The CMS in C I never wrote

In 2015 you were worth whatever language you used. This is the story of how I ignored that rule, and what happened next.

(This is an English adaptation of my Italian essay Il CMS in C che non ho mai scritto.)

The choice

In 2015 I spent a few days seriously considering writing a CMS from scratch in C.

It wasn't pure madness (okay, it was getting close). I had the skills to pull it off, if I'd wanted to. The real problem was something else: in the dev world of those years, you were worth whatever language you used. In the street-credibility narrative of the time, the steeper a language's learning curve, the more respect it commanded. C was, without question, respectable. PHP or JavaScript were for amateurs. WordPress wasn't even worth mentioning.

Then I said: naaaaaah.

I had to build a platform for selling digital comics. Users, content, payments, files protected from download. Alone, on a tight budget I needed to stretch to cover a place to sleep for me and my non-technical co-founder, plus the costs of landing the first contracts. Do-or-die deadline: six months to reach an investor day, presenting not just the app but user numbers to back it up.

I took WordPress and bent it to what I needed.

Not because it was the cooler choice. But because I knew that codebase by heart. I knew how it behaved under load, where it tended to break, how to extend it. And above all I knew where to look when something went wrong at 2 AM — which happens, always, invariably, at the worst possible moment.

A CMS in C, I'd probably still be writing it today.

What came out of it

In six months, with a dev, UI, UX, and DevOps team consisting of... me, alone:

• The platform's backend
• REST API for the mobile app
• iOS app
• Asymmetric-key encryption system for file protection, registered with SIAE (Italy's copyright collection agency)
• Infrastructure on two load-balanced, synchronized machines, managed with Ansible
• HHVM for JIT-compiling PHP, since PHP 8 obviously didn't exist yet

The back office was "free" — the same old WordPress everyone on the team already knew how to use. Essentially zero onboarding: whoever needed to use it, used it, without ever opening a support ticket.

First week after launch: 9,000 signups. App of the Month on Italy's App Store. €400K raised from investors.

The point

There's a difference between not knowing the alternatives and knowing all of them and choosing the one that works given the real context and the real constraints.

The risk isn't "I have to learn the syntax" or "the elegance of the solution." It's that you don't know the failure modes yet.

It's like choosing between racing with a car you've driven for ten years — you know the third gear grinds a little, how much force to put into the wheel, how it brakes on wet pavement — or with a brand-new vehicle you steer through something the manual calls trim vectoring.

You're already asking: what's trim vectoring?

Exactly.

And that's precisely the problem. With the new stack you're debugging in the dark, at 2 AM, something you don't even have a name for yet.

The CMS in C would have been a great story to tell at meetups.

The 9,000 signups were better.