Ask HN: Are you a fast developer?

leesalminen · on March 5, 2015

No, of course you didn't waste a week. Every minute spent honing your skills will be beneficial to you later on.

I've been told I'm a "fast" developer. I'm still unclear about what it means. To me, it seems as though they really mean "You say what you mean and you do what you say" or "You meet deadlines, don't bullshit and don't overbill".

Don't worry about what other people say. Focus on you - the rest will come.

bbcbasic · on March 5, 2015

You can't objectively measure your speed as a developer. Attempts to try and measure developers' speeds will cause perverse behaviors - like rushing code to get it checked in quickly and making things less generic, splitting up bugs (so more bugs are fixed per month) etc. And this would cause those developers to actually slow down in terms of business value delivered.

Don't worry about speed, focus on effectiveness.

The trial and error thing - next time see if people can help you either in your team or getting answers online etc. That may reduce the time spent in trial and error but sometimes it is unavoidable.

pjbrunet · on March 5, 2015

What makes me fast is I press Control-S and my code is live--working--in production. Clients love it.

You have an idea--I make that idea work while we're on the phone. Slow devs have six layers of red tape between their code editor and the production server (version control, dev servers, prebuilds...) and they call that agile.

Fast devs keep it simple.

Fast devs are not afraid of their code, they know what to do. Code works the first time, most of the time. That's a good feeling. With experience, you're confident, most changes are easy and predictable.

untog · on March 5, 2015

What makes me fast is I press Control-S and my code is live--working--in production. Clients love it.

One day they won't. I guarantee it.

smoyer · on March 5, 2015

I used to routinely work on code that could incur several millions of dollars per hour liability if the code was wrong (or even inoperable). I'm all for continuous delivery BUT you only deliver if your regression tests pass.

If you deliver, your monitoring systems will roll back if problems are detected. I'd strongly recommend you put at least one layer of "controls" between you and production ... agile (to me) is the lack of BDUF ("architect astronaut") but does not advocate "cowboy coding".

pjbrunet · on March 5, 2015

I agree we need extra careful people doing those important jobs. Most websites, they don't even want to pay for a failover server. So every market has different expectations.

sharmi · on March 5, 2015

I doubt you have worked for long and experienced atleast one episode of the cleanup following a substantial damage. If you are really as fast as you claim you are, I suspect you have never worked with anyone before :)

pjbrunet · on March 5, 2015

Been writing code for 30+ years. I did devops for OS/2. I have seen stuff break, like when Media Temple had a cluster fail and their backups failed. Which is why you always do nightly offsite backups. I never trust a host's onsite backup because they often fail. However, digging into backups for code never happens because with experience you learn how to introduce changes. Also remember, there's a 3rd and 4th backup copy: control-z plus the copy on the production server. 5th backup copy if you count your code editor's backup copy. 6th backup copy if you count the hosts onsite backup copy. And because I use Linode, that means 4 archived backups. So yeah, that's about 9 backups, just in case I forget a semicolon ;-)

sharmi · on March 6, 2015

I dont even have a quarter of your experience. I do realize we have multiple backups either intentionally or unintentionally. But one thing I found when I don't check in code, and unfortunately have to resort to one of the backups, is that a lots of time is spent finding out which is the latest and has all the changes. Under really unfortunate circumstances, I have ended up with different copies of the code having portions of the latest features and having to merge. I have also had the bad fortune of cursing myself for deleting a code file unintentionally and having to rewrite it from some other unfinished copy. Makes me love version control systems. I always use one even for my personal projects. That is my recipe for coding without fear. YMMV.

mod · on March 5, 2015

Damage can go far beyond "I can't find the old code."

The consequence of failures vary drastically with the type of business you're developing for--if you're just developing wordpress sites for mom & pop flower shops, your advice sounds pretty solid.

mathgeek · on March 5, 2015

This was also my first thought. Statement like "fast devs are not afraid of their code" are a huge red flag that someone has little to no experience working on a team.

leesalminen · on March 5, 2015

I don't use the Grunts, LESS, and asset pipelines like most "cool" devs, but Git is 100% worth the effort.

From side projects, to enterprise, to my own start up, Git has saved my butt more than once.

mod · on March 5, 2015

Outside of the "saving my butt," like saving dozens/hundreds of hours of re-writing (probably by enforcing a backup copy of deployed code), I find git helps me gain time in numerous other ways:

1. git blame 2. git log (to re-assert where I left off, among other things) 3. git diff before a commit helping me to remember to remove test/debug code 4. github/gitlab etc search features occasionally

I wouldn't dream of going back to a non-git workflow.

PS: less/sass are cool. not necessarily time-savers but they make nice, clean css source.

iseedeadpeople · on March 6, 2015

Absolutely. I can't even imagine not using git, let alone version control. The time saved and improvement in process pays much more than you put in by learning it.

CyberFonic · on March 5, 2015

Haste makes waste.

You can only make a comparison of speed if you compare developers on the basis of equal robustness, efficiency, reusability, etc.

Once you get on top of the learning curve with any given set of tools, libraries, etc - you would expect to become progressively quicker. But as you have experienced, the learning part can take a lot of time. Assuming that your eventual solution is better (on whatever criteria are important to the given project) than the quicker alternatives then the learning was worth it.

Lorenzo45 · on March 6, 2015

The way I see it is it's all about experience. Now that you've experienced that week of trial and error you can complete a similar problem in 5 minutes, whereas someone inexperienced who doesn't have the knowledge you do will certainly take much longer. The people who are fast coders have just done more trial and error in the past and know how to deal with problems effectively.

goodbye_world · on March 5, 2015

From my experience, I develop fastest when I have the following, in order of importance: 1. Adequate and frozen specifications. 2. Tools. 3. Libraries. 4. Testing tools.

Even if you are working on your own pet project, get off the computer and write some specifications whilst pretending you are the client. For the other three I mean the thing itself and your experience in using those things.

mrits · on March 5, 2015

I doubt it's much different than school. We were all fast before an assignment was due. Very few of us maintained our productive habits everyday.

So the final product has just as much to do with work ethic as ability.