Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Did the author mention C++?

Also, as a user of intelliJ, I would prefer it be written in C++. It's dog-slow.

> Facebook is originally PHP

I doubt it's success was based on its tech, but the fact it was rewritten tells a tale in itself, no?



Facebook in PHP may very well have been a key behind their success. At the time, Friendster was the big social network but was being overwhelmed by their "friend graph" code. Facebook came out to Harvard only, then any college email, then open to the world. I remember being impressed with how few outages the site had, which were MUCH more common then (pre-cloud).

That is mostly kudos to devops but PHP let them build stable code quickly and resolve issues fast. There was a source code leak at one point - a missing tag at the beginning of the page meant the source code rendered as text. That in itself meant their deploy process was garbage but the challenges at the time were mostly around handling scale and traffic spikes.

Anyway, the leaked code wasn't very clean or well-organized and definitely showed some of the growing pains. But the fact is, the site progressed and grew and was mostly stable the entire trajectory, no matter what was happening under the hood or behind closed doors.


> I doubt it's success was based on its tech

Never underestimate how powerful a low barrier to entry is for trying things quickly. In a winner take all market like social media, it’s very possible the choice of PHP over something like lisp gave them a huge advantage.


I actually used Lisp to implement an early (1995) social media platform, which provided a low barrier to entry and allowed me to try things quickly, beating the averages for a few years.

http://shiny.link/QpGMwJ


I wouldn't think of lisp as the next best choice.

Quickly written spaghetti code with massive tech debt could be delivered fast enough to gain a huge competitive advantage - does that mean modern SW dev should venerate that approach?


>> Facebook is originally PHP

> I doubt it's success was based on its tech,

It is, if it let them iterate quickly onto product market fit. Had they chosen something with the dev speed of Rust it is unlikely that they would have iterated fast enough to be successful.


I doubt IntelliJ would become much faster if it were written in C++. The bottleneck on all of my machines running it has always been I/O. If it wasn't I/O, it was waiting for other tools (gradle, maven, node, take your pick) to complete one of their build steps.

It's not an application that's meant to be light weight like VS Code, I believe Jetbrains Fleet is much more suitable for that.

Maybe there are caching advantages to be gained, but those can be implemented in Java as well. I think the most important optimizations wouldn't stand to gain much from a rewrite into native code.


They were at least aware of its reputation considering this is (was?) painted on one of the walls of MPK16.2 https://i.imgur.com/7unV7.png


> I doubt it's success was based on its tech

Success is seldom the result of tech choices, but failure often is.

It's a common misconception that being technically superior is worth more than having a rich, mature ecosystem to draw on.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: