Those of you surprised to see Java so prominent, where have you been all your careers? 10 people startups with nodejs backends? You must have been entirely shielded from enterprise software companies.
Java was the first language I learned in my CS degree, I still think this was a sensible choice by the CS department, but I don't think I've written a single piece of Java since I left 10 years ago!
It seems like a lot of Java usecases may be big and important but kinda isolated! Something about where they sit in the economic value chain perhaps?
There the world is not like at Google. Sap, Java and .net are what people work with.
In terms of programming languages, Google is very much a microcosm of the industry.
(.net is the exception though. Not much of that at Google).
Edit: And maybe some Dart and Kotlin too.
what are the main programming languages used at Google
Result (unformatted):
AI Overview Google utilizes a diverse set of programming languages across its various products and services. The main programming languages used at Google include:
C++: Widely used for performance-critical applications and system-level programming, such as in the core search engine, Google Chrome, and other backend infrastructure. Java: Essential for Android app development and significant portions of Google's backend systems. Python: Employed for a wide range of tasks including scripting, data analysis, machine learning, and web development (e.g., YouTube). JavaScript: Fundamental for web application development and frontend interactions across Google's web-based services. Go (Golang): Google's own open-source language, increasingly used for cloud-based projects, microservices, and network programming due to its efficiency and concurrency features.
SQL:
Crucial for managing and interacting with databases, which are integral to almost all data-driven applications at Google.
So it looks like the main ones I missed were Rust and SQL.
Dang! I should have thought of SQL, at least for their IT ops, but I was thinking only about their customer-facing apps.
Anyway ...
The financial sector, insurance sector, healthcare sector all jumped on Java a couple of decades ago, and they have massive repositories of Java code, and are actively migrating their COBOL code to Java.
What do you mean by this? To me it sounds like people are saying they are both "old" languages, but I don't know what you mean.
I work in a shop that has lots of both Java and COBOL. We are not "actively migrating" COBOL code to Java. It looks like mainframes will continue to exist for decades to come (i.e. >20 more years). Obviously, brand new applications here are not written in COBOL.
That was two decades ago – almost a generation! Interesting to think that some of those systems would now be considered “legacy”.
That would make OPs counter re netflix relevant. I don't understand your point
In its day, a lot of 'cool' companies used COBOL, back then. Because it was an ok solution, back then. So to say, today, Netflix is cool and uses Java, thus Java is different and still cool, is not valid. Does not invalidate the point. It is the same situation, just decades later.
Maybe shouldn't have conflated SAP, but they seem to be just all part of the same giant ecosystem of 'current/entrenched' solution that 'we use because we have to, not because it is better'. Not unlike COBOL.
I'm not saying COBOL is a bad language, far from it, which the billions of lines of code running production proably also attests to. The first COBOL program i ever edited, in 2008, was last edited in 1987. It had run flawlessly every day for 20 years. COBOL when invented was invented to allow business people to express business logic programatically, which is also why it has such a large footprint in finance, insurance, etc.
I'm not saying Java is a bad language either. Java is great, much like COBOL was, and like COBOL, Java still evolves today. It has flaws, but so does every other language, and most of the flaws in Java are understood. There is literally also nothing you can't do in Java that you can do in <insert fashionable language of the year>.
We probably shouldn't write web frontends in Java, and most people figured that out a decade or more ago, including the financial institutions.
The typical flow in a financial institution is something like "Angular (in some form) => Java Backend => COBOL on mainframe => DB2", where "=>" can be anything from REST to message queues (i was tempted to write MQ, as most will likely be IBM MQ, but others exist and are used).
Most companies migrating away from mainframe (and thereby often COBOL), have also started implementing microservices instead of giant monoliths, which is what has kept the mainframes of the world running for so long. Most companies i've worked with, have had 45,000 - 90,000 COBOL programs running every night, with almost as many running on demand, and each and every one depends heavily on the output of the previous part of the chain.
Thost giant chunks are now being migrated to microservices with well defined couplings, meaning that when it eventually becomes time to migrate away from Java, it will be somewhat easier as you can eat the elephant one mouthful at a time, and not have to reimplement 50+ years of legacy code and conventions in one go.
I've said it before, and will gladly say it again, if you choose a COBOL career today, you will most likely never be unemployed for long until you retire.