Clojurians Log v2

Bore da

Whole table on non-stop train to myself so laptop out, spread out and 2.5 hours to catch up on Clojure/Clojurescript/Machine learning related stuff!

nice

Morning. I love working on trains, I find it really productive

When you have room to work…absolutely

Morning ldn-Clojurians....

Morning my melty friends

In work now…reprogramming brain from small, discrete compossible functions to Java 1.5 imperative programming with as much logic crammed into one method/class as possible!

@agile_geek: i'm just gonna leave this here https://github.com/krukow/clj-ds

@mccraigmccraig I'll take a look for my own education and edification but for reasons why it's not something I could use in work see my #C04V12NPC thread!

@agile_geek: java 1.5 ! wow 😞

@agile_geek: do you have to enter your java 1.5 on punch cards?

@mccraigmccraig these people may be looking after your money for u! ;-)

Mind u I've said before code of mine written in COBOL over 25 years ago still runs in production

@thomas u know what...I might be happier if I did!

Bon dia

I love, and hate, coding on the train. Love it due to lack of distractions. Hate it due to lack of internets.

I used to have that problem but now I find a tethered 4G phone retains connectivity 80% of the journey.

Just wrote up http://blog.juxt.pro/posts/why-clojurescript-matters.html

@jonpither: well spoken sir.

@jonpither: bookmarks for reading later

@jonpither: we just had a conversation along these lines in the morning with some colleges. nice summary of my overall impression of euroclj as well in a way 😉

Was a great conference wasn't it

BTW @benedek Someone should do a talk soon or a blog on the state of clojure.mode, cljs-refactor and CIDER, and how they will evolve in tandem

it's rather fascinating

blog/talk: very true. i submitted an abstract but it got rejected (did not include cider tho — but bozhidar submitted one too afaik). only the one big missing piece in refactor-nrepl is cljs support

vanilla elisp clj-refactor stuff supposed to work for cljs too tho

like threading macro stuff and the like. even remove requires hopefully 😉

and perhaps we need to see the first few steps in merging cider and refactor

btw any opinions on this ladies and gents here?

would you boo or clap your hands on a mega clojure support bundle for emacs (and other editors perhaps…)?

so all the cider and refactor stuff in one package...

@jonpither: Thanks for the article. Sending it out to our team. :simple_smile: Sorry we didn’t cross paths while I was in London… Hakan had told me he wanted to introduce us but I guess the stars weren’t aligned. Hope you had fun at the conf.

Hi @steven. Next time defo. We did have a beer at Strange loop and I have photographic evidence

your company in India sounds cool - a Clojure consultancy of sorts..?

@benedek: FWIW, and I'm not nearly clued up enough - I'm conflicted. On the one hand I like the idea of clj-refactor folding into clojure-mode and CIDER. But then I like the free reign clj-refactor has as a new project, and the innovation it brings

yeah absolutely. you need to be more conservative with cider (for example with dependencies)...

Are there videos out for the conf yet?

https://vimeo.com/euroclojure/videos doesn't look like it 😞

they'll be on clojure tv in a few weeks

(at least that's what they told speakers)

@benedek: you helped solve the dependency problem with mranderson. I think it's more that with clj-refactor you can release new features that you're not sure will totally be awesome (i.e. experiment, and see what takes). They cannot do that with CIDER, no way, because it has more users, and people expect the feature set to be extremely well honed

true. dependencies: this time i meant elisp deps 😉

Also the tech debt and complexity of CIDER is greater, so they need to more careful with what they add

yeah, will be really nice move away from nrepl as well hopefully

(great article @jonpither !)

for those of you that couldn't make it to euroclojure, I hosted my own talk video because I hate waiting for conf videos: http://yellerapp.com/posts/2015-06-29-performance-and-lies.html

@tcrayford: awesome talk. Quite scary some of it, as I know the odd dev who advocated doing a lein run to fire up the app in prod

@jonpither is talking about me

@malcolmsparks: 😁

thanks @tcrayford :simple_smile:

I missed Tom's talk because of booth duty, but will definitely watch it on youtube when it's put up

@malcolmsparks: I already uploaded my own video 😉

http://yellerapp.com/posts/2015-06-29-performance-and-lies.html

nice one

sorry for the spam, thought folk in here might want to watch it

definitely

@jonpither: I mean using lein run in prod is ok if you turn on the "go faster jvm" settings 😉

go faster stripes?

I need to know those...

well there are 830+ options, many of them are go faster stripes 😉

just a lot of them are "this might make you go faster, but you'll want to tune it to a good number, and that depends on what code you're running, what machine you're on, what JVM you're running etc"

@tcrayford: random consultancy Q - on a previous project, clojure.set/subset? was a perf hotspot of comparing sets of keywords. At some point do you gain value using funky bitmap approaches over sets and keywords?

@jonpither: for sure you can. Depends on the app

(yeller uses some bitmaps internally in a few places and it couldn't do some features without them)

Cool

@tcrayford: the +830 options (I heard 866?) is one of the reasons why a lot of people with in depth knowledge of the JVM are saying Oracles plans to default G1 GC is not wise. Discussion centres around the G1 GC not being optimal for a lot of ‘defaulted’ apps. Oracles argument is anyone really concerned about performance will have used some of those switches.. really?

@agile_geek: I think g1gc by default does kinda makes sense though - most apps don't care so much about throughput of the GC these days, but they do care about max pause times a lot more imo

it is app specific though, which sucks 😞

great post @jonpither !

@tcrayford: not sure all legacy apps don’t care about throughput?

yeah, depends

but lots of apps really don't wanna have 1 minute+ GC times or whatever

hadoop obviously doesn't care though

True

Discussions I’ve read have been around it being premature for Oracle to do this without further consideration of the impact.

yeah, makes sense

oracle do work really hard on that stuff though, and I trust them pretty well these days

You have more faith in them than I do! Maybe I’ve been burnt by their DBMS division far too often!

hah. I think the jvm perf team has some smart folk on it, and the language design team is doing well despite a near-impossible challenge. Never talked to the DBMS team though 😉