something strange happened after implementing a first interpreter

I've been reading about programming language implementation for a while now as a hobby. Last week I had the knowledge, time and energy to implement in C an interpreter for a portion of Scheme. I'd long wanted to do just that. It was definitely the biggest thrill I've had in programming to implement an interpreter and see it run while interacting at the REPL. :-)

Then suddenly I had a "now what?" moment. Even though I didn't implemented an efficient interpreter, I now understand the basic, general principles of language implementation (interpreter or compiler.) Clearly I'm not a highly skilled language implementer after my little bits of book reading and experimentation. Even so, the big mystery of programming languages seems mostly gone. It is a strange feeling after years of thinking programming languages were something the gods handed down to some mortal on a hill.

Programming languages are certainly the part of computing that interest me the most. That said, I don't think I'd ever find a job that suits my life better than my current web programming job. From what I understand programming language implementation jobs are rare anyway. Stopping work to indulge in grad school for five years is not an option. Maybe when I retire. Developing a hobby language that fits the bill of dynamic languages like Scheme/Perl/Python/Ruby seems like an unnecessary reinventing of the wheel.

I'm not sure what will be next for me and programming language hobby time. Anyone else have similar experience wondering what to do with newly acquired abilities?

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

There are many paths before

It is a strange feeling after years of thinking programming languages were something the gods handed down to some mortal on a hill.

My initial languages were Basic, Logo Writer, Visual Basic, C++, Pascal, and Java. I can say for certain that I've never been under the impression that programming languages were handed down by 'gods'. Except maybe to curse us. Perhaps if my first languages were ML, Prolog, Lisp, and Smalltalk, I'd feel differently.

I'm not sure what will be next for me and programming language hobby time. Anyone else have similar experience wondering what to do with newly acquired abilities?

There are many paths before you...

Down the discovery path, you can learn, play with, implement parts of other languages and designs. Logic programming, functional reactive programming, rules based programming, constraint programming, learning systems, cellular automata, esoteric languages like Snusp or SKI calculus, software transactional memory, concurrency, distribution, capability model security, etc. You'll probably never finish a language on this path, but you'll learn a lot.

Down the implementation path, you can learn about optimizations, compacting vs. generational mark and sweep garbage collection, just-in-time compilation, integration with virtual memory and persistence, parallelization, resource management, networking. You'll learn about transparent 'intrinsic' functions and other hidden optimizations (e.g. perhaps replacing 'lists' transparently with 'ropes' under the hood will speed things up? it might be worth trying...). You'll learn about automated code transforms, i.e. to CPS or SSA. You'll learn about threads and fibers, about cache thrashing and affinity, about allocator designs to support multi-threading without hindering GC. You'll struggle with portability vs. performance concerns. You'll struggle to make a bignum library that works well with your allocator and GC. It can be exciting to set some high goals for an implementation then work to make them come together. A lot of people with an interest in programming languages lack the inspiration or motivation to come up with a new language, but do help others implement their visions.

Down the design path, you find yourself irritated with something the languages you've used have been doing poorly, or recognize that programmer tools are insufficient for some grander plan you have, and you start banging your head pounding out pages of brainstorm (mostly drivel) coming up with a good way to solve the problem. If you're smart, you try APIs and frameworks before resorting to a new language. It helps if you've worked on discovery for a while.

The problems I'm tackling for my own design are: weak code distribution, resulting in poor locality of code to resource; data-binding and latency issues between update of a data resource and seeing said update; cache coherence for large, structured calculations after an update to a sub-expression; disruption tolerance - in particular, the need to explicitly identify and recover from disruption that gets scattered throughout code; resumable exception (Lisp and Smalltalk seem like the only languages supporting it); dangerous concurrency rife with race-conditions (not even actors model helps); persistence, and the challenge of achieving persistence in a manner suitable for distributed cross-platform computation; DLL hell and fragile coupling between versions of libraries and plugins; gathering of data scattered across code or remote resources; security against both accident and malice, and effective management thereof; and serialization for IPC (esp. with encryption).

There is plenty more out there to target. Other people focus on languages that can, for example, compile for the GPU. There are languages designed for artistic endeavors, like PovRAY and Inform 7. There are languages to describe missions and military concerns (stealth, surprise, compromise, etc.).

So, if you're interested in programming languages - if you get a 'thrill' out of implementing bits of Scheme and understanding the language design - there's plenty of things to do, plenty more language designs to grok, plenty more implementation details to understand in context, plenty more challenges awaiting both design and implementation.

I've followed each of the above paths quite deeply in pursuit of my own 'grand vision' which (in my happy, perhaps delusional mind) will achieve something far more interactive, open, and secure than the current Internet. My initial pursuit of programming languages began only upon realizing I couldn't achieve it just through programming in existing languages, so I've never been at a loss for what I want to do with the skills I gain by perusing sites like this one.


It seems to me you ark asking about the kinds of jobs language lovers can enjoy and profit from?

dmbarbour and Ehud

dmbarbour and Ehud Lamm,

Both of your replies are perfect and each seems such an appropriate length. :-)

Programming language work is so interesting that it could consume all my time. There are so many things to investigate and implement. If something is going to take so much time then perhaps I'd want it to be a job so I don't end up with two jobs: the one for pay and the one for play.

Interesting replies. Thanks for both.

The highest concentration of

The highest concentration of PL PhDs/luminaries is probably in the Google cafeteria at lunch time. Turns out PL folks make the best high-end programmers, who would have thought!?

Learning PL will develop yourself as a programmer, and you could move into higher-end code work if you want. Its not all about PL implementation, its about thinking about problems in different ways, and designing the best abstractions.

I'm still a little surprised that Google

hasn't promulgated a Google programming language (or languages). By "promulgate", I mean market and publicize in the same manner that other googly technologies have been promoted (Gears, Android, the Chrome browser). Google certainly has lots of in-house tools, some of which may qualify as PLs, and doubtless Google engineers have numerous such projects.

Maybe its just wishful thinking, dreaming of a white knight on a fiery steed, looking to liberate mankind from various ill-formed and/or proprietary languages pushed by various system vendors... :)

Python, Sawzall

Google famously limits most of its projects to Python, Java, C++, and JavaScript. Apparently they believe standardization outweighs innovation in this area, for the most part.

However, one might consider Python to be the "Google programming language" given that they pay Guido and Alex Martelli (and others?) to work on it, and they've been a significant force promoting it through initiatives like Google App Engine.

And of course there's Sawzall, which is about as googly as a language can be (and unfortunately proprietary, which is also a fairly googly trait).

Python's growth

I've found the existence of Python to be much less annoying to me in the last couple of years, I think because the discussion of the language has grown up (Guido himself, in particular), and because some PL-interesting projects involving Python have come to light.

And RPython looks like a dialect that frees me from my sense of unease reading Python code. Forbidding TAB/SPC mixing in indentation would also be a good thing, to keep to meaty bike-shed-colour matters.

be careful what you wish for

I'm still a little surprised that Google hasn't [...]

Give it a little time (days, weeks, perhaps a few years).

Maybe its just wishful thinking, dreaming of a white knight [...]

Be careful what you wish for. The last verse of that song was sung by Sun in the key of Java. And look where that got us.


And look where it got Sun...

And look where it got Sun...

language design as second phase career

Most commercially successful programming languages come from someone who does some other systems programming task, but knows a bit about language design and implementation, and then makes a language that makes their previous job easier. Examples: C, Perl, Python, Lisp, Emacs Lisp, Erlang, etc.


In-house languages and DSLs

People often think about designing the next mainstream language, or working on a C++ compiler, but I guess most PL work is done on in-house languages and DSLs that are either part of a larger product or used in-house and are entirely unknown to outsiders. I try to mention both kinds of projects from time to time, so people not forget that these offer good opportunities for language implementation and design work.

Find an application

To me, the most fun part of language design is matching your application(s) of interest to the PL. If you don't have an app in mind, it's hard to come up with a good language; that could explain your feeling of "what next."