robert-stuttaford: Hmm. Don't have pro at the moment.

It's cool. I have a working solution. Was just wondering if there was a neat way to do it in a datalog query.

robert-stuttaford: Thanks anyway

cool

i’m sure that Rich or Stu or one of the other cognitects would have an answer for you pretty quickly. the trick is to put it in front of them

Can someone invite them here? 😉

they’re busy people. they probably won’t show up in here

@martintrojer: about your do’s and don’ts article: use dynamo. which other backends have you tried that failed?

we tried PSQL (with many peer nodes) obviously doesn’t scale like dynamo

wanted to make that clear to people trying to figure out datomic

i see

we’re intending to use cassandra since we’re running on google cloud

any SQL store would show similar characteristics

Yeah, we wanted to limit our infrastructure so introducing cassandra / riak would blow our devops budget

the big string thing, ran into that too. indexing would take forever after storing enough raw xml in datomic

how big were your strings, stijn?

yeah, I have a feeling people are putting json blobs into datomic without knowing the limitations.

ranging from a few KB to a couple of MB

we’re putting them in cassandra now, and store the key in datomic

you lose the ‘change detection’ of datomic though

still kind of off-putting if you ask me. It’s very nice to dump big documents into PSQL JSON columns (and to be able to query into them)

datomic needs to feature

also, the for love of god give us query planning.

i read somewhere it’s on the todo list 😉

the blob thing

If I read this example properly (don't have a few minutes to test) Datomic supports unbounded recursion: https://gist.github.com/stuarthalloway/2002582

nice tjg! @srihari , see tjg’s link

Though I hear datalog always terminates on finite data... If it supports sufficiently powerful recursion and always terminates... then I'd think something's gotta give.

tjg, robert-stuttaford: thanks! Will take a look.

@martintrojer: I've thought about writing a query planner for datomic a few times. A trivial one ain't even that hard (but obviously you can ramp up how good it is a whole bunch)

what would a planner actually do?

-curious-

determine the best order for the clauses on the fly?

that would be pretty awesome

yup

some orders will always be the same - e.g. when working from :in values to :find values

but sometimes changing the order of clauses in the middle makes a huge difference

using the AWS transactor appliance, what is the recommended way to do backups?

by default you can’t SSH in, or put files on the box, so do you modify the AMI, or something else?