5 ms·

I've been a software dev for 10 years. When I started I saw the transition from Perl to PHP and a lot of snobbishness from the former towards the latter. Seeing

by leonroy 10y ago

I've been a software dev for 10 years. When I started I saw the transition from Perl to PHP and a lot of snobbishness from the former towards the latter. Seeing the changing of the guard in web languages was pretty instructional and it's something I see again and again. I think basic CompSci courses should really have a course or two on managing software projects and handling the problems of what framework do I use to build my new software app? Because fundamental language or framework decisions have both a technical and a business component and even as a front line programmer it helps to be aware of both. Node.js is a great environment for getting a server side app going fast and it has very good tooling thanks to the rest of the JS community with additions like npm, gulp, bower, express etc. There's obvious benefit in having shareable libraries between client and server side and most importantly (to software companies) hiring coders who can work with it is far, far easier than say finding that rarest of unicorns - an experienced Haskell developer. If (and it's a damn big if) you outgrow Node.js you're doing well. Then (and only then) look at the alternatives like Play Framework, Spring Boot, Vert.x or whatever else floats your boat. Rants can be useful in giving a kick up the asses of the relevant community to go address certain bug bears. This rant though is so damn generic it reminds me of those Perl developers at college pouring cold water over the idea of using PHP because they felt threatened by it.

ido 10y ago
I think basic CompSci courses should really have a course or two on managing software projects and handling the problems of what framework do I use to build my new software app The problem (if it's a problem, depends on who's asking) is that undergrad CS courses mainly train you to be a CS graduate student (which in turn train you to be a CS academic), but most students choose to major in CS because they want to become professional programmers (aka Software Engineers). I've done both bachelor and master level CS studies and job-preparation-wise would probably have gotten as much (or more) from a 1-1.5 years (2-3 semesters) vocational training than I did from 8 years of university. Probably the most apparent perk my studies have gotten me career-wise was being invited to interviews at major tech companies like google and amazon.
- jalanb 10y ago
  > most students choose to major in CS because they want to become professional programmers (aka Software Engineers) Then why not major in Software Engineering? Is that not an option in American universities?
  - erklik 10y ago
    Not sure about America but it's more of lack of understanding in Australia. Also, in the university i am studying in. The university course for a Software Engineer is for 5 years since that is the minimum study time needed to be accredited as an engineer while the CS degree is 3 years. The more messed up thing is that everything taught in CS major is taught the Software Engineering major but Software Engineering has more electives and has a few general engineering units.
  - duncanawoods 10y ago
    One problem is that it does not have the standing of a pure subject like math, engineering or CS. Its more vocational, isn't offered by top-tier universities (in the UK) and therefore has lower status e.g. its like taking Media Studies at an ex-poly vs. reading English at Oxford.
    - ENTP 10y ago
      It was very much seen that way back in the day at Birmingham University. We wrote an x11 window manager with audio and graphics processor apps for our first term in 2nd year. I saw academia start to go down the pan at college with the introduction of "ICT". To me, this was nothing but subsidised Microsoft training.
      - dagw 10y ago
        this was nothing but subsidised Microsoft training. "Subsidized Microsoft training" is what most people actually want and need to get the sort of job they aspire to. People with this "Subsidized Microsoft training" are also what many companies are looking to hire. So why is it such a bad thing to offer people this option?
andybak 10y ago
I don't disagree with most of what you say but I can't shake the feeling that for the most common types of web project - nothing touches the productivity possible with frameworks such as Rails or Django. (I don't know enough about mature PHP MVC frameworks to comment on whether Laravel et al should also be in that list) Is there anything for node that offers that wide range of functionality and a mature ecosystem for content-driven sites?
hocuspocus 10y ago
> If (and it's a damn big if) you outgrow Node.js you're doing well. Then (and only then) look at the alternatives like Play Framework, Spring Boot, Vert.x or whatever else floats your boat. I couldn't disagree more. If you like the concept of type safety, don't waste your time.
jondubois 10y ago
There is no such a thing as 'outgrowing Node.js'. Often, people who think that they're "too good" for Node.js are people who had a single bad experience; they designed their Node.js app poorly (E.g. they didn't think through their architecture enough) and then instead of blaming themselves for their ignorance, they put all blame Node.js. It's always convenient to blame the tool. But I've used a lot of different tools and I can tell you that the tool is almost never the problem. I've found that people who don't like Node.js are often people who were forced to learn it because their company used it. These people had resentment against Node.js from the very beginning - And they kept this resentment with them while they were 'learning' it - And as a result, they never really learned it properly - They never fully understood it. Not everyone has to love Node.js but there is no reason to hate it either. It's a damn good tool.
- zaphar 10y ago
  It's always convenient to blame the tool. But I've used a lot of different tools and I can tell you that the tool is almost never the problem. This is both true and also misses the point. Given a sufficiently smart software engineer Javascript is fine. But like the old adage about compilers the joke is that there is no sufficiently smart engineer. Even the best will occasionally make mistakes. Over time those mistakes will accumulate. When enough of them accumulate you experience serious pain. Any argument that rests on the idea that all you have to do is: "Just hire engineers that can be perfect in perpetuity" is doomed to be a poor one. No language is perfect, however there do exist lanaguages that make it possible to fix problems after the fact with higher levels of confidence and more guarantees. Javascript is not one of those languages. Not to mention that single threaded callbacks have an inherently lower ceiling on concurrency that multithreaded approaches. Some times you have to make the decision on whether to save thousands-millions of dollars on infrastructure so you can stick with your current codebase or to rewrite and take a medium term hit on velocity instead. There most definitely is such a thing as outgrowing Node.js.
noddingham 10y ago
You called out the fact that you see repeating patterns in the industry, and go on to mention Node.js as a great environment for rapid prototyping. That made me chuckle because that's exactly what I heard about RoR just a few years ago.

by leonroy 10y ago

ido 10y ago

I think basic CompSci courses should really have a course or two on managing software projects and handling the problems of what framework do I use to build my new software app The problem (if it's a problem, depends on who's asking) is that undergrad CS courses mainly train you to be a CS graduate student (which in turn train you to be a CS academic), but most students choose to major in CS because they want to become professional programmers (aka Software Engineers). I've done both bachelor and master level CS studies and job-preparation-wise would probably have gotten as much (or more) from a 1-1.5 years (2-3 semesters) vocational training than I did from 8 years of university. Probably the most apparent perk my studies have gotten me career-wise was being invited to interviews at major tech companies like google and amazon.

jalanb 10y ago
> most students choose to major in CS because they want to become professional programmers (aka Software Engineers) Then why not major in Software Engineering? Is that not an option in American universities?
- erklik 10y ago
  Not sure about America but it's more of lack of understanding in Australia. Also, in the university i am studying in. The university course for a Software Engineer is for 5 years since that is the minimum study time needed to be accredited as an engineer while the CS degree is 3 years. The more messed up thing is that everything taught in CS major is taught the Software Engineering major but Software Engineering has more electives and has a few general engineering units.
- duncanawoods 10y ago
  One problem is that it does not have the standing of a pure subject like math, engineering or CS. Its more vocational, isn't offered by top-tier universities (in the UK) and therefore has lower status e.g. its like taking Media Studies at an ex-poly vs. reading English at Oxford.
  - ENTP 10y ago
    It was very much seen that way back in the day at Birmingham University. We wrote an x11 window manager with audio and graphics processor apps for our first term in 2nd year. I saw academia start to go down the pan at college with the introduction of "ICT". To me, this was nothing but subsidised Microsoft training.
    - dagw 10y ago
      this was nothing but subsidised Microsoft training. "Subsidized Microsoft training" is what most people actually want and need to get the sort of job they aspire to. People with this "Subsidized Microsoft training" are also what many companies are looking to hire. So why is it such a bad thing to offer people this option?

andybak 10y ago

I don't disagree with most of what you say but I can't shake the feeling that for the most common types of web project - nothing touches the productivity possible with frameworks such as Rails or Django. (I don't know enough about mature PHP MVC frameworks to comment on whether Laravel et al should also be in that list) Is there anything for node that offers that wide range of functionality and a mature ecosystem for content-driven sites?

hocuspocus 10y ago

> If (and it's a damn big if) you outgrow Node.js you're doing well. Then (and only then) look at the alternatives like Play Framework, Spring Boot, Vert.x or whatever else floats your boat. I couldn't disagree more. If you like the concept of type safety, don't waste your time.

jondubois 10y ago

There is no such a thing as 'outgrowing Node.js'. Often, people who think that they're "too good" for Node.js are people who had a single bad experience; they designed their Node.js app poorly (E.g. they didn't think through their architecture enough) and then instead of blaming themselves for their ignorance, they put all blame Node.js. It's always convenient to blame the tool. But I've used a lot of different tools and I can tell you that the tool is almost never the problem. I've found that people who don't like Node.js are often people who were forced to learn it because their company used it. These people had resentment against Node.js from the very beginning - And they kept this resentment with them while they were 'learning' it - And as a result, they never really learned it properly - They never fully understood it. Not everyone has to love Node.js but there is no reason to hate it either. It's a damn good tool.

zaphar 10y ago
It's always convenient to blame the tool. But I've used a lot of different tools and I can tell you that the tool is almost never the problem. This is both true and also misses the point. Given a sufficiently smart software engineer Javascript is fine. But like the old adage about compilers the joke is that there is no sufficiently smart engineer. Even the best will occasionally make mistakes. Over time those mistakes will accumulate. When enough of them accumulate you experience serious pain. Any argument that rests on the idea that all you have to do is: "Just hire engineers that can be perfect in perpetuity" is doomed to be a poor one. No language is perfect, however there do exist lanaguages that make it possible to fix problems after the fact with higher levels of confidence and more guarantees. Javascript is not one of those languages. Not to mention that single threaded callbacks have an inherently lower ceiling on concurrency that multithreaded approaches. Some times you have to make the decision on whether to save thousands-millions of dollars on infrastructure so you can stick with your current codebase or to rewrite and take a medium term hit on velocity instead. There most definitely is such a thing as outgrowing Node.js.

noddingham 10y ago

You called out the fact that you see repeating patterns in the industry, and go on to mention Node.js as a great environment for rapid prototyping. That made me chuckle because that's exactly what I heard about RoR just a few years ago.