Elixir and Phoenix_transcript.txt

[0.00 --> 15.08]  welcome back everyone this is the changelog and i'm your host adam stekowiak this is episode 147
[15.08 --> 20.96]  and today jared and i are talking to chris mccord talking about elixir on top of erlang
[20.96 --> 26.70]  phoenix the web framework definitely got me and jared thinking about concurrency and elixir for
[26.70 --> 32.26]  upcoming project phoenix sounds really cool and i love this conversation with chris we got some
[32.26 --> 38.64]  awesome sponsors code ship top towel and postico we'll tell you about top towel and postico later
[38.64 --> 43.16]  in the show but our friends at code ship have this awesome new feature that help you deploy the
[43.16 --> 48.44]  production so much faster than you're doing right now it's called parallel ci and if you want to run
[48.44 --> 53.80]  faster tests and have your builds built and in production quicker you've got to run your tests
[53.80 --> 59.94]  in parallel with parallel ci you can now split your test commands into 10 test pipelines or up to 10
[59.94 --> 64.76]  test pipelines this lets you run your test suite much faster than before and drastically reduce the
[64.76 --> 70.46]  time it takes for you to push to push your code to production and get your builds run they integrate
[70.46 --> 75.66]  with github and bitbucket you can deploy to cloud services like roku aws and many more and you can
[75.66 --> 80.38]  get started today by trying out their free plan which includes 100 builds a month and five private
[80.38 --> 86.62]  projects or you can use our offer code the changelaw podcast again that offer code is the changelaw podcast
[86.62 --> 93.38]  and with that code you'll get 20 off any plan you choose with code ship three months head to
[93.38 --> 97.98]  codeship.com slash the changelaw to get started and now on to the show
[97.98 --> 107.48]  all right everybody we're back we got chris mccord on the call today we're talking about some cool
[107.48 --> 113.78]  stuff today elixir the language uh phoenix the framework chris welcome to the show how's it going
[113.78 --> 119.34]  and we also got jared santon on jared how are you today man doing better better than i was yesterday
[119.34 --> 123.74]  yeah yesterday was a bad day that was a bad day bad day would you did you sleep a lot of a lot of
[123.74 --> 130.02]  time i was trying to sleep mostly tossing and turning you know but uh all good today feeling better
[130.02 --> 140.12]  did slack keep you up and whatnot ah everything was keeping me up it was a bad day so chris you hail
[140.12 --> 147.42]  from dayton ohio right that's right not phoenix arizona no not yet we established in the pre-show
[147.42 --> 153.10]  that chris would be slightly cooler well at least i did if you were from phoenix arizona because of
[153.10 --> 159.24]  phoenix the the framework it's just it would make sense you know maybe he had named the framework
[159.24 --> 166.12]  dayton framework see come on chris the logo wouldn't have been nearly as cool though so
[166.12 --> 171.00]  i don't know i mean you got some you got the date what the daytona 500 up there right
[171.00 --> 177.00]  no that's that's that's wrong state is that a different state that's uh florida so that's right
[177.00 --> 183.22]  florida florida what's in date well you know isn't there something racing there no we we like where
[183.22 --> 187.58]  the wright brothers were from so we were first in flight except they didn't they built it here but
[187.58 --> 193.64]  they flew it in north carolina so both states claim that we're both first in flight so that's
[193.64 --> 200.12]  probably our only claim to fame that's it well you got some other claims of fame here so let's let's
[200.12 --> 204.78]  dig deep here so you know for those who know who do not know who you are how do you introduce yourself
[204.78 --> 211.42]  to an audience like we got today um i created the phoenix framework uh it's an elixir web framework
[211.42 --> 219.26]  and i also i just recently uh authored uh metaprogramming elixir for pragprog and uh during
[219.26 --> 225.52]  the day i build web applications at little lines little lines i like that you know that you got a
[225.52 --> 233.20]  little shout out too in that uh in your talk recently it was elixir com yep yeah um was there
[233.20 --> 237.22]  people in the crowd that knew who little lines was i think we got yeah we had like that one little
[237.22 --> 241.08]  whoop a small following i don't know i don't know who that was we asked someone did
[241.08 --> 248.82]  so yes jared's singing in the chat room behind the scenes daytona 500 classic yeah that's that's just
[248.82 --> 253.08]  i i confuse it all the time with dayton and i'm sorry that's classic
[253.08 --> 259.34]  we get that a lot we get that a lot yeah you google date and you get daytona 500 every time
[259.34 --> 264.24]  every time all right well chris we've been excited to get you on the show we've had uh
[264.24 --> 270.26]  multiple requests to have shows on elixir uh show on phoenix we have to thank a few of our
[270.26 --> 275.40]  listeners alexander quine was the one who originally said hey let's let's talk about elixir on the show
[275.40 --> 282.80]  and then following up to that was uh sebastian cizduro cizduro sorry sebastian if i butcher that name
[282.80 --> 290.02]  cizduro um who said uh yes let's talk about elixir and specifically let's get chris on the show to
[290.02 --> 294.82]  talk about phoenix so thanks to those guys for requesting this and we're happy to have you here
[294.82 --> 301.02]  to talk about what did they request that at uh they request that on our ping repo which is github.com
[301.02 --> 306.32]  slash the changelog slash ping yep and i think they reached out to me someone reached out to me on irc
[306.32 --> 311.82]  elixir link and then that's right and opened up that issue made it happen i love that man
[311.82 --> 315.42]  elixir advocates out there apparently yeah and little lines too
[315.42 --> 323.88]  yep yeah so elixir is an interesting young language um seems to be growing quite the following
[323.88 --> 330.28]  and seems like it's got your uh it's caught your fancy and so why don't you just start off we'll
[330.28 --> 334.24]  talk about elixir and then we'll get into phoenix later but i'd like to hear about elixir from your
[334.24 --> 340.06]  perspective chris as one who is now developing lots of stuff uh in the language sure so yeah so
[340.06 --> 346.28]  you mentioned elixir is a young language and that's true uh but we are at a 1.0 uh stable release
[346.28 --> 353.16]  we've been there since uh july last year and kind of the cool thing about elixir is it's built on top of
[353.16 --> 360.00]  uh the erlang virtual machine so we are very young language but we're using um erlang to kind of
[360.00 --> 365.34]  bootstrap the language itself so we didn't have to kind of reinvent the wheel for a lot of the
[365.34 --> 370.52]  underpinnings and that let elixir kind of get out of the gates quickly and then we also have you know
[370.52 --> 375.82]  20 plus years of innovation and libraries under the hood that we can take advantage of uh today
[375.82 --> 384.04]  so for those that don't know uh elixir compiles down to um erlang bytecode and it runs on the erlang
[384.04 --> 389.76]  virtual machine and um erlang was kind of an obscure language at least for me a few years ago i hadn't
[389.76 --> 394.92]  i had heard of it but i wasn't really aware of it um never heard about it in college erlang kind of
[394.92 --> 401.00]  had this nugget of innovation for the last 25 years is what i tell people they basically run half
[401.00 --> 408.18]  the world's telecommunication systems and um so it's been operating um at high scale and high
[408.18 --> 415.54]  reliability for decades and only now recently people are starting to take notice of our most of our modern
[415.54 --> 419.58]  languages aren't really great at concurrency and now they're kind of finding out that erlang solved
[419.58 --> 425.74]  these problems 20 years ago um so it's pretty cool that like you know you might say that other
[425.74 --> 432.00]  modern languages like go kind of you know tackle concurrency but erlang from the very beginning had
[432.00 --> 437.62]  three core values which i think most not any language that i'm aware of targets it was a language that
[437.62 --> 442.78]  needed to be highly concurrent um highly distributed because they wanted to run out on multiple telecom
[442.78 --> 447.40]  switches in the 80s and then a highly fault tolerant because you wanted these things to stay online
[447.40 --> 452.54]  forever and they kind of built the language around those needs and developed kind of the standard
[452.54 --> 457.72]  library around these specific problems and that those specific problems actually map perfectly onto
[457.72 --> 463.78]  kind of the multi-core age that we have today um so elixir builds on top of that uh erlang innovation
[463.78 --> 470.30]  adding um some modern features that you would expect with a language like uh metaprogramming was big
[470.30 --> 476.80]  for me it adds um polymorphism through protocols so it borrows some ideas from closure and some other
[476.80 --> 482.02]  places to really give you this modern language with all this great innovation for the last couple
[482.02 --> 488.44]  decades so did you had you previously tried erlang directly or did you come through it from the elixir
[488.44 --> 493.70]  side so i went straight to elixir but so what got me into it initially is i've been doing ruby
[493.70 --> 501.00]  development professionally for the last uh six years or so and i wanted to do uh web sockets in my rails app
[501.00 --> 506.68]  i was building some kind of real-time features into uh my rails app i built this gym called sync
[506.68 --> 512.10]  that let you let you build let you do like real-time rails partials and it worked well but i had to
[512.10 --> 517.58]  jump through kind of all these hoops to push out real-time events i couldn't do it in my rails app
[517.58 --> 521.98]  itself i had to kind of offload it to an event machine thread or i had to offload it to like a
[521.98 --> 527.42]  separate fae server and it was like all this messiness and i got into it it wasn't going to
[527.42 --> 532.60]  scale scale well at all and so then i started looking at like you know what are some other people
[532.60 --> 537.28]  doing what languages are they using and that's when i looked at erlang and i think that was when
[537.28 --> 543.06]  that was when whatsapp was not yet sold for 20 billion dollars but at the time um there was an article
[543.06 --> 548.66]  i read a couple years ago two or three years ago that was about they were using erlang and they were
[548.66 --> 553.72]  getting a million concurrent connections per server and that kind of blew my mind because i was looking
[553.72 --> 560.64]  at getting maybe you know 100 connections uh on my rails app um so that kind of spiked my interest
[560.64 --> 566.40]  into erlang and then i remembered elixir that i had just kind of come across um so i kind of checked
[566.40 --> 572.38]  elixir out and then that kind of got the ball rolling on on everything so you came at it from
[572.38 --> 577.00]  the ruby perspective which it seems like it's fitting because elixir also has a kind of a ruby
[577.00 --> 581.74]  pedigree is that right yeah so we'll see we see a lot of people coming over from ruby i think you
[581.74 --> 588.06]  know a large part of that is uh jose valim the creator of elixir um was a rails core team member and
[588.06 --> 592.24]  he's written probably if you're doing rails development you use half of the gems he's ever
[592.24 --> 597.86]  created uh so i think his proximity to the ruby community has brought a lot of people over and then
[597.86 --> 603.20]  also yeah at a glance uh the syntax does look familiar i would say that that's kind of like a
[603.20 --> 607.42]  veneer because once you get into it the semantics are very different but at least at a glance you're
[607.42 --> 612.40]  like hey this looks kind of similar so it's kind of like a double double whammy to for rubius to
[612.40 --> 618.38]  kind of jump in awesome so um tell us a little bit more about the language itself so it builds on
[618.38 --> 623.76]  top of erlang it has some meta programming is it a functional language is an object oriented language
[623.76 --> 628.50]  give us some of the the aspects sure yeah so it's a it's a functional language and uh it's immutable
[628.50 --> 635.50]  and uh it interops it has full interoperability with erlang so you can call erlang from elixir and
[635.50 --> 641.48]  you can call elixir from erlang so any uh off-the-shelf erlang library um you can just drop in
[641.48 --> 647.22]  and it just works and that's what kind of let us um as a community move forward and get a release
[647.22 --> 651.90]  out soon because we didn't have to kind of you know ream at the wheel for everything um but it
[651.90 --> 659.46]  adds and it solves some pain points um that people have had with erlang uh one is at least for me some
[659.46 --> 663.72]  people really dislike meta programming but for me coming from like a ruby background i i love
[663.72 --> 670.32]  meta programming kind of lets me um distill my domain problems and write beautiful code uh so it adds
[670.32 --> 675.06]  meta programming which was kind of a big feature for me and then it adds um like i said polymorphism
[675.06 --> 681.58]  it borrowed the idea from closure where you can get um polymorphism but without object orientation so
[681.58 --> 686.04]  you can go to functional programming you don't have to throw away all this um all these ideas of
[686.04 --> 691.64]  polymorphism which are good and you can kind of get that but in a functional paradigm so for me that's
[691.64 --> 695.88]  been really nice and then it has really great like just simple things that you expect in a modern
[695.88 --> 700.74]  language like really great unicode string handling which has been historically problematic in erlang
[700.74 --> 707.70]  and it really focuses on great experiences like developer experience so it has a build tool that
[707.70 --> 713.76]  it ships with a project generator so you can just say uh mix new give it a project name and it will
[713.76 --> 719.62]  generate you know idiomatic project structure for you it ships with a test unit framework i can just say
[719.62 --> 724.80]  mix test to run my tests and then it also comes with a package manager and kind of all these things
[724.80 --> 731.34]  were uh missing or not missing they were not uh shipping with the base erling installation and it
[731.34 --> 736.72]  kind of erling didn't focus on getting up and running quickly because historically they've been solving i'd
[736.72 --> 743.60]  say like grittier problems uh so it came in i think solved some some filled in some big gaps and i think you
[743.60 --> 750.52]  know jose uh said it best that when he first looked into erling he loved everything he saw but he hated the
[750.52 --> 755.42]  things that he didn't see and that's what kind of elixir came about is filling in those gaps and still
[755.42 --> 761.22]  building off all the things that he loved that's interesting i have a little bit of a history with
[761.22 --> 767.42]  erling like a few days like a like a weekend with erling where i thought i'm gonna kind of
[767.42 --> 773.44]  bend my mind a little bit and and the syntax itself it was for me difficult to stretch my mind around
[773.44 --> 777.84]  started to get to it by the end of the weekend but didn't have a real real world use case for it back
[777.84 --> 784.32]  then so it didn't stick what's the syntax of uh of elixir i know it's kind of difficult to describe
[784.32 --> 788.88]  code in words but if you had to compare it to a language or to a few languages what would you say
[788.88 --> 795.16]  that it feels like writing well so i would say the syntax is beautiful and uh it it definitely uh
[795.16 --> 800.98]  feels uh it definitely feels like ruby to some degree but it has even it has even more of a natural
[800.98 --> 808.20]  syntax than ruby and um it has some neat it stole we stole the idea of the pipeline operator if you're
[808.20 --> 814.94]  familiar um closure has like threading macros and f sharp has the pipe operator um so when you write
[814.94 --> 820.04]  elixir you structure your code i'd say very differently than other functional programming languages that i've
[820.04 --> 826.22]  seen just by virtue of having the pipeline operator allows you to kind of flatten out a list of nested
[826.22 --> 832.10]  function calls so it pipes the argument of the result of one operation into the first argument
[832.10 --> 836.14]  of another function so instead of like kind of trying to read your code backwards in a bunch of
[836.14 --> 842.34]  parentheses it kind of flattens that out and just compiles down to the same nested function call um so
[842.34 --> 846.58]  it kind of it made when i first got into it made me feel like when i was writing ruby of it just it
[846.58 --> 852.90]  feels very natural and it's really pleasant to write um but even more which is surprising for me to say
[852.90 --> 858.10]  so like it gives you things like pattern matching which um anytime i go back to my ruby code i lack
[858.10 --> 863.44]  these features and it makes me kind of sad so it's like the first time i had language envy coming from
[863.44 --> 867.36]  you know a ruby mindset of you know code should be beautiful it should be a pleasant you know
[867.36 --> 874.14]  experience to write can you explain pattern matching a little bit question too yeah so explain pattern
[874.14 --> 882.46]  matching in words so i can write multiple uh functions with the same name and i can um
[882.46 --> 889.30]  give the value of an argument and that function will only be invoked if that pattern matches so if i
[889.30 --> 896.44]  wrote like a countdown function i can pattern match i could recursively call countdown and pattern match
[896.44 --> 901.88]  on zero and that would be like the case where i'm done counting down and then i could just pattern
[901.88 --> 908.14]  match on a variable like number and then recurse on countdown minus one that makes any sense that's a
[908.14 --> 914.88]  very simple example but the virtual machine lets you kind of uh destructure um any data structure so i can
[914.88 --> 920.72]  say i have a list i want to i could pattern match out the first three elements of that list and then get the
[920.72 --> 924.78]  head of or get the tail of the list like all the remaining elements and i would have those all in separate
[924.78 --> 933.32]  variables so it lets you kind of destructure things really naturally so does the arguments to the method or to the
[933.32 --> 939.02]  function they become part of the the method signature yes exactly so you'd have multiple ones of the same
[939.02 --> 944.50]  name and as long as the arguments vary they would call different methods yeah exactly fair to say yep
[944.50 --> 950.24]  okay and that's how you'll see interesting you'll see that's kind of like idiomatic code so you elixir has
[950.24 --> 955.98]  like standard branch like if statements unless or if expressions unless expressions but you use them a lot
[955.98 --> 961.28]  less often because you have pattern matching that's kind of one of the staples of functional programming
[961.28 --> 966.36]  correct you're going to find that in enclosure you're going to find that in you know lisps right
[966.36 --> 973.84]  yeah i think you'll find that in most functional areas and to me it's it's a it's huge i mean once
[973.84 --> 979.40]  you experience it it makes writing any code without it for me a kind of a painful process yeah it seems
[979.40 --> 983.98]  like it helped melt away conditionals that would otherwise be checking these things and then you know
[983.98 --> 989.02]  you know branching depending on the arguments um you could just melt those away with different
[989.02 --> 994.24]  methods yep so you'll use if occasionally but if i if i'm using if a couple times in a function
[994.24 --> 998.04]  i start sweating a little bit thinking like you know something here is not quite right
[998.04 --> 1006.64]  interesting so you're building a web framework elixir comes from a rails core team member
[1006.64 --> 1011.68]  is elixir built for the web or is it a more of a general purpose programming language
[1011.68 --> 1016.40]  it's built i would say it wasn't i mean jose comes from a web background
[1016.40 --> 1023.12]  but the neat thing about erlang is you can target um kind of like the embedded space
[1023.12 --> 1028.72]  or the web so it's kind of i would say a general purpose in that you can go both high level and
[1028.72 --> 1033.08]  low level with it because it came the erlang comes from you know running on telecom switches
[1033.08 --> 1039.56]  so people are using it um in the embedded space um pretty successfully today but then it's also
[1039.56 --> 1045.00]  great for you know higher level building any kind of thing that you would run you would consider a
[1045.00 --> 1050.64]  server uh so kind of web applications come into that naturally um you look at like whatsapp they're
[1050.64 --> 1055.90]  running the their entire operation on it um i think they were i mentioned a million connections
[1055.90 --> 1059.92]  per server but now i heard they're up to two million connections per server and i think they hit three
[1059.92 --> 1066.80]  million during a spike once and they're running 400 plus million users all on erlang and that you
[1066.80 --> 1071.58]  know that's yeah they had something like 30 engineers too or yes exactly so ridiculously small
[1071.58 --> 1076.66]  amount of engineers for how many users there was 400 million users and 30 engineers to kind of prove
[1076.66 --> 1083.24]  how robust erlang is and elixir builds on top of that so the standard library um kind of gives you
[1083.24 --> 1089.42]  these um tried and true mechanisms for building out distributed fault tolerant applications and um
[1089.42 --> 1095.22]  things uh intended just work cool so i saw your video on elixir conf which we'll link up to in the
[1095.22 --> 1099.82]  show notes and one thing you said is that you're you'd love if all of your consulting work
[1099.82 --> 1106.46]  not just your your you know personal projects but all your consulting projects hopefully eventually
[1106.46 --> 1113.42]  at some point will be written in elixir if you had to describe it you know in one in one phrase what
[1113.42 --> 1117.02]  is it about elixir that has you so excited you got to distill it down what would you say
[1117.02 --> 1126.74]  uh so i will two phrases um so one it lets me i i can write highly performant code without sacrificing
[1126.74 --> 1133.22]  productivity and i can be highly productive without sacrificing performance and it kind of a play on
[1133.22 --> 1138.54]  words but it's kind of the holy grail like i come from you know writing a lot of rails applications
[1138.54 --> 1145.70]  where i'm sacrificing that concurrency and that performance but the um productivity and the feeling
[1145.70 --> 1151.70]  of building these things is so great um elixir gives me both of those performance and productivity
[1151.70 --> 1159.38]  any drawbacks or things you've run into that have been not so great uh so because elixir is so
[1159.38 --> 1165.78]  young um you're definitely going to have access to a lot less off-the-shelf uh libraries so if you can't
[1165.78 --> 1172.10]  just do gem install device and have user authentication yet um so i'd say that lack of off-the-shelf
[1172.10 --> 1180.50]  packages and also the the learning curve is pretty steep compared to if you're coming from any kind of object-oriented
[1180.50 --> 1186.74]  background and you want to get into you know ruby or get into python you're it's pretty straightforward
[1186.74 --> 1192.58]  because you know an object's an object um but coming into functional programming at least for me elixir
[1192.58 --> 1198.18]  was my first functional language the learning curve initially i call it like the frustration gap where
[1198.18 --> 1202.74]  you're like you feel dumb you're not getting anything done something that should looks like it should be
[1202.74 --> 1206.90]  simple you can't figure out um it takes a little bit to get past that for things to actually like click in
[1206.90 --> 1212.98]  your mind thinking about programming differently awesome well we're going to take a break now to
[1212.98 --> 1218.02]  hear a word from a show sponsor but we'll be right back and when we come back we'll talk about phoenix
[1218.02 --> 1225.06]  the web framework top towel is the best place to work as a freelance software developer if you're
[1225.06 --> 1230.02]  freelancing right now as a software developer and you're looking for a way to work with top clients
[1230.02 --> 1236.02]  on projects that are interesting challenging and using the technologies you want to use top towel might just be
[1236.02 --> 1241.22]  the place for you working as a freelance software developer with top towel means that your days of searching
[1241.22 --> 1247.14]  for long-term high quality work and getting paid what you're worth will be over let's face it you're an awesome
[1247.14 --> 1251.94]  developer and you deserve to be composite like one joining top top means you'll have the opportunity to travel
[1251.94 --> 1258.82]  the world as an elite engineer on top of that top towel can help provide the software hardware and support you need
[1258.82 --> 1266.98]  to work effectively no matter where you are in the world head to top towel.com slash developers that's t-o-p-t-a-l.com
[1266.98 --> 1270.18]  slash developers to learn more and tell them the changelol sent you
[1272.34 --> 1278.34]  all right we are back with chris mccord chris you fell in love with elixir you've been doing a lot of rails
[1279.14 --> 1286.74]  um you decided there needs to be a rails for elixir that kind of the thought process there so uh
[1286.74 --> 1292.98]  yes and no so i would say we we borrow um we definitely borrow some ideas from rails um but
[1292.98 --> 1301.22]  we're not trying to recreate rails and elixir but i will say um we are trying to borrow a lot of the
[1301.22 --> 1307.46]  um a lot of the spirits of rails so like what rail showed is if you have a community get together around
[1307.46 --> 1315.62]  like common conventions you can group build great tooling uh so i would say i would caution to say that
[1315.62 --> 1322.02]  we are building we aren't building rails for elixir but in some ways we are um we want to build a
[1322.02 --> 1326.98]  full-featured framework that the community can get together and build great tooling around so if i come
[1326.98 --> 1333.14]  into your phoenix project i can kind of know where things are how to name them um i just so to that
[1333.14 --> 1341.06]  degree we are um kind of replicating that similar uh experience uh-huh it's not unique we've we've heard
[1341.06 --> 1346.26]  this before though i mean uh when we had taylor outwell on recently uh talking about laravel that
[1346.26 --> 1351.86]  was the case there too where you sort of go to the the camp you like to hang out best at and you
[1351.86 --> 1358.90]  create the worlds the rails of that world more or less i mean yep right so a couple of shows back and
[1359.54 --> 1364.98]  uh a lot of people love that show and we're hearing i think you hear it repetitively in a good way
[1364.98 --> 1369.78]  though it's a good testament back to a lot of the things he helped instill into good web frameworks
[1369.78 --> 1375.54]  oh definitely i i still do rails every other day and i love it so i think you know rails got a ton
[1375.54 --> 1382.18]  right and um you know so we borrow if you look at our router um we borrow some ideas there so they
[1382.18 --> 1385.54]  someone doing rails comes in and looks at phoenix without having any elixir experience they'll
[1385.54 --> 1391.06]  kind of be like hey if they squint a little bit um that looks pretty similar the dna of rails comes out
[1391.06 --> 1395.62]  yeah but we aren't just trying to go feature for feature saying okay like you know we need
[1395.62 --> 1403.70]  it's not a feature x right right gotcha kind of like those uh those movies that aren't they're not
[1403.70 --> 1409.38]  like a retelling of a of a non-fiction they're like based on a true story you know those where you know
[1409.38 --> 1413.38]  like this is not actually a true story it's just based on a true story it's kind of like you know you
[1413.38 --> 1418.10]  want inspired by yeah and you want to you want to think differently and um we've definitely borrowed
[1418.10 --> 1423.78]  some ideas um from other areas too so we're not just looking at rails we're looking at um like
[1423.78 --> 1429.62]  you know socket io was a big inspiration for our real-time layer and i think the problem that people
[1429.62 --> 1436.74]  coming into elixir initially from any other language will say okay well we don't have a device for phoenix
[1436.74 --> 1440.90]  yet let's make device so instead of which is the wrong way to think about it if you need user
[1440.90 --> 1445.14]  authentication then you should start thinking about how do i solve user authentication you know in a
[1445.14 --> 1449.46]  a functional mindset so i think um that's why i just want to caution that we're not just trying
[1449.46 --> 1453.94]  to say okay how do we make how do we clone rails in phoenix no it's like how do you how do we think
[1453.94 --> 1458.90]  about the spirit of rails and how do we apply that on some of the good stuff that we can bring over to
[1458.90 --> 1463.86]  what works best in elixir and what what we need to fulfill our needs yeah i mean you got to play the
[1463.86 --> 1468.66]  strengths of your of your language and and your environment where are you drawing those those different
[1468.66 --> 1474.58]  ideas from if not directly from rails from other experiences from other projects in the elixir
[1474.58 --> 1480.74]  ecosystem from elsewhere yeah so i mentioned so the big thing for me when i started getting into
[1480.74 --> 1487.62]  looking into this was doing real-time events from my rails app so when i saw elixir and right up on
[1487.62 --> 1493.30]  erling you know i immediately realized that this would be perfect for solving the real-time nature that i
[1493.30 --> 1498.82]  wanted with my application uh so socket io i had a little bit of like node experience played with socket io a
[1498.82 --> 1505.06]  little bit and um socket io is awesome because it just lets you gives you a browser client and on the
[1505.06 --> 1509.78]  node server side you can just push and receive events and it's super simple so i kind of drew on
[1509.78 --> 1516.26]  that api a little bit in that experience to say i want real-time events from my browser to my phoenix
[1516.26 --> 1522.10]  app to be as trivial as building like a rest back end like if you're building a rest back end and whatever
[1522.10 --> 1526.98]  framework you're using like you can just snap your fingers and you know instantly how you would build that
[1526.98 --> 1532.34]  so that's kind of a big push on our real-time layer was looking at socket io and then we kind of
[1532.34 --> 1537.94]  extended that a little bit um to kind of namespace events um but i'd say that's been a big uh inspiration
[1537.94 --> 1545.06]  for us so web socket support came in early and is like production ready or what's the status of your
[1545.06 --> 1551.06]  of your web socket support yes so web socket support is production ready and since then uh since elixir
[1551.06 --> 1556.90]  conf we've added it falls back to long polling uh so it's going to work in i think ie 8 plus now
[1557.54 --> 1564.74]  um falls back to long polling transparently and um you can just drop it in phoenix js and uh it just
[1564.74 --> 1570.74]  works awesome have you had uh have you had success with people out there in the community trying it
[1570.74 --> 1578.02]  have you built it have any of your own projects with it yeah so we're using it um one of our own
[1578.02 --> 1583.30]  uh projects which unfortunately i can't share publicly um but people are using it kind of all over the place
[1583.30 --> 1589.62]  uh someone uh recently built like a uh right around christmas time you could play uh jingle bells
[1590.34 --> 1595.78]  remotely with like a party of people so kind of like you know you could coordinate they they did this
[1595.78 --> 1600.26]  like timing algorithm so you could ring your bell and play like you know the jingle bells chime
[1601.30 --> 1608.50]  you know across iphone to iphone just using phoenix channels so i think gaming could be a big uh a big
[1608.50 --> 1614.42]  area like someone i haven't seen this come to fruition but someone was looking into using uh phoenix channels
[1614.42 --> 1620.10]  um over to like an unreal engine so like a native desktop app but you know pushing out the real-time
[1620.10 --> 1626.18]  events back and forth so kind of outside of the html server web space i think phoenix could have a
[1626.18 --> 1633.14]  a happy home there too yeah i even saw that you uh have some ios support going on want to talk about
[1633.14 --> 1640.90]  that sure yeah so i think you know as i mentioned at elixir conf um phoenix is great at building just
[1640.90 --> 1646.50]  what you would think of as html web applications and apis but i kind of wanted to go beyond that
[1646.50 --> 1650.74]  because the web is kind of transitioning changing a little bit you know i don't think browsers are going
[1650.74 --> 1657.78]  anywhere and i'm a web developer but i think we have all these um native apps native clients
[1657.78 --> 1662.98]  internet of thing movement coming on and we need a framework to be able to connect all these devices
[1662.98 --> 1668.66]  and push out events and talk to them so i think phoenix is well positioned for that and i kind of for
[1668.66 --> 1675.94]  version 1.0 we have a ios client um that is i think it works on it's a swift client um so it's not
[1675.94 --> 1681.06]  going to work on any of your uh objective c library libraries but someone has put together
[1681.06 --> 1685.86]  an objective c library and we're going to show up with an android client as well for 1.0 so you should
[1685.86 --> 1690.58]  be able to write mobile applications that are native and have them talking to the same back end that your
[1690.58 --> 1696.10]  web application is talking to from the browser that's pretty nice we're hearing lots more about
[1696.10 --> 1700.42]  the internet of things i mean i think that's we heard that jared with uh some things we haven't
[1700.42 --> 1705.62]  released it with beyond the code when we talked about what in software inspires and excites people was a
[1705.62 --> 1711.46]  lot of the things that ties into like devices um things like the watch for example or other things
[1711.46 --> 1719.06]  that are inside your home i'm a little bit wishy-washy on the internet of things myself it seems like it's
[1719.06 --> 1724.66]  very nebulous and kind of i don't know i don't think aspirational is the right word but
[1725.86 --> 1730.98]  pie in the sky but they're like the real world applications of it don't seem like they're actually
[1730.98 --> 1735.14]  all that useful like you know your toaster can tell you when your toast is done it's like yeah but i can
[1735.62 --> 1741.14]  i can look at my toast and see that it's burnt right um what or your toaster what are your thoughts
[1741.14 --> 1746.10]  your toaster can list itself on ebay did you see that there was someone made one where if you didn't
[1746.10 --> 1751.30]  use it in x amount of days it would automatically ebay itself so it could be sold to someone that would
[1751.30 --> 1756.34]  use it yeah that's hilarious yeah i mean most of our applications are just someone i somewhat agree
[1756.34 --> 1761.06]  with you they're they're they're very kind of silly kind of uh novelties right now i'm sure
[1761.06 --> 1767.46]  it's just because it's very fringe and just upcoming and very much a buzzword but um no doubt
[1767.46 --> 1772.34]  we'll need software to drive those you know those interactions um whether or not they're valuable or
[1772.34 --> 1776.82]  not we're gonna have to find out find out i guess well for you chris though what when you say internet
[1776.82 --> 1784.82]  of things as as in regards to phoenix and elixir what are you thinking of so for me i so i think i i
[1784.82 --> 1789.22]  kind of agree that you know internet of things is just a it's a hype right now that we haven't really
[1789.22 --> 1794.50]  seen great applications of it but i think even even in the mobile space if we just talk smartphones
[1794.50 --> 1799.06]  you know that is we already have the internet of things in my mind everyone's walking around
[1799.06 --> 1804.26]  permanently connected to the internet with a square in their hand uh so i think from that perspective
[1804.82 --> 1810.82]  um coming from again i know i'm not trying to pick on rails but coming from a rails background i just
[1811.54 --> 1815.94]  don't see it in that mindset of being able to have all these connected devices i can't do that in
[1815.94 --> 1820.98]  ruby just based on the concurrency model at least not super well so for me just from like the the
[1820.98 --> 1826.26]  mobile um landscape i want to have a framework that i can maintain a persistent connection to
[1826.26 --> 1833.30]  my devices and you know push out real-time updates like someone um recently wrote an app that uh they
[1833.30 --> 1838.90]  use phoenix channels to track shipments for a new startup so they're they have a web browser it's
[1838.90 --> 1844.82]  an ember app and it's listing you know where the drivers are at all times so it's a native application
[1844.82 --> 1849.22]  pushing out updates over phoenix channels and shows you in real time on a map in the browser where
[1849.22 --> 1854.66]  your drivers are so things like that um i think is what i'm talking about with internet of things
[1855.54 --> 1860.74]  but i think the the toaster and stuff i think maybe in four or five years we'll see useful applications
[1861.30 --> 1865.30]  i think a good application of what you're talking about just with the phone the device connectedness
[1865.94 --> 1871.78]  um is is the meerkat app which kind of has blown up in the last couple weeks um which is a twitter
[1871.78 --> 1876.66]  base it's an io it's an iphone app you guys probably have seen it posted to twitter where
[1876.66 --> 1881.70]  you can immediately start streaming live um just post a tweet out to twitter i think it's like a
[1881.70 --> 1887.06]  single click to start streaming video from your phone and anybody from their phones can just click
[1887.06 --> 1893.22]  and they're now streaming your video live to their phones very simply and you know this is the kind
[1893.22 --> 1898.50]  of thing that just wasn't even possible a couple years ago right i mean now we have the pervasiveness
[1898.50 --> 1904.58]  of lte we all have the devices we have the software there has to be infrastructure behind that right
[1905.46 --> 1909.06]  you get the bandwidth costs which is what i think about is like man how long can these guys
[1909.86 --> 1912.90]  just burn money before they're probably open twitter buys them or something but
[1913.54 --> 1918.02]  how long can they just burn money on bandwidth you know before they start making some money of their own
[1918.02 --> 1924.18]  but then also like their servers back there driving these things and the ability to all of a sudden have
[1924.18 --> 1934.10]  thousands of streaming video connections seems like erlang and the the otp stuff that it does is kind of
[1934.10 --> 1942.18]  well suited for those kinds of large um communities what do you think about that yeah exactly exactly i was
[1942.18 --> 1948.02]  gonna say the uh i went to erlang factory uh last year i think that's the first time i talked about phoenix
[1948.02 --> 1954.10]  publicly and uh one of the whatsapp engineers was there and they were he was talking about that where they
[1955.14 --> 1960.82]  turned on the feature for sharing image and video and they weren't sure how it was going to scale because he was talking about
[1961.30 --> 1966.74]  you can't really you can't really load test these things you can't fake it yeah yeah if you have 400 million users you can't say
[1966.74 --> 1971.86]  like hey let's see how this is going to work throw some servers behind it you can't really load test it so they had to do a live
[1971.86 --> 1976.90]  and uh just worked and they were able to you know turn on the and that was kind of you know sharing video
[1976.90 --> 1982.18]  and images so i think you know very similar to what you were just talking about yeah awesome so the
[1982.18 --> 1987.78]  channel support is really unique i think um and interesting what else does phoenix bring to the
[1987.78 --> 1994.02]  table as far as a framework goes do you consider it a full stack framework yes yeah so the goal is i
[1994.02 --> 1998.90]  think you know i it's not a micro framework i hate that word i think you know if you want to be a
[1998.90 --> 2004.42]  library libraries are fine some people like building their applications out from small
[2004.42 --> 2010.66]  composable pieces but i think in practice at least building a lot of rails applications and inheriting
[2010.66 --> 2015.62]  a lot of rails applications i think having a set of conventions and a set of full features and obvious
[2015.62 --> 2022.02]  ways to do things is great for community so yeah it's definitely the it's a batteries included framework
[2023.54 --> 2028.58]  give us some of those batteries what you got in there cool so we have uh
[2028.58 --> 2035.06]  similar to i mentioned our router is somewhat similar to rails so you can have this dsl for
[2035.06 --> 2042.34]  routing requests into your controllers and we have a our view layer is kind of neat because you can
[2042.34 --> 2047.94]  render we kind of separated the idea of views and templates so views are more your presentation layer
[2047.94 --> 2053.70]  and views render templates and a special thing at least about what we do is we pre-compile all the
[2053.70 --> 2058.90]  templates into the view module so runtime you're just calling a function and it's returning you it's
[2058.90 --> 2064.18]  basically doing string string concatenation whereas you know other frameworks can't really pre-compile
[2064.18 --> 2068.58]  these things so they have to read from disk and cache it and go through all this work but since we have
[2068.58 --> 2074.10]  metaprogramming elixir all that content in your template files is just pre-compiled and baked directly
[2074.10 --> 2079.46]  in as a function call is that how all the metaprogramming works in elixir it's all at compile time
[2079.46 --> 2084.42]  yeah all the compile time so we can do a bunch of work at compile time and at runtime it's super
[2084.42 --> 2087.78]  fast you don't have to worry about caching these things you're not doing any disk reads because it's
[2087.78 --> 2094.66]  all just in memory so but it's just as robust and um powerful as what you would think of in like a
[2095.30 --> 2099.78]  scripted language that's not being compiled so like people are seeing like to give you an idea people are
[2099.78 --> 2105.06]  seeing microsecond response times rendering templates in phoenix which is kind of cool
[2105.06 --> 2112.90]  what else you got you got so routing is i'm sure there's uh it seems like you have mvc but you
[2112.90 --> 2119.78]  also have a what you call a view layer and a template layer can you talk about that yeah views are
[2119.78 --> 2125.06]  basically like a presenter the presenter pattern if you're coming from an object-oriented mindset but
[2125.54 --> 2134.18]  we push it further too so i'm in a controller i can say you know render my show template and you can
[2134.18 --> 2140.02]  explicitly specify the content type or it will use the content type from the accept headers so if it
[2140.02 --> 2145.70]  was an html request it's going to render the html template but we also are promoting the mindset of
[2145.70 --> 2150.50]  your view should render everything so if you want to render json if you're just building like a json api
[2150.50 --> 2156.50]  your view is just a module that's going to construct that json uh for the caller so we kind of i think i
[2156.50 --> 2160.74]  at least haven't seen that in other areas where we're kind of pushing things out of the controller and
[2160.74 --> 2165.70]  keeping your controllers really clean and making the view present whatever you know was requested
[2166.50 --> 2172.10]  and then obviously um on top of that our channel layer is probably the biggest i'd say innovation and
[2172.10 --> 2179.30]  what draws people in and that's what you can do you can push out real-time pub so events to the browser
[2179.30 --> 2187.06]  today and hopefully you know ios and android coming soon cool so that view layer i mean it's it's kind of
[2187.06 --> 2194.34]  like a presenter or a decorator layer but it can itself just serialize json so you wouldn't even
[2194.34 --> 2198.98]  have to have templates at all is that what you're saying yeah exactly exactly but you would still be
[2198.98 --> 2203.54]  rendering a view you wouldn't be constructing the json in the controller as like a as a map in the
[2203.54 --> 2209.22]  controller what if you do want to have just straight up html templates with generated server-side
[2210.02 --> 2218.90]  markup that all yeah so then then you yeah you would just create a html.ex template and uh say render
[2218.90 --> 2223.94]  index.html and it would it would it would precompile that and render it uh so it'd be just the same
[2225.62 --> 2230.98]  and that i assume has your your typical helper methods or whatever they're called in elixir they
[2230.98 --> 2236.10]  called functions is everything a function yeah everything's a function i i tend i helpers is
[2236.10 --> 2240.98]  an accurate way to describe it too so your views can define um all the helper methods for its
[2240.98 --> 2246.66]  templates whether that template is whether that's going to be your html template or just you know
[2246.66 --> 2253.70]  json directly in the uh module body awesome so as far as full stack goes is there anything missing at
[2253.70 --> 2261.54]  this point so uh we're pretty close to 1.0 although i say that and i don't want to give any hard dates but
[2261.54 --> 2267.62]  i'd say what's missing video that you said uh end of by the end of the year you'd be at 1.0 but that
[2267.62 --> 2272.18]  was last year i didn't say which i didn't say which year he's pulling a pearl on us here
[2274.18 --> 2280.10]  he's pulling a software developer on us yeah i pushed it back to uh uh i've said quarter one this
[2280.10 --> 2284.58]  year but we're coming up on that but no really i think we're actually quite close now um what is
[2284.58 --> 2290.42]  missing is uh like internationalization we used to have uh what we tore it out um because we're working on
[2290.42 --> 2296.10]  something better uh so we might release 1.0 without internationalization um and then release like
[2296.10 --> 2303.54]  release like a 1.1 with it so that's been taking out now and um i think for now we're just trying to
[2303.54 --> 2311.22]  settle on our channel api and make sure our um you know the current apis that we have are are good
[2311.22 --> 2315.22]  because you know we don't want to lose we can't release 1.0 and immediately break things but the
[2315.22 --> 2321.70]  core features there you know are baked in so we have routers uh controllers views we even have a
[2321.70 --> 2327.62]  we refer to them as endpoints so we're not like a monolith like rails so you can kind of bring in
[2327.62 --> 2335.22]  phoenix into an existing otp app or you can run multiple phoenix endpoints within an app and everything
[2335.22 --> 2339.46]  kind of just works along the lines of you can have as many of these things running as you want
[2340.26 --> 2344.66]  and then along with that what we've recently done is you know we've had channels for a long time but we
[2345.22 --> 2350.50]  recently um supported different pub sub adapters so the channel channels operate on top of a pub sub
[2350.50 --> 2355.78]  layer and the problem that we used to have is if you weren't running elixir uh nodes clustered
[2355.78 --> 2361.06]  together like on heroku dynos can't talk to each other um you couldn't really make use of channels
[2361.06 --> 2366.74]  over dynos that weren't clustered so we recently added a redis adapter and anyone can implement their
[2366.74 --> 2373.46]  own pub sub adapter now where we kind of use a pub sub adapter to broker the um events so now on heroku
[2373.46 --> 2378.10]  cross your dynos it's just going to work or if you're running elixir distributively it's just
[2378.10 --> 2383.62]  going to fall back to the standard library so that was a big feature for us recently one term i think
[2383.62 --> 2389.78]  we've glazed over a little bit is otp yeah could you unpack that for the audience sure yeah and i you
[2389.78 --> 2395.06]  know when i first saw this too i was thinking like what the heck is that so otp stands for open telecom
[2395.06 --> 2400.90]  platform which doesn't really tell you anything about what it is um so it started off this is you know
[2400.90 --> 2405.94]  historically erlang was built for um telecommunication systems that's where the open telecom platform
[2405.94 --> 2411.14]  comes from but really the best way to describe it is it's erlang's standard library for building
[2411.78 --> 2416.90]  concurrent distributed fault tolerant applications and that's what elixir ships with as well so
[2416.90 --> 2422.82]  anything you're doing as far as holding state or we build applications we call them in like supervision
[2422.82 --> 2428.74]  trees so you have supervisors that are going to supervise different parts of your application so if
[2428.74 --> 2433.06]  something crashes or something goes wrong it's going to automatically be restarted by its supervisor
[2433.62 --> 2438.82]  and that will kind of travel up the chain and um all the way up to your application supervisor and
[2438.82 --> 2442.90]  you kind you know so if everything crashes the whole world's on fire it will keep trying to restart in
[2442.90 --> 2449.70]  the entire you know different chains of the application uh so that's what otp is and um you're going to use
[2449.70 --> 2454.98]  it um everywhere use it everywhere what what is it actually what's the benefit
[2454.98 --> 2462.50]  yeah so uh elixir's concurrency model is we call them uh our unit of concurrency is a process
[2462.50 --> 2467.62]  it's not an operating system process it's like a really lightweight thread and you can run like a
[2467.62 --> 2473.94]  million of them on like you're like on my macbook i can spin up a million processes so what otp is it's
[2473.94 --> 2480.82]  just uh it built out a set of conventions for um managing state in an application and responding to
[2480.82 --> 2487.62]  um failure and those are processes and uh they call we call them gen servers if we want to hold state
[2487.62 --> 2492.26]  in a process we have to basically recurse on a process with its state over and over since we're
[2492.26 --> 2497.54]  an immutable language so it's basically just a set of conventions that kind of naturally came about
[2497.54 --> 2502.98]  when uh erlang was solving these problems and they codified it into otp uh so actually the
[2502.98 --> 2507.70]  best way i've heard it described is it's it's the uh the rails of concurrency nice
[2507.70 --> 2513.54]  and that's pretty accurate all right let's take a break real quick and hear from a sponsor when we
[2513.54 --> 2518.98]  come back we'll talk about deploying phoenix to production so we'll come back in just a second
[2521.14 --> 2527.38]  i could not be more excited about this sponsor postico postico is a postgres client for os 10
[2527.38 --> 2533.86]  developed by a single independent developer named jacob egger jacob has open source chops as well you may
[2533.86 --> 2539.06]  know him as the maintainer of postgres.app and he even used to work on sql pro back in the day
[2539.86 --> 2546.10]  equal pro is a mysql client for os 10 that has a special place in my heart it is a great client with
[2546.10 --> 2553.94]  a capital g but unfortunately it only supports mysql i'm a postgres convert and i've long been clamoring
[2553.94 --> 2560.66]  for a high quality postgres gui for os 10 postico aims to be just that it has all the features you come
[2560.66 --> 2565.62]  to expect from an sql client and more but you know what if something that you want is missing
[2566.18 --> 2572.50]  you can contact jacob i can tell you from personal experience he takes feedback and is very responsive
[2572.50 --> 2578.42]  which is important to me when i'm investing in software but the best part about postico is that
[2578.42 --> 2585.06]  it is native to yosemite and it is designed well it fits right into the operating system copy and paste
[2585.06 --> 2590.82]  just works undo just works keyword shortcuts you bet the postico gone are the days of using
[2590.82 --> 2597.22]  sql guis that stick out like a sore thumb there's a free trial which has no time limit so check it out
[2597.22 --> 2604.58]  postico the best way is to google that word that's p-o-s-t-i-c-o you'll find it at the very first hit
[2604.58 --> 2610.50]  that's the simplest way to get where you need to go check out postico i love it and i think you will too
[2610.50 --> 2618.02]  thanks to jacob for supporting us and let's get back to the show all right so we're back chris let's
[2618.02 --> 2621.70]  talk about deploying i think that's one of the things that tends to hold people up you know getting
[2621.70 --> 2624.74]  into production is one of the things so what kind of environment do you have to deploy a phoenix
[2624.74 --> 2632.82]  framework app to so um it's actually funny because it's easier for me to deploy my phoenix apps these
[2632.82 --> 2638.02]  days to heroku and then to deploy my rails apps and i don't know why like i i did some benchmarking
[2638.02 --> 2642.98]  recently and it took me like two hours to get this rails app started up even though i do rails
[2642.98 --> 2648.74]  full-time um but elixir is actually pretty easy to deploy if we're not talking some of the more
[2648.74 --> 2656.02]  advanced features um it's pretty easy to build on any kind of you know unix linux uh box we ship with
[2656.02 --> 2660.26]  uh usually a lot of people if you're deploying yourself you'll run it behind like a nginx proxy
[2660.98 --> 2666.90]  um but it will happily serve you know directly over port 80 if you wanted to you know live on the
[2666.90 --> 2673.54]  dangerous side yeah but it's not it's not too hard to stand up um elixir i usually i tend to build
[2673.54 --> 2681.14]  from source and have any problems um erlang is available um for pretty much any distribution you
[2681.14 --> 2686.50]  can think of um and heroku has a build pack for it so it's pretty easy to stand an app up on heroku
[2687.38 --> 2692.02]  quickly and then if you want to get into like advanced appointments um we haven't mentioned this
[2692.02 --> 2699.14]  yet but erlang and elixir has this idea of uh hot code uh uploading so i can upgrade my code in
[2699.14 --> 2705.86]  production from like one state to the next and um for literally zero talent downtime deploys so not like
[2705.86 --> 2711.62]  you know some people are doing where you have like nginx like serve all the requests and then wait and
[2711.62 --> 2719.54]  shove the request on a new instance it's actually like literally i can update a module from one piece of
[2719.54 --> 2724.50]  code to the next and have what it was doing pass that off to the new code if that makes any sense
[2725.14 --> 2732.82]  how does that work man magic sounds like it sounds like magic to me yeah so so i don't have experience
[2732.82 --> 2737.86]  in this regard uh we have we do have a guide up but it's called uh they're called uh releases we do have
[2737.86 --> 2744.26]  a guide up on phoenix framework.org for performing these and it will work but you have to you have to
[2744.26 --> 2748.02]  think about a lot of things in your code to make sure they can work well with releases so that's the
[2748.02 --> 2754.58]  only thing like if you don't need that absolutely you know critical zero time downtime deploy uh type
[2754.58 --> 2759.54]  setup it's it's i would say not worth it so you have to kind of weigh the odds of development cost
[2759.54 --> 2763.94]  of maintaining that and then you also really have to test releases because um you want those things
[2763.94 --> 2769.54]  to work perfectly um so if you're like a telecommunication company um for example like the joke is
[2769.54 --> 2773.94]  you've never heard like at&t doesn't call you up and say you know sorry you can't make a phone call
[2773.94 --> 2779.30]  tonight from 6 to 10 p.m because we're going down for scheduled maintenance like you never that never
[2779.30 --> 2785.94]  happens so i think for for some um for some companies they can they can do that and they
[2785.94 --> 2790.50]  have been doing that successfully uh so you kind of have to just weigh the odds but that does work
[2790.50 --> 2796.90]  there's a tool called exrm which is like elixir uh release tool for building releases and performing
[2796.90 --> 2803.46]  hot code swapping but it definitely requires some more tlc i'd say so just to be clear we're not talking
[2803.46 --> 2810.98]  about a rolling restart we're talking about live code swapping just like bam yeah exactly so that
[2810.98 --> 2814.50]  yeah you can do rolling restart but yeah this is this is like a whole nother ball game of
[2815.06 --> 2820.98]  if i have like you know if i have a counter keeping track of number of active users and that thing's in
[2820.98 --> 2826.82]  the middle of like running a piece of code that module i update that counter module the new piece of
[2826.82 --> 2832.42]  code will get the previous state what that counter was and you actually write a callback in a function to
[2832.42 --> 2837.22]  say there's a code change like here's what here's the state that i had now you take it and make it
[2837.22 --> 2842.98]  into some new state that you want to handle that makes any sense uh so it's pretty i mean pretty
[2842.98 --> 2847.38]  amazing that these kind of things are built in but like i said it's going to take some more work
[2847.38 --> 2851.94]  is this the language level or phoenix level this is at the language level okay
[2852.74 --> 2857.14]  the advantage of being on erlang because like you said the telephone companies didn't have the
[2857.14 --> 2863.46]  luxury of downtime like we have with the web right so i think what i recommend to people is
[2863.46 --> 2868.50]  you know rolling restarts are totally fine and i think you know to some degree you you're going to
[2868.50 --> 2873.06]  have to write your application around you know if someone trips over the server power cord things are
[2873.06 --> 2879.30]  to go down anyway so you still want to be able to react to going offline anyway so i think unless you
[2879.30 --> 2885.86]  have the the peculiar you know use case for releases um i think at least getting started you don't have to
[2885.86 --> 2891.30]  really worry about it but if you do have those high demand i think there's a it's not whatsapp
[2891.30 --> 2896.02]  there's another early chat app that they're running on releases um so you know they're get they're like
[2896.02 --> 2900.98]  guaranteeing message delivery type stuff so it's really cool if you have a especially if you're running
[2900.98 --> 2906.10]  kind of some software as a service like uh like you know pusher or uh some of those companies where
[2906.10 --> 2910.98]  you know you can't really stand to have messages dropped you have to guarantee that there's not
[2910.98 --> 2916.42]  going to be any kind of blip but it's a nice option let's switch gears a little bit and talk
[2916.42 --> 2922.42]  about the community one thing i've noticed so far is you've just referred to it as we pretty much the
[2922.42 --> 2928.74]  entire conversation yeah when speaking about phoenix who's all involved in phoenix and then by extension
[2928.74 --> 2935.22]  the elixir community tell us about that sure so uh we've had the great fortune of jose
[2935.22 --> 2940.66]  valim has joined the phoenix core team uh so he's been working on phoenix quite a bit for the last
[2940.66 --> 2947.86]  several months and he's contributed some of the biggest features uh as of late um so it's jose is
[2947.86 --> 2954.18]  a big core contributor we've recently we have a jason steves um sunny scroggen and i'm going to miss
[2954.18 --> 2959.94]  a few people here but we have a set of i think five people on the core team right now and the community
[2959.94 --> 2965.46]  in general i think is one of elixir's biggest assets as far as um you know we're trying to establish
[2966.58 --> 2970.98]  um the meme in the community of you know we're very helpful helpful and welcoming because a lot of these
[2970.98 --> 2976.18]  concepts are difficult to get started in especially if you're coming into phoenix you're listening to
[2976.18 --> 2980.66]  this podcast and you're like wow this sounds great let me dive in so suddenly you're learning
[2980.66 --> 2984.98]  like a new web framework you're learning a new language then you have this like otp thing that
[2984.98 --> 2989.38]  you're trying to learn so it's like a new programming paradigm so like there's like seven different
[2989.38 --> 2994.66]  things that are like totally new to you um so we're trying to kind of build a community around
[2995.30 --> 2999.54]  being very supportive asking for help receiving help and i think we've been pretty successful with
[2999.54 --> 3006.18]  that where you said uh somebody pinged you earlier i think it was in the pre-call how you got ping for
[3006.18 --> 3012.10]  coming on the show was in an rc channel what was that rc channel yeah it's uh elixir lang
[3012.10 --> 3018.90]  elixir hyphen lang on uh free node that's that's where i direct anyone that's on twitter mentions
[3018.90 --> 3022.90]  anything about elixir i tell them to hop on if they have you know get stuck or have any questions
[3022.90 --> 3029.22]  is there one specific for phoenix uh so not yet so whenever jose gets tired of seeing all the phoenix
[3029.22 --> 3034.10]  stuff in the elixir channel that we'll make our own but for now it's um kind of one of the same right
[3034.10 --> 3041.54]  now yeah he's pretty savvy of jose as the language author and obviously highly invested in the success of
[3041.54 --> 3049.30]  elixir to participate in the development of phoenix because um adoption especially of a language that
[3049.30 --> 3056.10]  suits itself so well to the web um could be highly dependent upon a popular and very high quality
[3056.10 --> 3060.26]  web framework do you think that was kind of the deciding factor of why he got involved
[3060.82 --> 3066.26]  or was it just pure excitement for what you're up to yeah i think um that that probably has something
[3066.26 --> 3071.38]  to do with i know from like from the beginning um jose wanted to you know he comes from a web
[3071.38 --> 3076.82]  background and he had started a prototype web framework called dynamo um why he was building
[3076.82 --> 3082.58]  elixir so from the beginning he was definitely thinking about um how elixir is well suited to web
[3083.30 --> 3088.02]  but i think it was uh bruce tate uh had convinced him that he couldn't do he couldn't write a web
[3088.02 --> 3094.18]  framework and a new programming language at the same time just it was too much so instead he uh he
[3094.18 --> 3100.42]  built plug which was like a uh http uh middleware and like web server abstraction kind of like rack
[3100.42 --> 3107.06]  you're familiar from ruby so instead of he kind of used that dynamo framework to prototype some ideas
[3107.06 --> 3112.58]  and why he was building elixir to see how they could coexist and then um step back from that built
[3112.58 --> 3117.30]  that middleware library and then that that way i was able to come in and build on build on top of his
[3117.30 --> 3121.54]  middleware uh so i think it was just kind of like you know the timing all together you know while he was
[3121.54 --> 3127.30]  finishing elixir i had found plug and started building phoenix on top of plug and then he was
[3127.30 --> 3132.34]  able to finish elixir and then he hopped on board so kind of timelines worked out really well and i
[3132.34 --> 3137.94]  think um yeah the phoenix's co-bases is definitely it's been hugely beneficial to have him on board it's
[3137.94 --> 3144.02]  awesome so you'd say that phoenix is still built on top of plug and still has this middleware thing
[3144.02 --> 3151.30]  going on yeah so and that's the big thing for us is i i didn't want to hide the the plug layer so
[3151.30 --> 3155.38]  like if you're coming you know i've been talking about rails a lot but i come from a rails background
[3155.38 --> 3159.94]  like uh you don't really you know you're running on top of a rack but unless you're using rack
[3159.94 --> 3164.74]  middleware explicitly you're you know in day-to-day you're not thinking about it whereas in phoenix
[3164.74 --> 3170.42]  we very we coexist with plug at a much finer level so you have to be kind of aware of the plug
[3170.42 --> 3176.34]  concepts and any kind of middleware or any kind of like uh filtering you do on a request is all
[3176.34 --> 3184.50]  happening um with a plug api interesting so it looks like you just released 0.10
[3185.78 --> 3188.82]  um with a whole bunch of goodies in there including live reload support which
[3189.30 --> 3194.34]  kind of amazes me because i've never seen live reload done without an extension
[3195.22 --> 3199.62]  which i'm assuming you're just using the web socket support to kind of shove that in there
[3200.42 --> 3204.66]  yeah so that yeah 0.10 is super exciting for me just because we've we've gotten we've like
[3204.66 --> 3209.06]  streamlined the development experience so like it feels like like i have a i have a more enjoyable
[3209.06 --> 3212.74]  development experience um today with phoenix than i do rails at least you know starting out
[3213.38 --> 3220.02]  so with live reload since we have the channel web socket layer built in we basically thought you know
[3220.02 --> 3224.50]  last month i was like wait a second why can't i just inject that a little bit of javascript on the
[3224.50 --> 3230.82]  into the body of your html page and then watch the file system for changes and and just basically
[3230.82 --> 3236.58]  dog food our own real-time system so it ended up being like you know 20 lines of code to have live
[3236.58 --> 3242.10]  live reload work and it doesn't require node.js doesn't require a browser plugin it's all just
[3242.10 --> 3249.06]  kind of dog food in our own real-time apis that is really cool you also have some asset pipeline stuff
[3249.06 --> 3254.34]  you've got um some new form helpers looks like it's you're really adding the goodies now tell us
[3254.34 --> 3261.22]  about 1.0 and then give us a just a brief overview of what you see for the future of phoenix sure
[3261.22 --> 3268.50]  yeah so uh 1.0 is basically like we said going to be a full stack uh framework we have we are using uh
[3268.50 --> 3274.98]  we're using brunch which is a node uh build tool for asset compilation um so we didn't want to write our
[3274.98 --> 3279.70]  own asset pipeline because we didn't think i mean there are mature uh solutions out there so we
[3279.70 --> 3286.10]  evaluated like half a dozen of the popular node options i have like huge amount of hours in that
[3286.10 --> 3292.26]  but we ship with you know you just throw your css and your es6 javascript uh sas into a specific
[3292.26 --> 3296.90]  folder it just gets compiled so you don't really have to think about um brunch or node or anything it
[3296.90 --> 3302.10]  just works a little bit pre-installed on your machine or something yeah if you install phoenix it
[3302.10 --> 3306.18]  will detect that you have node installed and automatically set it up so if you have node
[3306.18 --> 3311.78]  installed already it would just work um otherwise you can forego the brunch integration but then you
[3311.78 --> 3318.90]  just have to build your own build tool um but we have a uh good asset story uh real-time events is in
[3318.90 --> 3325.30]  for 1.0 so i think you know for now from here until i think the end of april which is elixirconf eu
[3325.94 --> 3330.18]  is what we're targeting for a 1.0 release it's just going to be about stabilizing the apis
[3330.18 --> 3334.42]  there's some channel work that i want to get in place that is going to make channels uh a little
[3334.42 --> 3339.78]  nicer to deal with at like the concurrency level but it's kind of a uh implementation detail uh but
[3339.78 --> 3344.66]  but the biggest feature remaining for 1.0 uh i haven't mentioned yet is being able to replay channel
[3344.66 --> 3350.10]  messages so if i'm online if i've written a chat app in phoenix and i drive under a bridge on my cell
[3350.10 --> 3356.98]  phone right now you'd miss any of those messages that were broadcasted out through the server so we want to
[3356.98 --> 3360.98]  be able to buffer those messages messages on the server and then when the client reconnects after
[3360.98 --> 3366.10]  they get out of the tunnel it just flushes the buffer so we're going to include that hopefully as
[3366.10 --> 3372.10]  a 1.0 release with mobile clients and i think that would be a solid 1.0 release to compete with
[3372.82 --> 3377.54]  with most of the frameworks out there except we should be able to take on the world so to speak as far as
[3378.34 --> 3386.90]  scalability goes and beyond 1.0 what you got so i have some ambitious goals i'd say
[3386.90 --> 3392.66]  with my elixir conf talk i talked about distributed services and this is be something that we probably
[3392.66 --> 3396.50]  uh this would be something that lived under the phoenix framework organization on github but not
[3396.50 --> 3403.22]  part of phoenix core itself but the idea is we have you know elixir i haven't mentioned this but you can
[3403.22 --> 3408.90]  you write code to run on one machine and you can run that on 50 machines connected together and it's
[3408.90 --> 3414.18]  pretty much the same code so the concurrency model is distributed at its heart so you don't really
[3414.18 --> 3419.62]  have to think about um receiving a message from a different um server on your cluster it's all just
[3420.18 --> 3426.10]  baked into the concurrency model so the idea there is i want to be able to write a service layer that
[3426.10 --> 3432.82]  i can spin up multiple nodes running kind of different services so if i have a i'm writing a search engine
[3432.82 --> 3439.30]  uh in elixir and i'm using phoenix i want to be able to say okay i have a page indexer service that is
[3439.30 --> 3444.42]  pretty expensive it has to make page requests you know parts the html and i know that i can do
[3445.06 --> 3450.66]  a thousand of those on one box at a time so i want to be able to spin up 10 nodes running that code
[3450.66 --> 3455.86]  and have them register themselves on the cluster saying hey i can perform the page indexing work
[3455.86 --> 3462.34]  or hey i can perform the page crawling work and then at my uh at my application level code i just want
[3462.34 --> 3468.26]  to be able to say hey service page crawler crawl this page and it will find a node on the cluster that can
[3468.26 --> 3473.46]  do the work that's available do the work for me and give me a result so there's the gist of kind
[3473.46 --> 3478.26]  of the ideas i'm playing with um is there i think there's a lot of hard problems to solve there around
[3478.26 --> 3483.70]  like you know cap theorem and distributed programming but elixir enables all that it's really exciting to
[3483.70 --> 3489.54]  me that that's all possible you're getting a lot of benefits into phoenix itself from elixir
[3490.42 --> 3495.14]  oh yeah you know choosing the right kind of language and then getting a lot of the benefits from the
[3495.14 --> 3500.66]  the limits itself yep like i mentioned in my uh in elixir conf talk too like if you're running
[3500.66 --> 3505.78]  channels you can i'm running channels on 10 different machines i could have one machine dedicated to
[3506.82 --> 3511.38]  streaming twitter results and if i find something interesting on twitter it can broadcast it out on
[3511.38 --> 3515.62]  a channel and that would show up in the browser so as long as it's connected to the cluster i could you
[3515.62 --> 3520.82]  know like i said have one machine dedicated to doing these things and you just push out an event and it
[3520.82 --> 3525.94]  gets really rich uh broadcast out to any listeners on the cluster and that's just built in that's just
[3525.94 --> 3532.10]  the concurrency model is i send you a message it's just gonna go over the network but the process could
[3532.10 --> 3536.90]  be you know located on any machine but it's the same as if i was writing that for just my laptop
[3537.70 --> 3543.30]  it's pretty cool it does sound pretty cool so let's talk about getting into the closing here for the call
[3543.30 --> 3548.82]  just uh for listeners sake i've been waiting to ask this because i've been biting my tongue over here on on
[3548.82 --> 3553.70]  getting started because i feel like you got so much stuff that we've talked about for those who
[3553.70 --> 3559.54]  could be coming from ruby so maybe a uh you know a similar look to the syntax but once you get deeper
[3559.54 --> 3564.98]  in is as we said before it's not the same um but still some of the fundamentals some of the dna of rails
[3564.98 --> 3570.50]  is in phoenix and so you kind of have a home away from home but how do you get started i know that elixir
[3570.50 --> 3577.06]  1.0.2 or plus is a requirement is elixir on most machines how do you yeah well how do you get to
[3577.06 --> 3583.30]  start with elixir or with elixir now so phoenix sure so elixir yeah elixir lang elixir hyphen
[3583.30 --> 3589.30]  lang.org has getting started guides for pretty much all major platforms so if it's if you're on a mac
[3589.30 --> 3595.86]  you can do a brew install elixir uh if you're on linux it's just you know apt um so it's easy to get
[3595.86 --> 3601.54]  running um elixir itself and then for phoenix we have phoenix framework.org getting started guides as well
[3601.54 --> 3608.82]  um and just takes you step by step um you should have a um an app up and running um with you know
[3608.82 --> 3612.90]  live reload and assets and in under you know a couple minutes i think you know in the in the latest
[3612.90 --> 3618.18]  010 release video i showed that where you just run mix phoenix new it generates a new project and
[3618.74 --> 3624.10]  sets you up and you're ready to go well i'm just about a minute or two what's the what's the hello
[3624.10 --> 3627.78]  world the perfect hello world for someone listening right now to jump into
[3627.78 --> 3633.30]  the perfect go towards if they're going to build something with phoenix i'd say like a little a
[3633.30 --> 3640.10]  little chat app i have i put a little example out um like the the chat app is like the uh the real
[3640.10 --> 3645.30]  time to-do list type thing like if you're on if you're coming from like node or meteor js you're like
[3645.30 --> 3648.66]  oh i can make a chat app so easily but i think that's a good one because it just kind of shows you
[3648.66 --> 3653.46]  how easy it is to push events back and forth uh so i'd say yeah check out a phoenix chat app kind
[3653.46 --> 3660.74]  of fun to get get your feet wet awesome we uh i do want to take a time a second to mention this
[3660.74 --> 3665.46]  too for those who are listening that are members um we've been working with elixir sips for quite a
[3665.46 --> 3672.50]  while we've never had a show obviously on elixir or phoenix for that matter but um for those who are
[3672.50 --> 3681.22]  members we do have a discount with elixir sips it's 77 off 77 off to be more clear um it's about six bucks
[3681.22 --> 3687.22]  for the first three months so to get started on some of these things i know that i do such a great
[3687.22 --> 3694.18]  job on uh on you know just exposing a lot of this knowledge so diving into getting deployment started
[3694.18 --> 3699.70]  all sorts of stuff so you remember taking advantage of that for sure but we do ask some really awesome
[3699.70 --> 3705.86]  questions heading out of the show and our favorite to ask is uh who's your programming hero programming
[3705.86 --> 3712.66]  hero so many doesn't matter i mean just share gotcha so i'd say uh matt's uh from ruby would
[3712.66 --> 3721.38]  probably be up there um i think just because it kind of embraced being um happy programmer and having
[3721.38 --> 3726.34]  a language that kind of put that at the forefront for me was like a breath of fresh air coming into
[3726.34 --> 3731.78]  ruby where um you know i don't think at least i had no experience where before you know the creator was
[3731.78 --> 3736.10]  like programming should be you should be happy we should um all be nice to each other and it should
[3736.10 --> 3739.94]  be a enjoyable experience and that's kind of the way i feel about programming it's like a it's a
[3739.94 --> 3746.18]  creative process and uh brings me a lot of joy so i think that um has really shaped a lot of people's
[3746.18 --> 3752.90]  opinions about programming so he he'd probably be up there with that another one we like to ask is uh
[3753.78 --> 3759.54]  if you weren't doing what you're doing now which is building out phoenix on elixir uh using elixir
[3759.54 --> 3765.30]  what if you weren't doing that what would you be doing um probably something uh like aerospace
[3765.30 --> 3772.34]  industry i'm a huge space nerd so uh space has always been a passion of mine um so i think something
[3772.34 --> 3778.34]  in the you know planetary science aerospace industry but um from a programming perspective i always think
[3778.34 --> 3786.10]  that like the mars rover for example like they they update the code on that thing on mars so i always
[3786.10 --> 3791.22]  think like you know if there's a bug in my code i get like a something went wrong error page on the
[3791.22 --> 3795.30]  web browser but i don't know that i don't know that i would have the stomach for that industry
[3795.30 --> 3801.30]  just like you know rake deploy and you know it's running on mars so i think um yeah definitely space
[3801.30 --> 3805.14]  is a huge interest of mine and i wait like two and a half years or something for your code to get out
[3805.14 --> 3809.78]  there right yeah well you have to yeah like best case like some people you write code and then 10 years
[3809.78 --> 3814.42]  later you're yeah you're hoping it works so i think uh well i'm i'm super passionate about that so
[3814.42 --> 3822.02]  something along those lines that's funny um i guess a good another good one for closing out is a
[3822.02 --> 3828.26]  call to arms to your projects elixir phoenix um you know where are some ways that the community
[3828.26 --> 3833.38]  listening to the show now could step in where's a good place to either help out with the language itself
[3833.38 --> 3839.22]  or help out with the framework and join the core team or help out where's a good place to start
[3839.22 --> 3846.02]  yes so uh phoenix framework.org is uh the best place to start and also elixir lang irc
[3846.66 --> 3851.06]  is um you know i live on there i'm probably on there too much but i'm happy to help anyone out
[3851.06 --> 3857.14]  that wants to jump in and i think um if you already have some experience in elixir um you can always
[3857.14 --> 3860.82]  you know contribute and give back whether it's helping someone else out that has questions or
[3861.54 --> 3865.70]  now is kind of a great time to get in early and start putting together third-party packages
[3865.70 --> 3871.14]  so like you know the plug middleware that we talked about um that's kind of the open season
[3871.14 --> 3875.70]  for if you want to build you know a first class authentication system kind of all these different
[3876.34 --> 3880.82]  big uh checklist items that the community doesn't have yet you know start building something and
[3881.30 --> 3884.66]  might find a problem that's not solved and you can solve it and kind of give back
[3886.10 --> 3890.90]  awesome and uh what is your github what is your twitter how can people follow you to kind of keep
[3890.90 --> 3896.82]  up with you in general besides irc when you're sure it's uh chris underscore mccord on twitter
[3896.82 --> 3902.34]  and then just chris mccord on github awesome chris anything else you want to mention as we close out
[3903.38 --> 3907.94]  uh like i said i just published uh metaprogramming elixir on pragprog oh yeah so check that out
[3907.94 --> 3912.74]  uh if you is that pre-order now or is it is it uh available now it's out it's been out for about
[3912.74 --> 3919.06]  a month and a half so nice check it out we'll put a link in the show notes for that uh maybe uh
[3919.06 --> 3923.78]  i'll ask you now do you have any discount codes you can give our listeners i don't at the moment
[3923.78 --> 3928.34]  but uh i can see what i can do awesome so if you're listening check the show notes we'll see
[3928.34 --> 3933.46]  if we can get one if we can get one cool if not we'll just uh we'll just flame chris and rc or
[3933.46 --> 3941.22]  something like that no problem also uh check out elixir conf uh eu it's uh end of april in krakow
[3941.22 --> 3945.62]  poland and i'll be there uh talking about phoenix so it should be a lot of fun we've got the
[3945.62 --> 3950.98]  jose creator of elixir be there and the creator of erling joe armstrong is also giving a keynote
[3951.54 --> 3956.34]  so it should be pretty cool stuff all right well that's it for this show everybody thanks for
[3956.34 --> 3963.22]  listening to uh this great conversation i i love the you know i was quite silent in this one just
[3963.22 --> 3967.46]  because some of the stuff is over my head when you said otp i was thinking something else but uh
[3968.42 --> 3972.90]  you know great conversation today so thanks for coming on the show and uh with that let's say goodbye
[3972.90 --> 3983.30]  see you
[3991.54 --> 3993.38]  oh
[3993.38 --> 4023.36]  We'll be right back.