5 ms·
Sounds interesting, but I don't think PHP & MySQL are a good stack. While it's possible to write secure software in PHP (I'd probably trust the Paragon IE) guy
by zeveb 9y ago
Sounds interesting, but I don't think PHP & MySQL are a good stack. While it's possible to write secure software in PHP (I'd probably trust the Paragon IE) guys to do it), it's not very likely. As for MySQL — as far as I'm concerned there is essentially no reason to choose it over PostgreSQL, and a project which does so is thus instantly suspect. If they've made an error that grave, what other errors have they made?
It'd be like someone choosing Windows as a server OS: there's a small chance it was a good decision, and a far, far greater probability that it was the result of a toxic engineering & management culture.
- nix0n 9y ago> If you can run WordPress, you can run Friendica. The LAMP stack, for all its problems (you're not wrong) is the bog standard and is easy to rent. If you'd rather use Rails, check out Diaspora: the two systems federate together, anyway.
- seanf 9y agoWhy do you consider choosing MySQL over Postgres a grave error? There are some cool features in Postgres. I am going to learn more about partial indexes, that looks useful! But there is plenty of precedent for large, successful MySQL projects. Uber, for example, switched from PostgreSQL to MySQL in 2016. https://eng.uber.com/mysql-migration/ https://eng.uber.com/mysql-migration/
- zeveb 9y ago> Why do you consider choosing MySQL over Postgres a grave error? Because PostgreSQL focuses on correctness & MySQL focuses on performance & ease of use. With a database, I want correctness first, and performance & ease of use second. Many projects chose MySQL because it was easy to use & somewhat faster, and they didn't realise how broken it was initially. I am sure that many projects failed due to MySQL, and are now lost to history. Choosing MySQL over PostgreSQL betokened the kind of culture which chooses convenience over correctness, which is not the kind of culture I wish to associate myself with. Over time, MySQL added more & more correctness, and PostgreSQL got faster & faster. As it turned out, PostgreSQL with the right performance flags is faster than MySQL with the right correctness flags. In the specific case of Uber, it sounds like they're essentially using MySQL as the underlying storage mechanism for a key-value store, and are able to build the mechanisms they need atop it. Also, PostgreSQL less efficient for their usage patterns than MySQL. That makes sense, and seems like a sensible engineering decision. 'Essentially no reason to choose [MySQL] over PostgreSQL' doesn't mean no reason at all. It's kinda like how sometimes it makes sense to write an inner loop in assembly, but it extraordinarily rarely makes sense to write a word processor in assembly in 2018.