly given that it's a UK company, the request may be arguably wrong, but it's at least semi-reasonable. Why make such a huge deal over it from either side?

given that the complaint is how inflexible the senior tech was on the issue, why were you just as inflexible?"

Software engineers are a stubborn bunch. It's kind of necessary given the nature of programming. However, if promoted to an architect, stubbornness is a real problem, with little to no upside.

Architects will usually stick mostly to design and not do a lot of debugging. So that stubbornness is not useful. But it sure can rear its ugly head when it comes to pointless conflicts like this one. And the problem is, all the junior developers are also stubborn.

Welcome to software development. What you want in in a software architect is primarily a diplomat. Not someone who is stubborn.

I disagree. Before I went back to academia, I had both job titles at various times. The whole point of having someone in one of these "architect" roles is that they're supposed to be there to exercise judgment they have demonstrated to you as a company to be reliable.

If there's a disagreement that can't be resolved between the two of them, the architect is the one whose stubbornness I value more. You have managers to be diplomats. You have technical leadership to say, "they pay me more than you because they trust my judgment more than yours. I've heard your argument; I found it not compelling enough to overrule my own experience and judgment, and the decision has been made."

Ideally you have more than one such person to help surface the cases where the architect was wrong, but if you're routinely treating them as though they're no more reliable than junior developers, why are you paying them so much?

As an architect myself, I'm sure that some of the decisions I make look like bikeshedding - and maybe they are, but they're born out of a desire to make our services easier to implement and manage at scale. I do try to avoid arguments over trivia, but it's a fine line to walk. When you have a small system, the kinds of convention, indirection, and abstraction you'd see in large ones don't make a lot of sense, except when it comes time to scale out.

Maybe he figured the work was done and wanted to move on to something else. It should have been a trivial issue, either way, and he wasn't the one making it a non-trivial issue. But the architect's propensity for blowing it out of proportion was a good signal that his working relationship with that architect was never going to be good, if something as simple as that was a problem.

I would have done the same thing. If someone is hanging your job over your head over spelling in source code, then the issue is no longer about spelling in source code. It's about dick wagging.

> Especially as one can't force CSS to stop calling it color.

Actually... It's (jokingly) possible :)

http://spiffingcss.com/

Strange, their 'Download it, Sire!' button isn't clickable, despite it being a link. They also have 4 h1s on the page.

Heh, you've basically avoided fully answering this question twice, possibly because it wasn't properly phrased. Let me give it one more shot.

What was _your_ reasoning for losing a job over a completely semantic argument?

Did the illogicality of it all really irk you so much that you couldn't stand to change it? What do you mean by saying that you couldn't 'comprehend' why variables should be written in UK English? It seems pretty clear that they should be in UK English because that's what you were told to write them in... at your job...

If this happened as described then I would have loved to be a coworker watching this hilariously petty feud unfold.

> What was _your_ reasoning for losing a job over a completely semantic argument?

I wanted a reasonable work environment that was delivering product to the customers, for the business.

The work in question was 20k lines and involved the whole stack. As it bled into web page stuff that actually had calls to browser open "dialog" commands, no find and replace was going to safely work to now make it comply with the clarification on the coding standards (UK English var names). We would spend weeks changing it and re-testing, weeks in which we were not delivering it.

I felt we were no longer working for either the customer or business, when we were willing to hold back an improvement that would immediately create revenue, for a petty argument.

It wasn't the first time I'd seen this architect do that to others, but I never thought it would happen to me so long as the product was good, the code was good, deadlines were met.

I was no longer convinced it was possible to create work that wouldn't fall foul of some rule or other. And the architect had managed to position himself on the org structure outside of a chain of command, so there was no-one to appeal to.

There are too many good jobs, and good companies, that want to ship product to their customers and build a great business to even consider staying somewhere that doesn't.

It was a very easy decision.

Wow, yeah, seems like it was.

I still would like to see someone that ridiculous operate... Sounds like the most incompetent architect/engineer I have ever heard of.

Sorry for the snark fellow human.

> Especially as one can't force CSS to stop calling it color.

You can get pretty close :)

http://spiffingcss.com/

Was it that way before you worked on the code? If so" that's perfectly reasonable.

What's the business justification for using "dialog" when the project leader(s) have requested "dialogue", however?

Let's assume that the debate was legitimate, doesn't that still put us one "replace all" away from fixing the problem?

I don't understand how these guys were "deadlocked" for weeks.

Because the spelling wasn't the root cause it was a symptom of a culture that had gone so badly wrong that people engaged in petty arguments just as a way to score points.

To be fair, a search and replace for "dialog" would also match patterns like, y'know, "dialogue" too, at least in all the naive replace-all algos I've seen in IDEs.

So, the traditional CTRL+F, or :%s/dialog/dialogue/g would have resulted in at least some instances of 'dialoguegue' as a result. Yeah, the obvious remedy there is :%s/guegue/gue/g, but that too could lead to unintended results, etc.

Not that I don't agree with the intent of your statement, search and replace across 20k lines of code, over however many files, is likely a more involved process than it sounds like.

Maybe some thing like s/dialog$[^u]|$$/dialogue\1/g

Or use variable renaming in an IDE that actually parses the code.

To match dialog but not dialogue in vim:

    dialog\>

...in other regex systems:

    dialog\b

"dialog" as part of a variable name with underscores:

    :%s/dialog_/dialogue_/g

"dialog" as part of a variable name with CamelCase:

    :%s/\([dD]\)ialog\([A-Z][A-z]*\)/\1ialogue\2/g

You don't need to capture every case in a single regex substitution, you can use a handful to cover pretty much every case though. This sort of change should not take more than a few minutes max of developer time.

I don't have experience with them, but I imagine an IDE with refactoring support should make these sort of changes trivial.

You can always replace dialoguegue with dialogue again afterwards.

Or use an AST and do a structural replace, rather than use text.

did you know about rule before start writing piece of code in question ? If yes, I really don't see how you can say in good faith that it wasn't your mistake to be fixed.

It's completely OK to have such a rule, your behavior is much more disturbing. It doesn't really matter if you use UK or US english, but you should use one and that must be pressured from above. If project standard is UK you use UK, not what you like.

Quitting because of that is just stupid. In my company CS rules are almost opposite to my preferences (and to common style for that language community of course). I use every opportunity to grumble about it, but it would be much worse if everybody (me included) shaped the code based on personal preferences, not style guide.

I once worked at a company that made code to be owned by the gov't of Canada. I'm not sure if the government imposed the rules, but we did have strict rules to use Canadian word versions. Though this was like 1999 so I'd be shocked if that's still the case.

Edit: not to imply a reason for the noted situation, just an example to lend some credence/reasoning elsewhere

Sounds like two stubborn people. Unless you were taking a "stand" to make a point about a larger issue I don't see that you did anything better to fulfill the business goals.

It's a good thing for hackers/programmers/engineers to be stubborn.... up to a point, but the one thing an architect should not be is stubborn.

And yet architects usually originate from promoted engineers. And that means they are stubborn, and the result is all kinds pointless conflicts with all the other developers.

The fact that the tech architect spent any time at all contemplating this issue is a massive red flag to me.

Technical Architecture should be about making the right set of trade offs given the business goals of the system being developed. That involves spending a lot of time talking to people both inside and outside of dev team about what the goals are and how the decisions that devs are making are impacting everyone else in the company. That stuff is time consuming, but if there's any time left you should really be keeping up with the latest frameworks and tools.

If you're spending your time reviewing commits and enforcing coding standards, you're doing it wrong.

"Ignoring that every word in programming is in US English: color, gray, program."

The parochialism of the wannabe "hacker" (entreprenerd) here used to be cute.

Since these were variable names they didn't have to be real words. So you should have explained that your "dialog" was not the American spelling of "dialogue", but a nonsense non-word that you made up (that happened to be spelled the same way as the American version of "dialogue").

At Sun, where there was a structured management chain, there was a time when one of the folks in my group reverted a big chunk of changes I had committed, and I went to our common manager and demanded satisfaction. The response was 'if we left them in he threatened to quit, there isn't anything we can do', and the only conclusion from that statement was that if one of us quit, the manager would rather it were me than this other person. Needless to say it was really annoying.

And thus "threatening to quit" becomes the way for the crazy dudes to get whatever they want.

They're not crazy, and it only works if they have more political capital than you do. If I was forced to accept substandard work, and had no other choice but either maintain that bad code, or quit, I'd threaten to quit too.

No, they are crazy. Reverting someone else's changes without explaining why is crazy. No one is telling these people they can't explain their actions.

If any of these people has a disagreement with a co-worker over the co-worker's changes, it's their duty as a fellow employee to explain to the co-worker why they disagree, or at least explain to the manager why they disagree with the change, so that the manager can explain it to the co-worker. But just making the change and then threatening to quit... that is most definitely crazy.

Just want to clarify that my response here is not related to the GitHub story. I have no knowledge of the circumstances that aren't in the story.

> No, they are crazy.

That may be.

> Reverting someone else's changes without explaining why is crazy.

Not necessarily. There are times when you are so wrong you aren't even wrong.

> If any of these people has a disagreement with a co-worker over the co-worker's changes, it's their duty as a fellow employee to explain to the co-worker why they disagree, or at least explain to the manager why they disagree with the change, so that the manager can explain it to the co-worker.

Let's imagine a scenario where they had done this, and yet nothing had corrected the problem. At that point, it would be quite reasonable to revert someone's changes without explaining why, and threatening to quit if one wasn't allowed to.

You're right, we don't know enough from the original story to say for sure. It's not productive to argue over incomplete stories.

Well, why was someone willing to quit over changes you made? And why was the manager willing to side with him over you?

Those are pretty big flags that you screwed up.

Well in this particular case it was pretty much all politics. Had it been about the code it would have had a different outcome.

The person making the threat felt they were immune from backlash, but they were threatened by where I was working in the code base because they lacked the expertise to compete on a technical level. What I found even more interesting was that our mutual manager felt that his position was an even more tenuous political position so he wasn't going to do anything he wasn't told to do by someone above him in the chain of command.

If someone working for me threatens to quit, I first ask them if their issue can be resolved rationally, and if it can't I ask them when will their last day be. But that is because even if my boss then comes to me and says "You told this guy who is friends with <important person> to quit? Your fired!" I am totally ok with that. Not everyone is.

To look at the flip side though as to whether I screwed up or not, I'm reasonably self aware enough to know when I do. And prior to this incident going 'nuclear', as it did, I had come at it from several different directions to try to eliminate bias. Every third party consulted felt my reasoning was pretty sound. But we all know that being "right" doesn't mean you get what you want in a politicized environment. Just ask the Ukrainians living in Crimea, life is what it is. We move past it.

> Well in this particular case it was pretty much all politics.

I had assumed as much. Sun seemed to be very political except for a very few areas; I avoided it for that exact reason.

It's always a hard lesson for a junior person to realize that politics exists. I had my introduction to that at a very big company in a very hard way.

> It really surprises me that somebody can get away with such destructive behaviour as reverting someone else's commits without reasonable cause. What justification is there for not firing on the spot anybody who puts their own personal vendettas over the good of the project?

And yet it happens everywhere. The question is what narratives can be told about it more than what actually happened.

I've never been anywhere where this sort of behavior happened. If management can't nip things like that in the bud, they aren't doing their jobs.

I have seen vendettas handled in crazy ways in a few different places. People figure out what they can get away with.

If Evil Coder sends an email to Noble Coder describing the vendetta-motivated reversions, then sure, something might happen. When would it ever be so obvious, however? If this is done quietly, and Noble happens to work in different parts of the codebase every cycle, she might not even realize the reversions have occurred for a few cycles. Even then, many people would assume there was some fair-minded rationale the first few times. After it is clear that something is going on, Noble can make a stink, but she might also just want to cut her losses and make a lateral move.

Hmm. I suppose that's a disadvantage of having a less formal code-review process, compared to one in which the reverter would have to specify a reason for each reversion, making it easier to detect spurious reversions.

You'd think that a site like GitHub would have mandated code review for all commits by now. Especially after that time a few years ago when they managed to accidentally delete all the pull requests in the entire site, or something like that!

Wait, they're a ruby shop, they don't even believe in letting the compiler help you avoid making basic mistakes. dons flame-retardant suit

It should be obvious that that's a false equality.

Reply on news.ycombinator.com