Hacker News new | past | comments | ask | show | jobs | submit login
Ask PG: How is Arc coming along?
116 points by shawndumas on Aug 12, 2015 | hide | past | favorite | 47 comments
I noticed that 500 days ago you said that you'd be working on Arc again [1]. How's it going?

-----

[1] https://news.ycombinator.com/item?id=7493993





Perhaps it is already complete.

Inspiration:

A manager went to the Master Programmer and showed him the requirements document for a new application. The manager asked the Master: "How long will it take to design this system if I assign five programmers to it?"

"It will take one year," said the Master promptly.

"But we need this system immediately or even sooner! How long will it take if I assign ten programmers to it?"

The Master Programmer frowned. "In that case, it will take two years."

"And what if I assign a hundred programmers to it?"

The Master Programmer shrugged. "Then the design will never be completed," he said.


He handed over the Hacker News leadership to a team of people a while ago[1], so that might be a reason for why he hasn't been working on it.

[1] http://blog.ycombinator.com/meet-the-people-taking-over-hack...


PG's comment was made on the post you are referencing.


Without deadline there will never be need to compromise. You can keep thinking about it year after year and seek perfection that is The Hundred-Year Language.

http://www.paulgraham.com/hundred.html


To achieve great things, two things are needed; a plan, and not quite enough time.

-Leonard Bernstein


Plans without deadlines are just dreams.


  > Plans without *action* are just dreams.
There. I fixed that for you.


Not really. If the action drags forever, they remain just dreams (like some great unfinished novel a guy writes for 50 years and gets to 60.000 pages).


Sometimes dreams are long-term plans


Sometimes plans are short-term dreams.


Waste of time. In 100 years the question will be what lifeform will you be working on / with.

Programming languages will look like stone tablets.


> Programming languages will look like stone tablets.

Trend look the opposite actually... it's more like everything would be programmable, maybe even virus and bacteria


But why would YOU program it? The AI can do that.

It's kind of going full-circle really. You don't need to program a horse in C or Lisp to ride her.


You either would program it (even by nature language) or the human race will cease to exists.

However, you seem overly optimistic about our future in 100 years. Assembly language is roughly 70 years old, Lisp is in its middle age already (56) so that means by 2040 we will have our first 100 years language and by 2060s (depending if Lisp or Fortran are still alive) we would have our first 100 years hight level programming languages.

Honestly I don't see any AI going even close to what you suggest by any of those milestones.


I could still be in the workforce at 2040, working in 100 year old languages. Ugh.


A dream is a goal without a deadline.


Meanwhile, JS shipped inside of a month, and look how far it's spread.


It did, but then look at the languages it borrowed from: Self and Scheme.

Yes it has warts borne of ridiculous deadlines, but it also has elegance learned from prior art. All things considered, JS is a good advert for standing on the shoulders of giants, imho.


Good projects last for decades and outlive companies. See: LibreOffice ;)


Some context (say, for example, a brief description of Arc) might be helpful.


LISP dialect invented by Paul Graham. This site is written in Arc. Afaik that is the only mayor application so far.

http://arclanguage.org/

http://www.paulgraham.com/arc.html


> This site is written in Arc.

Is that why nobody can ever add any useful new features to it?


BURN


Meh, even if it was true, I consider that in itself a feature.


That's what a closed-source fork of a forum written in a language almost no one cares about[0] will get you.

[0] and a community which harbors a pathological fear of making Hacker News look and act like Reddit


There's also more up-to-date documentation/support by the community: http://arclanguage.github.io


Clojure won.


You're getting downvoted, but I think there might be some truth to this.


As likely one of the few people outside of pg to use arc in production, I agree Clojure won- It has about 80% of the features that arc pioneered, and many additional features that leave arc in the dust. (Most of the remaining "good ideas" in arc are syntactic sugar which are just too controversial to see mainstream adoption.)


What features did arc "pioneer"?


Shoot, it's been too long for me to sort out the various features now, but you can read about the cooler parts of arc here: http://old.ycombinator.com/arc/tut.txt which won't look very exciting to any Clojure programmer nowadays.

Probably its most useful innovation, even if it sounds mundane now (and I'm not 100% sure it was first on this- feel free to correct me) is that it was the first Lisp to realize that sequences and key-value associations are both equally important and should both have first-class status in the core language (with proper literal and shortcut accessor support) ... so like clojure, arc is more of a MLISP (Map and LISt Processor)

However, the posted link doesn't have examples of the crazy syntactic sugar stuff pg did that I think was added in a later revision.


You're definitely not the only person using Arc in production. Hubski is written in Arc and still under active development.


In my defense, I wrote "one of the few people" :-)


In that case, is there any prospect of an Arc-flavored dialect of Closure with the syntactic sugar put in?


That would require some very ugly "under the hood" modifications to clojure that are not easy.


Could you list the features that Arc pioneered?


He should probably rephrase. Clojure has won as an acceptable Lisp in business. Lisp'ers should probably get behind it to help to grow it even more. Once it grows, other Lisps will have an easier time.


I think a lot of Lispers are; it got a great reception at Lisp50 from at least some of the oldbies in attendance.


I'm not really convinced Clojure is even one of the reasons, though. It seems like they have different goals and mindsets.


I'm fairly certain Arc influenced Clojure in some way.

Especially the naming of functions such as fn, pr, prn and the likes.


I can't help but imagine a party, not entirely unlike the opening scene of Animal House except that everyone is older. Our protagonist enters and a hipster host quickly leads him away from the action and into a side room saying "Rich, Paul is working on a new dialect of lisp, too."


I'm not sure, I've used 'fn' in toy programming language concepts before I ever heard of arc or clojure. I mean, its not impossible and I certainly don't know about the others, but its also not impossible that they were independently named.

However, I do think its very likely that arc influenced Clojure, even if just a little.


SML uses fn for lambda.


Ah. Actually, now that you mention it, I probably saw it there before I ever thought of using it.


PHP won.


lol.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: