So, this happened...
@JacobyDave you mean the same implementation language as in cassandra, elasticsearch and hadoop ? ;)
— Michael Hunger (@mesirii) May 8, 2015
On the one hand, I wrote this on a day when I woke up after watching this fantastic conference talk on Neo4J and RNeo4J from Nicole White before bed. This had served to stoke the fires of my Big Data / Data Science jones and had moved graph databases from the corner of NoSQL that I just didn't get to something I had the concepts for but needed to play with to understand.
On the other hand, I wrote this on a day where a tech in my lab was upgrading an instrument machine (a computer that controls a data-generating scientific instrument), and in the process, upgraded Java. This lead to that corner of the lab to enter a time of complete shutdown while nothing worked.
I told someone that day, saying "one of my users updated Java", and without any further explanation, he knew exactly what the problem was, exactly how bad it could be. "They didn't", he said.
This is not a case of me hating Java as a language. I mean, I do, but nobody is asking me to write it, at least not in this context. With Neo4J, I there are modules such as RNeo4J, serving as an interface between the database and the language you're doing your work in. I don't care what language MySQL or MongoDB are written in; I just care that they'll store my data and give it back when I need it.
The one thing that distinguishes Java is "write once, run anywhere"; you write it and compile it so it runs in a VM, abstracting away all connections to underlying systems. Except, there are systems created years ago, where the software was written for a version of the JVM that has been replaced and abandoned a long time ago, where they even deploy computers running XP that were built for Windows Vista or even Windows 7. These exist to some extent because for schools of engineering and computer science, you have to teach the basics of computing using something, and somewhere along the line, the go-to something for introductory courses became Java. So, if you're making complex, computer-controlled scientific instruments, you're needing to design for what students know, rather than spend time and money teaching them a new language as well as the interface to your science. And, because once you sell it, you don't get money for upgrading the interface, so you just don't.
That is my biggest fear. That running a big program to run a big program will tend to choke a computer is an issue, but not as big an issue in this case. It'd run on an older computer, but it's a far newer computer than I started to develop my Java hate on, decades ago. I mostly think that, in context of modern, well-backed projects like Neo4J and Cassandra, I wouldn't end up where I can't upgrade the JVM because it'll break my tools and but every time I use the mouse, a popup shows up saying I'm several major versions behind, and I couldn't download and store the "right" version for my software if I wanted to.
Those are the two Javas: The bright shiny one that promises a good future, and the previous version that keeps you stuck in the past.
On the other hand, I wrote this on a day where a tech in my lab was upgrading an instrument machine (a computer that controls a data-generating scientific instrument), and in the process, upgraded Java. This lead to that corner of the lab to enter a time of complete shutdown while nothing worked.
I told someone that day, saying "one of my users updated Java", and without any further explanation, he knew exactly what the problem was, exactly how bad it could be. "They didn't", he said.
This is not a case of me hating Java as a language. I mean, I do, but nobody is asking me to write it, at least not in this context. With Neo4J, I there are modules such as RNeo4J, serving as an interface between the database and the language you're doing your work in. I don't care what language MySQL or MongoDB are written in; I just care that they'll store my data and give it back when I need it.
The one thing that distinguishes Java is "write once, run anywhere"; you write it and compile it so it runs in a VM, abstracting away all connections to underlying systems. Except, there are systems created years ago, where the software was written for a version of the JVM that has been replaced and abandoned a long time ago, where they even deploy computers running XP that were built for Windows Vista or even Windows 7. These exist to some extent because for schools of engineering and computer science, you have to teach the basics of computing using something, and somewhere along the line, the go-to something for introductory courses became Java. So, if you're making complex, computer-controlled scientific instruments, you're needing to design for what students know, rather than spend time and money teaching them a new language as well as the interface to your science. And, because once you sell it, you don't get money for upgrading the interface, so you just don't.
That is my biggest fear. That running a big program to run a big program will tend to choke a computer is an issue, but not as big an issue in this case. It'd run on an older computer, but it's a far newer computer than I started to develop my Java hate on, decades ago. I mostly think that, in context of modern, well-backed projects like Neo4J and Cassandra, I wouldn't end up where I can't upgrade the JVM because it'll break my tools and but every time I use the mouse, a popup shows up saying I'm several major versions behind, and I couldn't download and store the "right" version for my software if I wanted to.
Those are the two Javas: The bright shiny one that promises a good future, and the previous version that keeps you stuck in the past.
No comments:
Post a Comment