EDIT: may the downvotes come from the idealists that hammer down pragmatists, like our opinion doesn't even count
- You are hijacking a top comment to insert your off-topic opinion.
- You appear to be more territorial than presenting some sort of argument.
- You are overly negative and somewhat dismissive/disrespectful.
- You are commenting on downvotes, which is generally frowned upon
If by "you appear" you wanted to say "it appears to me" OK. It didnt appear to me.
OP is a little big negative, so that's great to temper the "a little too positive" comments.So it's a win in my book.
Yeah I agree with your last point, but I also consider the ones who downvote because that equally as petty, adding more noise to the site.
Rust isn't going to replace Java, but it may supplant Java in a lot of fields where one wants both memory safety and an OO language with high performance. No one is (rationally) suggesting that Rust will replace every language.
The tone of your post, however, makes it sound as though you have something against Rust, and that you're more interested in pointing out all the things it's not good at, or shouldn't be used for, rather than contributing to a constructive discussion.
> I also pointed out that Rust has to win the mindshare of C++ programmers first, before even reaching other ecosystems
Not entirely true; Rust is an alternative to C++, but it's vastly more accessible to people who've used other languages like Java, Python, Ruby, and so on, where you get types, OO, and memory safety. Instead of simply resulting in a 1:1 replacement of C++ with Rust, it may result in a minimal reduction in C++ programming and a notable increase in Rust programming which would have been done in another safe OO language instead, but at dramatically worse performance.
C++ is just a tool, as is Rust. Rust does not need to win the mindshare of C++ programmers first to become successful. It only needs to itself as a viable alternative in whatever fields people find it useful. I believe it has.
Between WASM, embedded, and even Web development, I see a lot of "mindshare" being attracted to Rust.
They're replying to a line about a Java comeback by saying there's a lot of big data libraries/systems used by Java developers that it'll be difficult to compete with.
I don't know the big data space well enough to judge.
You mean one is already shipping Rust and another seems to be on track to shipping it soon?
> For now, Chrome investment in Rust will remain a background investigation (mostly directed towards prototyping these tools and techniques). If we become convinced this sort of interoperability is possible, we’ll revisit widespread use of Rust in Chrome
[0]: https://www.chromium.org/Home/chromium-security/memory-safet...
It would be unfair to blame rust. But that example doesn't provide confidence that switching from c would be a great idea.
Java is somewhat of a special case though, especially the Big Data ecosystem. Much of the fastest and most important Java tools are written in pure Java (or other JVM language) and as such can't be called from Rust in a reasonable way.
Writing Big Data code in Rust is certainly possible from a performance perspective but is an uphill battle of lack of libraries, integration, etc. Some of this might change in time but I don't see Rust ever displacing Java from Big Data. One of the big reasons for this is the main downside of Java is it's GC, which is mainly problematic for latency sensitive applications. In the Big Data field this is substantially reduced as a problem as the code is usually throughput bound and executed outside of interactive contexts, hiding the cost of the GC where it might be more apparent in other applications.
I love Rust and I think it has a lot of potential to replace some of the databases and networked services used in data infrastructure, just don't see it displacing Java as the top dog.
Rust isn't aiming to displace Java. They're two different levels of abstraction. It's the same with Python. Rust isn't aiming to displace Python just as Java isn't aiming to displace Python either.
Rust one day could displace C++, but the odds are low, because in response to Rust C++ has stepped up its game and has changed as a language drastically. Most modern C++ features are concepts from Rust.
To be fair, that doesn't put Rust in a worse position than any other language binding a C++ API. It's just that Rustaceans tend to have higher expectations in terms of safety.
That said, I don't think anyone was suggesting Rust would displace Java. The OP was pretty clear IMO that his firm had experienced exponential growth in interest in Rust skills while he had been hoping that interest in Java would grow instead. This doesn't mean that Rust will replace Java for data engineering or any other domain in particular.
Give it 20 years and I can see it doing to Python what Python did to R.
Yes, Java applications and Rust mostly don't overlap, but there are exceptions in both ecosystems.
Rust might offer some advantages for Data Engineers (the same way as CUDA applications in C++), not all of them as with every language else, and yes you can write Data centric libraries in Rust for Python (easier than C/C++ IIRC).
And no, there's no need to rewrite existing services to Rust, it can interface with all the examples you cited.
To me Rust is just a simpler and safer C++, if you said you would program C++, you probably will want to program Rust, writing C++ bindings for Rust is basic stuff nowadays
I was mainly responding to this statement which came across quite aggressive. Java and C++ do not need comebacks and my comment was intended to be a wake-up call to reality. Yes, it is great to learn Rust on the side, but advocating something to be used everywhere is another story.
Where is that in the comment you replied to?
Both are general purpose languages, so of course they overlap.
> Good luck rewritting all the big data tools from Java (Elasticsearch, Spark, Kafka, Hadoop, Neo4j, Deeplearning4j, Cassandra, Solr, Arrow, OrientDB)
Cassandra (https://scylladb.com) and Kafka (https://vectorized.io) have already been rewritten once in C++, with massive latency and throughput improvements. No reason why they couldn't get their superior Rust clones in the observable future.
Materialize (https://materialize.com), Noria (https://github.com/mit-pdos/noria), and Sled (https://github.com/spacejam/sled) are just some of the Rust database projects that are aiming at unseating the de facto standard implementations in the space. InfluxDB (https://www.influxdata.com) is now doing major Rust development as well.
The future is almost here. It's just not evenly distributed yet.
It is not a single dimension problem as they try to frame it. Performance is a single dimension of systems. Java with GC is a memory safe solution compare to C++ (hello segmentation faults). The reason why the C++ implementation is faster is not that it is written in C++ but it is written using a different approach. There are plenty of Java HFT systems that run circles around C++ systems yet we cannot draw a conclusion that Java is better suited than C++ for HFT. Implementation details matter, code quality matters and performance does not come purely because you pick a particular programming language. Btw. most of the systems which I was working on using Cassandra instead of ScyllaDB because the companies were not ok switch up for some performance gains (because Cassandra was performant enough for their business use case). I think what you receive as superiority is not as clear cut as you try to make it.
From a current perspective I may agree, but in the ‘90 a number of programmers like me was really excited to be able to develop at a much higher abstraction level than C using classes and, later, templates, preserving efficiency and performances.
More or less because there is some overlap: for example, people write low-latency trading systems in both. But it's very hard to get Java to perform well at the kind of work where Rust shines naturally.
There is also very little overlap between Rust and Go. There's far more between Java and Go than Rust and either!
Yep, this is true. Java has an upper hand in data engineering. There are better languages for targeting the Java libraries and JVM than Java itself. I always use Clojure instead of Java when doing data crunching because it is much easier to create parallel programs with it. Kotlin is also coming up in this space.
Having said that. I like Rust for some reasons:
- targeting multiple libc (musl, no libc at all)
- targeting multiple platforms (ppc, mips, etc., even though some of those are tierN)
- having ML features
Why I do not like Rust:
- accidental complexity and convoluted syntax (maybe it is only me, but it is very hard to follow what is happening)
For me the ideal language would be something like F# with the good parts of Rust.
And as for browsers, you know Mozilla created Rust specifically for Firefox?
I agree that Rust is still early for most adopters.
There are a subset of people who need to solve problems that would normally be solved in C or C++, but they don't know or don't want to use those languages. Those people generally love Rust (I am in this camp).
Then there is everyone else, who either like Rust because it's a cool language, or don't for any number of reasons all of which are fine!
We are heading towards distributed high available SQL and NoSQL databases (with Redis, Postgres and MySQL compatibility for easier migration) designed to work exceptionally well in cloud environments like CockroachDB (Go), TiKV (Rust), Vitess (Go) to name a few.
That is because SRE teams want to be able to scale horizontally and make a node, rack or even region failure to be a normal thing, their company don't need to worry about.
> I always hoped Java would make a comeback, but Rust is looking like the real deal.
To me it seems like SoSoRoCoCo is hoping their preferred language would make a comeback in popularity while also acknowledging that Rust is gaining popularity and real-world demand.
I assume that this is meant to imply that Rust doesn't provide advantages in the data engineering space. So I'll add in here -- as a data engineer myself -- that Rust is definitely providing an advantage for me. Combining it with Python (using PyO3 bindings) lets me write performant, safe code that interops with Python in a project for which Scala+Spark isn't feasible. So while Rust may not provide an advantage for you, that doesn't mean it doesn't have a potential niche for data engineering.
As for the others, if it can be shown that porting them comes with a decrease in latency and infrastructure costs then they'll be ported. I've often wondered why they haven't been already.
