This page is not created by, affiliated with, or supported by Slack Technologies, Inc.
2018-06-18
Channels
- # aws-lambda (3)
- # beginners (57)
- # boot (3)
- # bristol-clojurians (1)
- # cider (38)
- # cljs-dev (23)
- # clojure (35)
- # clojure-italy (32)
- # clojure-nl (6)
- # clojure-spec (35)
- # clojure-uk (132)
- # clojurescript (34)
- # cursive (22)
- # data-science (1)
- # datomic (54)
- # devcards (1)
- # duct (5)
- # editors (7)
- # euroclojure (4)
- # fulcro (40)
- # graphql (1)
- # hoplon (6)
- # immutant (5)
- # jobs (1)
- # off-topic (22)
- # om (1)
- # planck (17)
- # portkey (1)
- # protorepl (12)
- # re-frame (97)
- # reagent (67)
- # reitit (16)
- # ring-swagger (1)
- # shadow-cljs (98)
- # spacemacs (8)
- # sql (20)
- # tools-deps (60)
Bore da
subhothuyum
we have a papers we love meet up in belfast, I joined it last week. Thanks to the group, I am finally reading SICP
beautiful book. if they had taught that in college..
I see people try to compare clojure with elixir
not sure why the talk is not about clojure vs python or typescript
surely the reason is that both python and typescript are not functional languages; are not expression-based; are not extensible; and so forth. plus typescript is, naturally, statically typed, whereas clojure is not. compared to clojure's statically typed peers (haskell, f#, ocaml), its type system is weaker also
somebody asked should they use clojure in the clojure channel.
not sure what answer than yes they were expecting in this community.
@manas.marthi well... the answer is of course.... it depends π
that's what I said, but upon insisting, I said use clojure
such as?
btw I think there is a real-time JVM from Oracle (IBM used to have one, not sure if they do anymore though). But I don't know if how well they work
Just back from paternity leave to see that Iβve missed a thread on offal-based dad jokes!
But everybody likes a good pun, no?
Coming up with good puns entrails a lot of work.
If at first you donβt succeed, tripe, tripe again.
Mind you, I told a group of my co-workers almost a dozen puns the other day thinking that at least one would make them laugh, but no pun in ten did.
i haven't got the heart to vent any more spleen over these tongue-in-cheek puns, i think i should just hoof it on out of this thread
Some of the time, yes.
In this case, sheβs a very chilled baby - which is exactly what we were hoping for.
That always seems far more important than what everyone else wants to know - is it a boy or a girl?
@peterwestmacott I thought you could get up to 2 weeks in the UK
London Clojurian event this evening - very interesting talks on introducing Clojure for work as well as using Clojure for managing important life events. https://www.meetup.com/London-Clojurians/events/rpbqcpyxjbhb/
@jr0cket will the meet up be streamed
@manas.marthi No live streaming of talks, but they will be recorded and should be available in less that a weeks time. You can also find previously recorded talks in our https://skillsmatter.com/groups/87-london-clojure-community#past_events
Noted. thank you!
congratulations @peterwestmacott and welcome to the sleep-deprivation appreciation society
Yeah, almost 6 years now.
i'm just out the other side, and now i can use my sleep-deprivation induced superpowers for good
yeah - you should get 2 weeks paternity leave, then you can split the maternity leave with your partner (if they are getting maternity leave!) - but itβs paid at the lovely (low) mandated maternity leave rate. Unless you have an employer that boosts it above the minimum.
If we have Karsten Schmidt as a speaker at one of our events in London, what would you like them to talk about? Here is there twitter if you are not familiar with them: https://twitter.com/toxi
@jr0cket - I would love to hear Karsten on: 1) data visualisation in ClojureScript 2) Data Science in Clojure 3) 3D rendering in Clojure / ClojureScript or pretty much anything that he wants to talk about as long as he'll do a break-out session on th.ing at some point during or before the Conference π
(not announced pricing this year, just a date to save, but I am negotiating with my COO...)
@mccraigmccraig - What is freaking you out..?
this freakout was about decommissioning a bunch of nodes from our live production cassandra cluster
Oh, yeah, ok, I can see that being a nail-biter, no matter how much preparation and precaution was involved...
yeah, all went totally fine... but given the potential for knackering everything, i don't think it could ever be anything less than terrifying
@mccraigmccraig just never turn anything off... π
@mccraigmccraig - Well I am glad it went well, I suppose I now know what I am not going to do, and that is build a Cassandra cluster... π
@thomas not turning off a bunch of m4.2xlarge
s does get a bit expensive!
@maleghast well, on the upside it is quite possible to reconfigure the cluster, add nodes, remove nodes, restart everything, all with zero downtime
yep. ops is terrifying!
I like RDS (I realise that there are downsides to RDBMS, but having the service-provision be someone else's problem is compelling)
if there was a cassandra RDS i would probably use it... i've used RDS for postgresql before and been very happy with it
yep - a combination of [1] an excellent resilience story and [2] that cassandra forces you to not do things which don't scale well, by virtue of not implementing any db features which don't scale well
I had a client that wanted Cassandra on a project from day 1 - I tried strongly to suggest they build iteration 1 with PostgreSQL, get it working, then see if they really needed cassandra after that. No luck. (I rolled off before the cassandra implementation was working, so who knows - maybe it was fine? π€· )
c* is just a different way of thinking really. Once you've had your head broken by it I'm not sure it is any trickier (esp if you think in terms of events or time series)
c* = cassandra?
yep, it's just different. you get some significant restrictions and some amazing superpowers in return, and it takes a while to grok what the differences are and what the consequences of the differences are
OK, I will have to look into it further to see if the "superpowers" are stuff I want and that the restrictions are liveable in return.
my problem with postgres (which I think is great btw) is that people treat it like mutable state and don't think about events in their system, but then I tend to build things from events that exist or create events b/c I want auditability and other derived bits of data
@otfrom - I can see that... My general feeling is that I am going to use Datomic for the "domain" data in our system, so that I get auditability / immutable history from that, but I am putting a lot of value-data (i.e. data that will NEVER have an UPDATE query run against it) into PostgreSQL for ease-of-use and because the Data Science team (such as it is at the moment) are all R-speakers who go funny colours when I talk about Clojure.
So bulk data that they only read goes in PostgreSQL 'cos they understand how to read it from there.
@maleghast that makes sense given your users
I am, as yet, unsure how to handle their outputs so that there is an audit trail and no update-in-place shenanigans, but I have ideas...
repeatable data science is hard w/people who create production code for a living. It is well nigh impossible with people who hack out R scripts (R is awesome. Lots of R practice isn't)
Yeah... I am pleased to say that the Head of the department would agree with you and wants to limit R use to exploration and prototyping. He wants to make it a core practice of the team that production code is implemented in Python, under much more rigorous coding / programming approaches.
Anyway, my intention in the first instance is to just insist that every time they run a model the outputs have to be written to a new record, either in a bulk / de-normalised table, so that it is easily converted into an immutable data structure and there is no update-in-place only writing new rows, or as a new JSONB document using PostgreSQL's JSON / Document handling features. As long as the practice is drummed into them that they only ever write new records, and that previous results are effectively permanent I reckon I have a shot at keeping things sane in terms of managing the data we generate as well as the data we take in.
a function that take the current value and then 1-n others, but often just the current value and a new value.
I think I mean more in discussion w/other coders rather than in the code. Tho it would be nice to have them be both the same.
Anyone watching the match? π
I guess that's a yes @seancorfield
Bit surprised at the lack of responses there π since it was on at a reasonable UK time...
...and I figured there must be some footie fans here? β½