Datomic Update: Client API, Unlimited Peers, Enterprise Edition, and More
blog.datomic.com
blog.datomic.com
Bummer.
$5k is less than my side project makes in a year.
And the average annual system maintenance budget for a small client.
This is just not something we've got a good, reliable answer for. E.g. I'm pretty confident that his creating and being the BDFL of Clojure wasn't sufficiently remunerative.
That said I also think it is worth to regularly revisit what an open sourced Datomic could do for adoption of Datomic itself as well as for Clojure (edit: not to mention what the impact would be like for the world at large). I'm optimistic that there are monetization opportunities through SaaS, consulting and other licensing concepts.
Ideas?
If Datomic achieves widespread success with both community and enterprise editions, Rich Hickey would have an impressive range of opportunities.
The man, however, is far more accomplished than I so I can only assume he's thought of that and firmly rejected it. Perhaps, although he wants to see his kids go through college, he also wants to be a big part of their lives. Perhaps 2 massive, popular over source projects would get in the way of that.
I think it would. What it wouldn't do is give Rich Hickey the control he currently has over his life and his projects. He'd likely end up getting paid to work on Datomic by some big company instead of running his own as he does right now.
Unity - free community version, pay monthly for extra features and to remove the splash screen, option to stop paying and keep license but stop receiving updates
UE4 - free, source code is on github (but not FOSS), takes 5% of your gross revenue after first $3000
Lumberyard - free, source code available (but not FOSS), you must either run your own private servers or use AWS
Early adopters will take most of the financial burden in order to recover the development costs, and open sourcing will happen once it's more strategic to use Datomic as a way to promote their knowledge and services broadly than it is for them to make money directly from the product.
Support tiers will remain even after open sourcing, of course. I don't expect it to turn into abandonware.
But as a developer I think the current license is a bit of a shame. I've played with Datomic, think it is fantastic, and would love to use it.
However:
- After the closure of products such as FoundationDB and Parse, I will never have a significant dependancy on a closed source product again. (maybe if the company was a behemoth like Microsoft or Oracle, but these days probably not even then.) [1]
- As a little guy, the $5000 I would need to pay after one year to keep getting updates to the product is just too much. One could argue that Datomic are looking for bigger fish, but something to consider is how important it is to get people using your technology (al la Microsoft 'developers developers developers!') Little guys can go on to be bigger guys making the important technical decisions. There is post above looking at the licensing done by game engines such as Unity, where larger companies pay more and the little guys are given a break until they are making a profit. Hopefully we will see more of this in the future.
[1] https://www.wired.com/2015/03/apple-pulls-plug-tech-company-...
I'll just keep using the free edition for my toy/side projects.
Datomic free edition also sounds great for small open source projects.
It wouldn't be suitable for large scale free SaaS websites but I feel that's a tiny slice of developers.
Right now, I'm sifting through the internet to find any other (up to date) similar system supporting datalog querying.
Let me know if you know about one!
(Only one I found so far being Cascalog [1], which is unfortunately only supporting old sluggish hadoop)
Will think hard about the suggestion of using some functional language instead also ... you for sure got me thinking! :)
Feel free to share any pointers regarding this!
In the meanwhile, sharing my thoughts so far: What I have so far thought was pretty "unique" about datalog is that it support "named queries", which can be used as a kind of "duck typing" classification, to build up a model of your domain of interest that is always computed on-demand, leaving the raw data unchanged in it's default ("normal"?) form.
Because of the "named queries", you can combine lower-level queries in higher-level ones, to create a hierarchy of "dynamically generated concepts", which I haven't seen any other technology be able to do.
(While named queries are conceptually similar to views in SQL, they are so much more natural to work with.)
I know this by experience from my dabblings in prolog [1], but haven't found any large-scale prolog data system supporting distributed computing (I understood it has some features making it unpractical for distributed computing and more, in a way that datalog doesn't).
[1] http://www.diva-portal.org/smash/record.jsf?pid=diva2%3A3988... (PDF: http://www.diva-portal.org/smash/get/diva2:398839/FULLTEXT01... )
Read this section from the Elixir docs: http://elixir-lang.org/getting-started/pattern-matching.html
Also this page from the Haskell tutorial: https://www.haskell.org/tutorial/patterns.html
Fun fact: early versions of Erlang (which Elixir is built upon) were implemented in Prolog, and IIRC what would become Erlang later was originally intended to be an extension of Prolog. There are no new ideas.
This is the purpose of functions in most programming languages. A function is kind of like an HTML template, but for code. So rather than have `<h1 title="{{bar}}">{{foo}}</h1>` you have `function(foo, bar) { return h1({title: bar, text: foo}); }`. DOM tree, AST ("abstract syntax tree"), relational data, same thing: a graph, and most programming languages can represent raw graph nodes/edges easily as lists of strings. You can write a simple tuple-matching engine in Javascript: https://gist.github.com/notduncansmith/7858d27a2b875ff10b84
Obviously that's not a production-ready piece of software, but it highlights an important lack of distinction between "database" and "program" that wasn't readily apparent to me as a developer for a while.
HTML, source code, and user input are all the same thing (just data, really). One program's source code is another's input data, all the way down.
If you want to get existential, you can extend this even down to matter itself as simply a representation of information: imagine a physical machine in a factory that winds jack-in-the-box toys on their way out the door. Same thing as a higher-order function.
function wind(toy) {
return toy.lever.rotateByDegrees(360);
}
Moving from the abstract to your concrete case:
> build up a model of your domain of interest that is always computed on-demand, leaving the raw data unchanged in it's default ("normal"?) form. [...] (While named queries are conceptually similar to views in SQL, they are so much more natural to work with.)
SQL views don't compose at the same level that Datalog expressions do, but that's because they aren't typically expressed as raw graph data (i.e. lists of strings where each list item is a node in the graph). If you could just name arbitrary chunks of SQL AST and mash them together, it would be just as composable (though perhaps more cumbersome due to boilerplate).
One final thing before this already-too-long comment comes to a close: you might give CouchDB a look. It's a document datastore that offers map/reduce views written in Javascript (use as much or as little composition/abstraction as you want). The distributed features are pretty cool (bidirectional sync, versioned documents) and writing apps with it is awesome, especially if you can leverage supporting projects like PouchDB or Couchbase Lite. Not sure if it fits your use-case but it sounds like you might find it personally interesting.
Hopefully this has been helpful!
Posted a separate HN link: https://news.ycombinator.com/item?id=13064674
This for a product which required a big leap of faith already (designing your data access around a radically different DB with enormous lock-in, supplied by a small consultancy).
Thankfully on superficial inspection it seems like some of the most egregious clauses have gone (Ctrl+F for "disrepute" and "immoral" here: https://blog.goodstuff.im/datomic_license to get a flavour).
I'm wondering though: has anyone here ever used datomic after some laywer looked at their previous license? In the unlikely case anyone from datomic cares to comment: did changing the license help sales?
Most of the cited examples in the blog post are common standard language used in license agreements. I'd say the interpretation of what they means is idiosyncratic. One could argue that the clauses were not nice, but that's a different argument to the one presented.
Their current Eula you linked is pretty short, I'd probably argue with the indemnity, but not unless I was paying enough money that someone would care. Average legal fee to discuss a license ~600 USD an hour, their license is 2,500 USD. Most people have tougher license terms in their office lease!
To be clear, I don't have a bad opinion about Cognitect (other than their choice of license).
But, I did run a commercial team (lawyers and negotiators) for a software company that handled hundreds of incoming and outgoing licenses: including ones to major software vendors like Salesforce, Oracle, Microsoft and IBM. I am not a lawyer though ... but, sometimes I play one on the Web.
I'm not trying to tell you the terms are "reasonable", that's a matter of opinion. I'm telling you that the way in which they are interpreted in the blog post is not in line with how they would commonly be read. It is not a fair reading of the terms as far as I can see - consequently saying they have a "bad choice" of license makes no sense since the clauses are being cherry-picked.
Second, people often think licenses have to be 'reasonable', they don't. They have to be enforceable and in your businesses interests - Cognitect clearly thought those terms were in their interests.
I'm not going to comment on the clauses further, you've clearly read them one way and I was offering the observation that I would not have come at them from that perspective. I guess the best way to have found out their intent would have been to call them up and ask them, getting clarity added when you negotiated the deal terms where it was required. Most commonly, people have terms in their licenses because of specific issues they are concerned about and when you understand that something can be done to "fix" the difference of perspectives.
Finally, the new license looks eminently reasonable, so perhaps they noted the concerns and have now fixed them, no?
* No limits on process count
* Datomic Pro Starter is now limited to one year of usage, then you need to pay $5,000 for Datomic Pro
* Riak, Couchbase, and Infinispan are now legacy storage and only available through Datomic Enterprise.
Edit: sounds like that's no longer the case. See sibling comment.
As was discussed in Datomic Slack:
"Starter was always intended as a path for customers to explore and use Datomic in a low-risk, low-cost approach as they developed their applications and moved toward production. We feel that a year is generally sufficient time to evaluate a product and develop a business application around that product. If you feel that you require a longer period to evaluate or develop against Datomic, please contact us."
I don't understand why Rich doesn't understand that he'd make far more getting a tiny slice of the huge pie that open sourcing would provide. Yes, they've changed the licensing approach because it's obvious the previous one didn't work out at all. But they haven't done enough, and by the time they don't figure this out, people will be stuffing everything into cloud services, even if they are inferior approaches.
No updates after a year? To the new wave of developers, this is foreign. Most won't know what that means, but at first glance, it looks like you'll be paying $5,000 in a year. Even the most green developer knows you need updates. Also, it's unclear what a 'system' is without digging around, and many potential users won't bother.
The problem is that Cognitect knows how awesome Datomic is, but the world doesn't. Many potential users are not going to try software they know they're definitely going to have to pay $5,000 for in a year. Updates are critical. Cognitect should go for massive adoption and nothing else. They'll make far more money.
This thread shows continued confusion and uncertainty about the license, and if a potential user googles 'datomic license' they are likely to find this thread. After reading up and down it, they will be deterred from getting into it all. Why try Datomic and get into a potential mess with your employer over licensing issues/costs in a year when you could play it safe? I'm sure those who think this licensing approach is the right one are clear the advantages of Datomic are worth the risk, but they are not paying attention to how software is actually adopted at scale.
It's so sad it's almost 2017 and this issue's been around since 2014 ( https://news.ycombinator.com/item?id=7657701 ) and this is the solution. The word Datomic means nothing to 95% of mainstream tech. This change is a modification of their old model. They need an overhaul of how they are doing this.
I was lucky in that my first job was at a small non-software company that forked over huge amounts of cash for Oracle because it literally enabled their business to succeed (mysql and postgresql were nowhere close to being able to do it at the time). And yet I've seen fellow developers waffle over spending $60 on something they will use literally every day.
Yes. Datomic Starter (nee. Datomic Pro Starter) now gives you one year of updates for free after which you need to pay to receive further updates. Previously you could renew your (free) Datomic Pro Starter license perpetually.
Datomic Free is unchanged and is still free and receives continuous updates.
There's been an open issue for years to add support for Google Datastore to bring Datomic Google Cloud support into parity with AWS.
But what constitutes a system? Will this end up mostly just being $5k per customer per year for cognitect?
https://www.reddit.com/r/Clojure/comments/5fc706/datomic_upd...
Throw out everything you ever thought about databases. It is from here you will learn a completely new way of dealing with data especially in environments where you need an immutable audit trail.
I almost feel like Datomic solves what blockchain does (immutability) really well without the drama.
Clojure was also another culture shock. I like it personally but the syntax at first was scary (but it's really simple and straightforward).
Having said that I'm not 100% sure whether I realized any productivity boost working with clojure. Finding clojure developers is tough & expensive vs Java or .NET or PHP.
Datomic has a bit of a learning curve and I still haven't fully crossed over, relying on google groups discussions and it's not a user friendly experience which I feel may be their limiting factor (they've made it far too difficult for the wandering developer to approach it) and the fact that it's closed source and proprietary but this is most likely because a lone developer hobbyist has no business using Datomic because it really makes sense in industries where you need a fully auditable trail, something that was typically available in event sourcing frameworks with some sort of CQRS wizardry that I've had trouble justifying using.
All in all, unlimited peer is a welcome news. However, datomic in all in it's shape or form, only applies to a small group of developers it seems, which is a shame because datalog queries blows SQL out of the water.
The stuff one can do with datalog on datomic tightly integrated with clojure, it's game changing, but won't apply to all use cases for folks on HN.
- Declarative and pattern matching make it elegant to read.
E.g., find me the ids of all customers invited by a customer named "Joe":
{:find [?customer]
:where [
[?inviter :customer/name "Joe"]
[?customer :customer/invited-by ?inviter]]}
- Allows recursive queries.
- You can reference metadata associated to any transaction that ever touched any attribute in your database within the same query constructs.
E.g, find me the time customer name was last modified:
{:find [?customer ?name ?last-modified]
:where [
[?customer :customer/name ?name ?tx]
[?tx :db/txInstant ?last-modified]]}
E.g, find me the name every customer ever had, w/ date of modification
{:find [?customer ?name ?last-modified]
:where [
[?customer :customer/name ?name ?tx true]
[?tx :db/txInstant ?last-modified]]}
- You can easily execute a query against how your database looked like months ago.
Same thing: would look like the query above, but executed using a different API call (http://docs.datomic.com/clojure/#datomic.api/as-of)
I would say these are the main ones, but there's more.
It's a pity. I'd be very happy to use Neo4J if it had.
Datomic focuses on modeling facts (entity-attribute-value). This is a very stable data representation that allows modelling tables, documents or graphs [1], on top of carrying transaction explicitly on an additional index for the time-travel magic.
I guess Neo4j might make a good DB for ETL-ing into and running certain kinds of analysis.
[1] https://hashrocket.com/blog/posts/using-datomic-as-a-graph-d...
Disclaimer: 12 years ruby/rails experience 3-4 years clojure experience
Well said. Unless you need to solve the Byzantine general problem there are better ways to database audit trails than Blockchain. e.g. Datomic or time travel queries.
and here's the link to byzantine generals' problem
https://en.wikipedia.org/wiki/Byzantine_fault_tolerance#The_...