I didn't realize Facebook only hired software engineers from the '80s who are allergic to coming up with good identifiers

Nah, things just refer to him as fbid 4.

Who are 1, 2 and 3?

Test accounts which ended up getting deleted before (presumably) auth worked.

Holy smokes I just had flashbacks to working at a place that tested things in production and ended up with a couple of these production-db IDs in the code. Shudders

Your codebase only has a couple?

You're already hitting a database, you can add a column for whatever information you want. I've heard people are spooked by having too many columns, but more spooked than having essentially untestable production IDs in their code? No thanks.

I remember one place I worked, someone released code which only returned data permissions for the test user (id 5).

fbid 1 is former zuck. fbid 2 is Facebook itself (the software, not the Facebook's own page in FB). fbid 3 is yet another former test account.

dog it’s a PHP shop, nobody re-uses code

PHP's Packagist disagrees with over 1.4m packages and around a half billion package downloads a month. (More than NPM FYI)

https://packagist.org/statistics

> (More than NPM FYI)

I found your claim hard to believe so I went digging but couldn't find any recent resource on the total number of downloads npmjs.org handles per month.

I could have sworn they used to publish aggregate stats prominently on their home page somewhere ... but can't seem to find it.

Anyway, aggregating* the download numbers of just 12 of the top npm packages comes to about ~504.4m, or half a billion package downloads a month.

*: using September 2018 numbers.

1: https://npm-stat.com/charts.html?package=express - 21m/mo

2: https://npm-stat.com/charts.html?package=body-parser - 24.4m/mo

3: https://npm-stat.com/charts.html?package=lodash - 64m/mo

4: https://npm-stat.com/charts.html?package=async - 64.7m/mo

5: https://npm-stat.com/charts.html?package=debug - 87.6m/mo

6: https://npm-stat.com/charts.html?package=request - 50.8m/mo

7: https://npm-stat.com/charts.html?package=chalk - 64m/mo

8: https://npm-stat.com/charts.html?package=commander - 61.8m/mo

9: https://npm-stat.com/charts.html?package=underscore - 21.5m/mo

10: https://npm-stat.com/charts.html?package=babel-core - 17.8m/mo

11: https://npm-stat.com/charts.html?package=webpack - 16m/mo

12: https://npm-stat.com/charts.html?package=react - 10.8m/mo

it cannot be more than npm.

npm is basically 1 line of code = 1 package

Don't forget the dependencies :).

lol, made me chuckle!

I was referring to the number of packages available as being more. It wouldn't be very useful to compare number of downloads since php does need an entire package to do things easily like check if something is an array.

Facebook recently became the principal sponsor of the Python Software Foundation[1]. Although I don't have any concrete numbers to cite, there's a lot of Python code at Facebook. Calling it a PHP shop would be highly inaccurate and myopic.

[1]: https://www.python.org/psf/sponsorship/sponsors/

Disclaimer: I work there on python runtime optimization, which mostly involves writing C (and C++) code. Also, I've never had to write any PHP/Hack code.