Is native outdated? Debate!
Life has its ups and downs, but this one is a case that really makes me stop and think. They say the moment you reach 30, you lose your immortality. In the sense that you no longer regard yourself as immortal, infallible and a list of other power-words. It’s the age when you realize that you are no longer in the front-seat of all things new and bright, and that in fact you are heading for death.
With that gloomy intro, here is the pickle that has made me really question what programming and being a “native” developer means: Is native code outdated? I write that because the more young programmers I meet, the less native code I encounter. And it’s really scary! I mean how do you build a custom server? Not just a server that does something different, but a truly custom, never before done, truly unique and brand new – without the speed power that native code provides?
But these kids have a completely different take on it, which makes me feel so very, very old. But it also makes me want to educate them about the differences, with limited success so far.
Once in a while I meet up with younger programmers, some at work and others through IRC or Facebook to check out the latest developments. It’s mostly open forum so everyone posts ideas, links, code, references – and we all bitch and moan about what we don’t like (or envy I guess, people do that as well online).
What throws me is that native code is taking up less and less space in the minds of young programmers. Programmers who, unlike our generation I presume, have primarily grown up with free development tools and open-source libraries. So in their mind paying for anything related to programming is just not where it’s at. Perhaps this is why Microsoft has suddenly transformed into the sugar daddy of high-tech? Time will tell perhaps.
But either way, the majority of these free tools must be running on what ultimately has be, it bloody has to be (!) native code. I accept that the platform itself can be script based, perhaps generating bytecodes or p-code. But native must by virtue of how the hardware is designed, remain relevant. Right?
The p-code thing
If you don’t know the difference between byte-codes and p-codes, they are essentially the same thing. But instead of representing your commands as bytes and small records, like say java or dot net does; P-codes have bytes which represents offset values into a runtime library. So instruction $0001 would execute command #1, and $000F would execute command number 15 (or 16 if you count from zero). The downside of p-code compilers is that they are extremely fragile. If you alter even a single byte the whole system can crash, so modern implementations use CRC and checksum validation before executing.
Since a p-code represents an offset into an array of pointers, they execute extremely fast. In many cases it can be hard to distinguish native programs from well written p-code programs. It depends wildly on the language at hand naturally, but all in all p-codes just pounds every ounce of clock cycles out of the processor.
The downside is that it’s fragile like hell. A single altered byte in a p-code program can crash the program with spectacular access-violations. Hence modern implementations are often cluttered with CRC checks and identifiers. But they are fast, much faster than translating byte-codes.
Right, now back to the topic at hand.
Is the future scripted?
So no. You can’t get through to them.
The smart thing
Since myself and the group of programmers I work with are marketing object pascal as the best language for web and cloud development; and we have adapted object pascal to be more in-tune with how development in our day and age works, like anonymous procedures, records, properties and whatnot — i should be pro 100% script right?
Well I’m not. Smart Mobile Studio represents, in my view, one of the best solutions for writing cloud based software. Primarily because you have full access to nodeJS, which in turn mean you can write both the server and client from the same codebase.
The view I have is that scripting is not really programming is it? I mean “really” programming. Scripting for me is a bit like sculpting. You sit down and sculpt data structures and setup boundaries using a language which, subjectively speaking, is the intellectual equivalent of putty rather than brick. So it’s a bit like carving a madonna out of a piece of soap versus a solid piece of engineering forged in steel.
Or how about this: native programming is a bit like coding a gene-pool, while scripting is the same as body-building. With scripts you carve out the form, the relations and setup the pathways between them. But without the underlying generic programming, there would be no intelligence or program to carry out the build-plans to begin with.
Native languages feel more edgy to me, more solid and concrete. You can’t cheat and get away with it under native object pascal or C, because it wont even compile. You are also closer to the hardware since the code you write is ultimately the binary pattern fed to the processor. While scripting languages are.. well, fed to a dispatcher after going through a lookup table.
A young solution
I really am amazed by how the younger generation solve things, because they are – without knowing it – doing a better job than we did 20 years ago. Instead of writing a server from scratch like we would probably do for a new piece of technology, they implement the “new” bit using the protocols already in place. So they get the job done, but in a way very different from ours.
So the new server is not a new server, it’s the same old HTTP server you have been using for 15 years, except now the URL’s are command and form-fields are parameters. Things like REST was not invented by a seasoned developer, it’s actually a perversion of the HTTP protocol if you like. But it works, and it’s even turned into a standard now.
Same with python. You dont setup a cluster with C or object pascal, no you use python and it will execute parts of the same program on different computers to spread the payload. And that makes sense, since native code would be much harder to disperse over X number of nodes. So suddenly scripting makes sense over native. At least in that particular scenario.
The only real problem with “young thinking” is that it doesn’t generate money the same way we are used to. When you use only free compilers, editors and server kits – how will a young developer approach systems like Delphi?
I really want Delphi to survive and I want the object pascal language, be it Smart Pascal, FPC or Delphi or Oxygene; I want it to thrive. I want young programmers to see and experience how rich and beautiful object pascal is and how much it can do for them; and also how much they can deliver through that language.
But how can you even hope to persuade a young man or woman who makes less than $800 a month to fork out $3000 for a development system all their friends call “old and outdated”?
AppMethod is cool. It’s cheap, its affordable and students and hobbyist programmers can pick it up. It’s still a bit pricey compared to Smart Mobile Studio, but I hope Embarcadero makes enough money to keep going.
Perhaps I should just forget trying to make sense of their thinking. Trying to enumerate all the technological changes I have gone through from childhood to now is exhausting. I tried to tell my son what a modem was and how we connected to the internet no longer than 15 years ago — he just looked at me with big eyes and went “Dad. Please. Its ancient history”. And my daughter found a cassette tape and could not imagine what it was. She thought it was scotch tape 🙂
It’s just so weird realizing that my kids have no idea what a cassette tape is and will never have first hand experience of a modem!
I guess we all play our part.
My part is to try to get object pascal kicking and screaming onto the cloud. I am content with that role and think it’s a privilege to even be mentioned in the history of object pascal. But a future based on scripting? I sure hope not!
What are the kids using?
When it comes to languages, I can honestly say that the official list is wrong. Dead wrong. One language may have more source online that others, but that doesnt mean it’s less used. It depends completely on what group of people use it and what it’s used for. A language used primarily for commercial applications will by nature have less code in the public domain than a language which is free, open-source and generally available.
This is the case we see with Delphi. There are probably hundreds of thousands of companies using Delphi in the world, but they use Delphi to produce closed-source products. As such their code never leaves the house (so to speak) and you wont find it on github, sourceforge or google code.
But having talking to probably hundreds of teenagers coding lately, here is my general impression, in order of magnitude
- C# and dot net
- C script
- Erlang and friends
Of native languages the list havent really changed that much over the years:
- Objective C
- Object Pascal
The second aspect and one I find very exciting myself, is the nodeJS side of things. Here we are seeing a tremendous growth compared to other technology. And the reason is exactly because it’s connected to the first-hand experience people have. Those who master the browser and the DOM will eventually be able to reach customers and create systems which are platform independent, international and dynamic.
Do we have anything to learn from this? Bucket loads! But we should never lose sight of the fact that we also have much to teach.
And with that my pondering stops for today 🙂