5 ms·

There's a joke that every web developer makes a personal page by first writing a custom bespoke CMS over a couple of months and then only using it for one "Hell

There's a joke that every web developer makes a personal page by first writing a custom bespoke CMS over a couple of months and then only using it for one "Hello, World!" post, but honestly, I've found that's been a great use of time for me. When I do real work projects, I've ended up saying "Didn't I solve this for my blog a year or two ago?" and then I dig up the code and reuse it. It's a great learning experience, and I think every web developer should do it.

larrywright 6y ago
Exactly. If I’m curious about a technology, no matter what it is, I know that if I build something with it or even just set it up myself and play around with it, I will learn a lot more than if I just read the README or a tutorial written by someone else.
munificent 6y ago
> There's a joke that every web developer makes a personal page by first writing a custom bespoke CMS over a couple of months and then only using it for one "Hello, World!" post I might be able to top that. About eleven years ago, I went through this process: 1. I wanted to redo my web site. But I hated using PHP which is all my shared host allowed. 2. Maybe I'll design a little programming language that transpiles to PHP. Then I can program my site using that. 3. I have the parser working and producing an AST. You know, writing a little interpreter that walks that AST is practically trivial. I'll do the PHP backend later, but let's give that a try. 4. Whoa, I have a working language! Neat! But, man, that tree-walking interpreter is slow. Maybe I could translate the tree to bytecode and write a little bytecode VM... 5. Bytecode is so fun! But I'm relying on the .NET VM for most of the heavy lifting like the GC and object representation. I wonder how that stuff works. Maybe I should try writing a little VM in C++... 6. That wasn't as bad as I expected. I learned a whole bunch. Maybe I should write some of this down. So, uh, now I work at Google on a programming language and wrote a textbook on building interpreters [1]. I never did finish that website. (I just ported my blog to Jekyll at some point instead.) [1]: http://craftinginterpreters.com/ http://craftinginterpreters.com/
- RotANobot 6y ago
  You must have really hated PHP.
  - munificent 6y ago
    Literally laughing out loud now. Yes, this is a pretty fair observation. I read once that a famous author got started because they read some trashy novel and was so incensed that it had been published that they decided they could do better.
    - a1369209993 6y ago
      > I read once that a famous author got started because they read some trashy novel and was so incensed that it had been published that they decided they could do better. Jim Butcher with Dresden Files, or someone else? Because that pattern-matches to "highly promising book recommendation" for me.
      - munificent 6y ago
        I want to say it was Richard Matheson who said that, but I could be misremembering.
      - gcapell 6y ago
        AIUI the scenario with Jim Butcher and Dresden Files was that he was _trying_ to write some "genre hack" novel, and succeeded despite himself: https://www.tor.com/2008/07/25/sdccjimbutcherinterview/ https://www.tor.com/2008/07/25/sdccjimbutcherinterview/ "... struggled mightily until deciding (out of frustration) to reverse course and follow Chester’s rigorous writing techniques to the letter, doing all the plot outlining and character background sheets and so on that the English literature folks often scoff at as the workmanlike tools of genre hacks."
- joekim 6y ago
  Awesome story. Thanks for sharing! Checking out your book.
hinkley 6y ago
It used to be said all applications evolve until they can handle email, but maybe that should be CMS, or CMS and email. I am fairly certain that 'bespoke CMS' is consuming enough developer-hours to largely account for the 'developer shortage' we have. If you supply an extra dozen developers, eight of them will end up working on another bespoke CMS. Off the shelf CMS ends up being a mix of feature factories and lock-in traps, though, which probably explains why NIH is so particularly pernicious in this case. We need a SQLite for CMS.
larrywright 6y ago
I'd be lying if I said I hadn't built my own CMS. Twice.

larrywright 6y ago

Exactly. If I’m curious about a technology, no matter what it is, I know that if I build something with it or even just set it up myself and play around with it, I will learn a lot more than if I just read the README or a tutorial written by someone else.

munificent 6y ago

> There's a joke that every web developer makes a personal page by first writing a custom bespoke CMS over a couple of months and then only using it for one "Hello, World!" post I might be able to top that. About eleven years ago, I went through this process: 1. I wanted to redo my web site. But I hated using PHP which is all my shared host allowed. 2. Maybe I'll design a little programming language that transpiles to PHP. Then I can program my site using that. 3. I have the parser working and producing an AST. You know, writing a little interpreter that walks that AST is practically trivial. I'll do the PHP backend later, but let's give that a try. 4. Whoa, I have a working language! Neat! But, man, that tree-walking interpreter is slow. Maybe I could translate the tree to bytecode and write a little bytecode VM... 5. Bytecode is so fun! But I'm relying on the .NET VM for most of the heavy lifting like the GC and object representation. I wonder how that stuff works. Maybe I should try writing a little VM in C++... 6. That wasn't as bad as I expected. I learned a whole bunch. Maybe I should write some of this down. So, uh, now I work at Google on a programming language and wrote a textbook on building interpreters [1]. I never did finish that website. (I just ported my blog to Jekyll at some point instead.) [1]: http://craftinginterpreters.com/ http://craftinginterpreters.com/

RotANobot 6y ago
You must have really hated PHP.
- munificent 6y ago
  Literally laughing out loud now. Yes, this is a pretty fair observation. I read once that a famous author got started because they read some trashy novel and was so incensed that it had been published that they decided they could do better.
  - a1369209993 6y ago
    > I read once that a famous author got started because they read some trashy novel and was so incensed that it had been published that they decided they could do better. Jim Butcher with Dresden Files, or someone else? Because that pattern-matches to "highly promising book recommendation" for me.
    - munificent 6y ago
      I want to say it was Richard Matheson who said that, but I could be misremembering.
    - gcapell 6y ago
      AIUI the scenario with Jim Butcher and Dresden Files was that he was _trying_ to write some "genre hack" novel, and succeeded despite himself: https://www.tor.com/2008/07/25/sdccjimbutcherinterview/ https://www.tor.com/2008/07/25/sdccjimbutcherinterview/ "... struggled mightily until deciding (out of frustration) to reverse course and follow Chester’s rigorous writing techniques to the letter, doing all the plot outlining and character background sheets and so on that the English literature folks often scoff at as the workmanlike tools of genre hacks."
joekim 6y ago
Awesome story. Thanks for sharing! Checking out your book.

hinkley 6y ago

It used to be said all applications evolve until they can handle email, but maybe that should be CMS, or CMS and email. I am fairly certain that 'bespoke CMS' is consuming enough developer-hours to largely account for the 'developer shortage' we have. If you supply an extra dozen developers, eight of them will end up working on another bespoke CMS. Off the shelf CMS ends up being a mix of feature factories and lock-in traps, though, which probably explains why NIH is so particularly pernicious in this case. We need a SQLite for CMS.

larrywright 6y ago

I'd be lying if I said I hadn't built my own CMS. Twice.