HN Theater @HNTheaterMonth

For anyone interested in background on Datomic, these are fun and informative resources.

The talk where I realized Rich is a data head too. Introducing his design goals and life experiences with db(s). https://www.infoq.com/presentations/The-Design-of-Datomic

"Clojure for the Brave and True" author, Dan Higginbotham, on key themes and overview. http://www.flyingmachinestudios.com/programming/datomic-for-...

Congrats to the Datomic team and Cognitect. I hope this move to cloud opens Datomic's design principals and ideas to a wider user base.

I would suggest to you that Clojure has a very clear "the Clojure Way" I would say, more so then almost any other language. It goes beyond syntax to a pretty deep design level. Its not even all that unique to clojure, you could do 'the clojure way' in other languages, but they usally dont.

Some of the importent keynotes from some of the early conjs are really importent, and they really went deep in the community, you can see it in almost every major library.

Another commenter has mentioned the notion of 'simple' but thats not really all that clear. I not enougth of a writer to tell you about it, but I can give you the resources:

- This classic got Clojure on the map for many people: Are We There Yet? - Rich Hickey www.infoq.com/presentations/Are-We-There-Yet-Rich-Hickey

- Simple Made Easy - Rich Hickey - http://www.infoq.com/presentations/Simple-Made-Easy-QCon-Lon...

- Simplicity Ain't Easy - Stuart Halloway - https://www.youtube.com/watch?v=cidchWg74Y4

- Ousterhout's Dichotomy Isn't (Part 2 of the Simplicity/Power/Focus Series) - Stuart Halloway - https://www.youtube.com/watch?v=cidchWg74Y4

- The Language of the System - Rich Hickey https://www.youtube.com/watch?v=ROor6_NGIWU

- Hammock Driven Development - Rich Hickey https://www.youtube.com/watch?v=f84n5oFoZBc

In addition to these, there are many talk that are about people trying to run with these ideas. Datomic is a example of taking these ideas to the max, it really standas as a nice example. You can find others in talks by David Nolan for example, but there are tons of applied talks.

- The Design of Datomic - Rich Hickey http://www.infoq.com/presentations/The-Design-of-Datomic

This model is somewhat similar to how Datomic models transactions -- Datomic uses immutable datoms and includes a transaction ID with each datom: "A datom consists of an entity, attribute, value and transaction (time)" (http://www.datomic.com/rationale.html).

However, Datomic's design is based on a single, centralized transactor that pushes out all the new transaction information to an index that is distributed to all its clients (peers), whereas NBTA would enable distributed transactors.

See "The Datomic Data Model" http://www.datomic.com/rationale.html#DataModel

From Datomic's FAQ http://www.datomic.com/faq.html ...

How does Datomic provide ACID guarantees?

The transactor serializes all transactions, and each transaction runs against a stable view of the database, and succeed or fail in their entirety. Transactions are not acknowledged until after they are logged in storage. Atomic read/modify/write operations can be performed by database functions, which run on the transactor within the transaction. Note that Datomic can provide ACID guarantees without utilizing read-transactions, nor read locks, due to the presentation to the query engine(s) of the database as an immutable value.

---//---

Rich Hickey on Datomic's design...

Intro to Datomic (12 min) http://www.youtube.com/watch?v=RKcqYZZ9RDY#!

The Design of Datomic (60 min) http://www.infoq.com/presentations/The-Design-of-Datomic

The Datomic Architecture and Data Model (46 min) https://vimeo.com/45136212

⬐ pbailis

Very cool, thanks! In general, immutable data items makes NBTA much easier; there's only one element in either 'good' or 'pending' for each data item at a time. The benefit of a single centralized transactor/ID generator is that it provides a simple total ordering on updates. The downside is that updates must (often) complete sequentially to guarantee safety and that the centralized service can become a bottleneck/challenge during failures. That said, I'm a big fan of the Datomic vision and look forward to seeing what they're able to accomplish.

This talk by Rich Hickey is what got me into Datomic.

http://www.infoq.com/presentations/The-Design-of-Datomic

I would probably have used it for the system I'm making now, as it's a very good match to our requirements, but 0.1 is 0.1, and the docs are kind of sparse so the learning curve for Datalog queries was too steep.

I believe this is the Clojure/West talk from which Hickey is being quoted:

http://www.infoq.com/presentations/The-Design-of-Datomic

I don't think we need a slogan to promote Clojure, we need more killer platforms. We need problem-spaces where when someone asks "What's the highest quality implementation of X?" people say, well, clojure has XYZ, it's way ahead of the pack.

Datomic (http://www.infoq.com/presentations/The-Design-of-Datomic) is going to be one of Clojure's killer apps -- it adds durability to Clojure's immutable, persistent data structures, and it rethinks datastore design, from the ground up, to take advantage of modern storage services.

⬐ Estragon

Datomic is super cool design-wise. It'd be great to see some apps actually using it.

Evidently you may be trapped inside yours -- I made no mention of VertexDB nor did I suggest any particular database.

However, for example, Datomic (http://www.infoq.com/presentations/The-Design-of-Datomic) is interesting, which is distributed and uses Amazon's DynamoDB storage service.

⬐ cheatercheater

I'm yet to see a Django based website that outgrows any old rdbms for auth. Care to back up your claim that it's bad? Same for admin. You do realize that Django's admin is just a search result of objects defined in its ORM, and that objects defined in its ORM are very obviously stored in an rdb - right? What would you like admin to use for managing data stored in an rdb? There's barely anything other than the models being managed that touches the database.

Edit: just before my comment gets skewed, my question is: why would you ever use django for a website big enough to warrant this sort of worry? Let's recap: big websites make money which is used to hire coders. Those people are at that point probably porting your website to Agda, or some other cloudscale technology. Sounds like you're the same kind of person who complains that Bash doesn't support OpenMPI and that JavaScript doesn't do fib() well enough and that this here carpentry hammer can't break diamond. Wrong application, dude. Django is for medium-small websites, not for your Cloneddit, Clonebook, CloneSpace, or news.yclonebinator.com; get a different hammer.

⬐ espeed

It looks like you're new here. Make sure you understand the context before you go trolling else you might be mistaken for a chatterbot with all the non sequiturs.

⬐ cheatercheater

I'm not new to Django or programming in general, though. Making generic statements like your original post and catering to hip trends is one thing, backing up your bs is another. Obviously you can't answer the direct and very simple questions I stated, so you resort to pointing out your paltry 3.79 karma average in an attempt to sage me. Nice.

⬐ espeed

My comment was about simplicity and paradigms, with allusions to Thomas Khun's The Structure of Scientific Revolutions (http://en.wikipedia.org/wiki/The_Structure_of_Scientific_Rev...). You must have missed that because you're trying to pick a fight about scalability, which has nothing to do with my original comment.

Most of my work is with graphs so, for example, if you want to use a graph database to build a social graph with a Bayesian network, you wouldn't use most of Django's components such as auth and admin because they're tied to the ORM -- most of Django would just get in your way. It has nothing to with a Django-based site "outgrowing" auth.

Try to break out of your relational mindset and understand the thread before you decide to go off. And regarding my "paltry 3.79 karma average," you do realize the median for the leaderboard (http://news.ycombinator.com/leaders) is 3.975, right? Again, know what you're talking about.

⬐ cheatercheater

Given a very specific need of working with a bayesian network based on a social graph, i can see how your point was almost-valid. However, there's no mention of this sort of thing in your top post. The original link is about breaking up web apps into services, a'la amazon. I think it is you who does not understand the topic at hand, but that's a fairly trite way of argumentation, so let's just say that you didn't mention your very specific needs which were needed to appreciate just why not using an SQL db for auth is simpler. The fact is that it's probably not, it's just that it fits your solution better, so the total expense is lower. Having auth or admin backed by an rdb is not more expensive, auth supporting your idea or admin supporting your idea which are two executions of the concept that are very unusual compared to what is normally done with Django. Again, you're complaining your hammer can't be used to screw in things. Get the right tool, or if it doesn't exist make it. Django is not the right tool. So the whole thing is a composition of "why didn't you say so" and barking up the wrong tree.

Of course, the lack of actual motivation in your top post can be explained through being inarticulate as in the paragraph above; it can also be explained by you trying to mesh together a few easy and hip claims. The later motivation could well be something you worked hard to muster up only after initial critique, and given that its connection to the top post hinged on a shaky premise I think that's what has happened here.

Regarding your "paltry 3.79 karma average", the leaderboard is sorted by karma amount, not karma average. There are people on that list who have about 10x as much karma average as you do. And even the guy with the least karma on that list still has over 4x as much karma as you do. Well done computing the median of a biased subset of people. Biased because it selects people with the largest amount of points which easily comes at the expense of post quality. Please read about the bias of controlled, not-randomized studies before you go throwing around statistics you can't interpret. I suggest looking at the tree graph found on page 7 of "Statistics" by Freedman, Pisani, Purves, since you like graphs so much. Coming up with meaningless numbers will only shut up someone troubled by innumeracy, and I am not affected.

⬐ espeed

Damian, I think I see the disconnect here.

The OP/thread is part of an ongoing conversation about the Python community at large and Django's stronghold on the community mindshare because a large portion of development resources goes toward the Django platform rather than to decoupled, external libraries (as noted in the talk).

Most new Python Web developers choose Django as the default option because that's where the mindshare is (hence the repeated "we'll use Django, of course" in the slides), and Django's design and momentum lock the Python community into a certain paradigm, even though new paradigms are emerging.

Kenneth's talk, my comments, and similar comments by people like Mike Bayer (http://news.ycombinator.com/item?id=4079892) were about breaking that stronghold. My example regarding graphs was just an example for you, but my original comment was about the bigger picture.

BTW you were the one throwing around karma average. Have you looked at yours, and are you aware how it's calculated? -- it's not really a good indicator to use when trying to discredit someone.

⬐ cheatercheater

At first you have tried argumenting with completely unrelated facts, then tried ad-hominem, then you have tried and failed at coming up with a speculative interpretation of your own original post in a failed attempt to pretend those were your motives. Finally, in a last-ditch attempt, you have strongly detracted from your top post.

> The OP/thread is part of an ongoing conversation about the Python community at large and Django's stronghold on the community mindshare

No, it's not. The highest rated top posts for this link talk about complex monolithic code vs flexible code made out of bits which is however lacking in features. This includes the post you're trying to defend using this detraction strategy. The question of Django having or not having mindshare in the python community is secondary to this, and isn't even a topic of the largest minority of comments I've seen on this link.

To support your claim that it's all about community you come up with a short post that is several levels deep, and purport that the original talk bemoans the fact that Django is domineering the community. The talk barely mentions that Django is just a popular choice, it is your completely disconnected analysis that he was complaining about Django's harvesting of the "python mindshare". In fact it mentions Django because according to the talker it's just the top competitor to what he's selling. This is standard course of action when you're presenting a new contender in a space and has nothing to do with "the state of the python community".

In the same way as you try to paint over the past repeatedly changing what you've meant with the original post, you do the same with the side discussion of karma. If it were really as unimportant as you say, why were you defending it in a post just above? You lack consistency.

Yes, I have looked at my karma average. The same page that displays it will also show you that I barely, if ever, post here, and if you try harder you will find out that I've registered about a year ago to post on my own content after it submitted here. This should in fact display to you that this website is not as important to me as you think it is. It's not that I'm new here - it's just that I'm not, you know, "a part of it". However, your comment was just so disconnected from the discussion, I felt compelled to point it out, and I'm glad I did so, because the resulting trainwreck should give you, and other people here, some fodder for thought - not everyone buys your junk logic. You can see it as a pedestrian bystander jumping in to rescue people from a car crash. Not a member of emergency services per se, but the situation warrants action strongly enough that some bystander felt the need to do something.

Threatening people with your amount of experience, jumping around in the supposed meaning of your point, ad hominem (oh, now my comment means this.. no, it means that and you don't get it because i'm so much better than you.. oh, no, it means something else; oh btw, I've been stalking you, watch out!) don't really form a way to have creative and intellectual discourse with anyone. Glad to stomp that out for you, you may thank me later once you've become accustomed to actually admitting when you had made an error, rather than feeling the need to spin it, pretending it's something else.

In my mind, when I made the original comment to your top post, the fallout looked like this:

(original premise): you say that it's very bad that you have to use an rdb for admin

(baiting answer): yeah, so let's just use (nosql database chosen as an especially ridiculous example)

(your answer): you would need to use a nosql database for admin to do (fringe application)

(my answer): but (fringe application) is not what Django was made for. Wrong tool.

(your answer): ok, here's a better illustration. We truly do need Django to be able to do (some thing which is tied to an rdb), but it's much better done if it were in its stead using (some nosql technology).

This workflow has happened (except for the last part), but it came with a lot of bickering and manipulative speculation, which makes me think that you hadn't even noticed it, much like someone who after getting a speeding ticket attributes it to police depravity and oppression of the common man. Therefore, I decided to point it out. Is it really so difficult to admit when you've shot off on a tangent? I liked the link in your top post, quite a lot in fact, but the comment that followed it was of no or negative value because you have chosen to illustrate with non-examples: concepts that do not support your claim. In addition you did this because you really really like nosql databases and probably feel the need to bash on everything that uses SQL from the angle of it using SQL, and sometimes can't control this need. In this way your nature has really messed up the execution of your intentions. I'd have totally upvoted you had you just made a link and no comment, and probably defended the link adamantly from anyone criticizing it. I'd have loved it even more had you accepted that maybe what you thought wasn't entirely correct. But neither of those two things happened. Instead, you bicker and manipulate, finally resorting to trolling through stalking, hoping that you'll find out my name (it's not Rumpelstiltskin) and somehow shock me or make it more personal. And even worse, you fail at stalking, but while doing that show yourself as a big jerk. And I'm not saying I wasn't being negative, but I'm trying to keep form, whereas you display somewhat of a sleazy, vaguely adversarial, win at all costs quality in your discourse, which shows lacks of consistency. Consistency is the most important thing when conveying information, and without it you end up being viewed as a charlatan. Think about it.

⬐ espeed

Here's the problem -- you fixated on this:

(original premise): you say that it's very bad that you have to use an rdb for admin

...but it's not what I said; this is what I said...

"Django locks you into a RDBMS if you want to hook into all of Django's components like auth, admin, etc"

I'm not saying that it's bad that you have to use a relational database for auth, I'm saying that if you don't use a relational database and the ORM then you lose admin, auth, third-party apps, etc. Strip all of that out and what do you have left? See slide 71 (https://speakerdeck.com/u/kennethreitz/p/flasky-goodness).

And I have said this several times before (http://news.ycombinator.com/item?id=2911275), so no, I am not trying to retrofit my argument.

The talk barely mentions that Django is just a popular choice, it is your completely disconnected analysis that he was complaining about Django's harvesting of the "python mindshare".

If you don't think that's at least the subtext of what the presentation was about, look at the slide for Kenneth's primary thesis: "Open Source Everything" (slide 10 - https://speakerdeck.com/u/kennethreitz/p/flasky-goodness). And then go through the presentation again to see what he means -- "Single Code Bases Are Evil" (slide 45).

⬐ cheatercheater

The slide just says "open source everything". How you conceived that this in turn means "django has ingested the python community" is beyond me. Probably the same mental flaw that makes you a stalker. Why should I be replying to a stalker again?

BTW, did I mention: "stalker"?

⬐ None

None

⬐ None

None