datomic has been announced
261 views
Skip to first unread message
kovas boguta
Mar 5, 2012, 1:46:12 PM3/5/12
to clo...@googlegroups.com
Since not everyone reads twitter or hacker news, http://datomic.com/
has been updated with an unveiling of Rich's new project.
has been updated with an unveiling of Rich's new project.
Linus Ericsson
Mar 5, 2012, 6:08:53 PM3/5/12
to clo...@googlegroups.com
Mindblowing! I can imagine many applications for datomic, it looks like a very powerful abstraction which is overcoming many of the problems I've been encountering when I've built applications. Thanks, Rich!
Chris Grangers awesome experimental UI for developing javascript games is covered in Wired's blog Webmonkey: http://www.webmonkey.com/ (I found it browsing the http://Wired.com!) Congratulations!
/Linus
Since not everyone reads twitter or hacker news, http://datomic.com/
has been updated with an unveiling of Rich's new project.
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clo...@googlegroups.com
Note that posts from new members are moderated - please be patient with your first post.
To unsubscribe from this group, send email to
clojure+u...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
Hank
Mar 5, 2012, 9:01:17 PM3/5/12
to clo...@googlegroups.com
From watching the presentation, it seems to include (but not be limited to!) the some of the good parts of Prevayler and CouchDB, so anyone familiar with those will find the Datomic concept familiar, too.
Prevayler (www.prevayler.org) was announced on Slashdot in 2003 (!) and they basically made the point that RAM had gotten cheap enough to keep most "working data" in RAM that you could have your data just be POJOs and run your queries on in RAM using some in-language query mechanism that runs on the object graph. Near the start of the above presentation Rich talks about "hardware and architectures have changed substantially" which is what this is talking about. BTW, even back then before "NoSQL" became a meme this unmasked some of the SQL fanbois who saw their consulting revenues in jeopardy. :)
Prevayler asks you to express all your changes to the object graph in transactions reified as Java objects, there called 'commands' from what I remember, and pass them through this bottleneck that serializes them to disk. This is similar to the "transactor" that Datomic seems to have. On the disk side of things, Prevayler would use a "logging and snapshot" combination that was a known design pattern before (http://www.ibm.com/developerworks/library/wa-objprev/) but popularized by Prevayler.
CouchDB exists since 2005 or so and offers the following concepts similar to Datomic: Replication across peers, except CouchDB also persists to disk on peers. The "receive queue of transactions that can be filtered" and "reactive, no polling" application design that Datomic offers its peers is similar to the CouchDB _changes stream of updates. CouchDB can filter _changes server-side to reduce load on the network, not sure Datomic can do that. CouchDB also implements the "add facts, don't change data in place" idea by keeping revisions of the data structures.
All in all very exciting to see this confluence of ideas.
-- hank
Prevayler (www.prevayler.org) was announced on Slashdot in 2003 (!) and they basically made the point that RAM had gotten cheap enough to keep most "working data" in RAM that you could have your data just be POJOs and run your queries on in RAM using some in-language query mechanism that runs on the object graph. Near the start of the above presentation Rich talks about "hardware and architectures have changed substantially" which is what this is talking about. BTW, even back then before "NoSQL" became a meme this unmasked some of the SQL fanbois who saw their consulting revenues in jeopardy. :)
Prevayler asks you to express all your changes to the object graph in transactions reified as Java objects, there called 'commands' from what I remember, and pass them through this bottleneck that serializes them to disk. This is similar to the "transactor" that Datomic seems to have. On the disk side of things, Prevayler would use a "logging and snapshot" combination that was a known design pattern before (http://www.ibm.com/developerworks/library/wa-objprev/) but popularized by Prevayler.
CouchDB exists since 2005 or so and offers the following concepts similar to Datomic: Replication across peers, except CouchDB also persists to disk on peers. The "receive queue of transactions that can be filtered" and "reactive, no polling" application design that Datomic offers its peers is similar to the CouchDB _changes stream of updates. CouchDB can filter _changes server-side to reduce load on the network, not sure Datomic can do that. CouchDB also implements the "add facts, don't change data in place" idea by keeping revisions of the data structures.
All in all very exciting to see this confluence of ideas.
-- hank
Reply all
Reply to author
Forward
0 new messages