Player is loading...
How are we going to migrate to Scala 3.0, aka Dotty?
Lukas Rytz, Lightbend
Wednesday, June 12, 2019 · 4:46 p.m. · 44m 43s
Scala 3.0 has been in the works for a number of years. We have probably all seen Martin's keynotes showing off exciting new features and explaining consolidations in the language design. The future of Scala is bright! This talk focuses on the path that will lead us to Scala 3.0. First we will recap the most important changes in Scala 3.0. Then we will look for incompatibilites and possible complicaitons, and show how they can be addressed. I will also talk about the Scala 2.14 roadmap, as the main goal of this release is to smoothen the transition to 3.0.
Embed
Transcriptions
Note: this content has been automatically generated.
00:00:00
thank you everyone for coming i'm really happy to talk about the uh scot free today so
00:00:07
um if you look at the session at the session on a you know what the schedule
00:00:13
and yep then you see that i slightly change the title so in in in uh originally i was the question
00:00:20
how are we going to migrate discover three and while preparing this talk i got you know
00:00:25
i i obviously myself started looking more more into what's called three is going to be like
00:00:30
and myself i get quite a lot more confident that this thing is really going to be to
00:00:34
go very well so i decided to change the title to be more you know more positive more
00:00:40
certain that this thing is going to be a very nice transition into solid three
00:00:46
um so really confident about the future and happy to talk about that day
00:00:52
um i'm working at in this got emit light and so that's team meetings got to nowadays
00:01:00
um does make try to handle and for you know if you want to reach out
00:01:05
in anytime um after the conference i'm okay and burns on one hour away from here
00:01:15
other than to my um i did a p. h. d. in
00:01:18
martin's lap um i since five years now i'm working for like band
00:01:23
i worked on various aspects of this colour out to compiler like the most in
00:01:28
the back and the optimise your first two thirteen also worked on the standard libraries so
00:01:34
what kind of things uh here's the outline of the talk so
00:01:38
there are three parts first part is a perspective on scott three
00:01:43
from somebody that wants to migrate to it so if you're an ordinary scott to user
00:01:48
and you want to migrate discover three you know what what is it going to be like
00:01:53
impart to part ways called working together um
00:01:58
they're different aspects of working together so there's obviously how people
00:02:04
and teams are working together to makes colour three a success
00:02:08
but there's also very technical aspects in working together so how is this colour three compiler
00:02:14
working together with you know all the existing ecosystem around scholar
00:02:20
so that's um like the big title for the second part of the talk and the last part is a bit more practical
00:02:26
questions you know but maybe come up when somebody wants to migrate discovered three l. talk a bit about time line as well
00:02:38
alright part one a perspective once colour three
00:02:42
so the goal of this talk is not to show you all the new nice and ah elegant
00:02:47
features that's got three will put bring along you've seen many of those in the keynote of course
00:02:54
but still i want to present a few futures um you know that's a few features that will
00:02:59
stella three we will bring just to show how they relate to use colour to code and also to
00:03:05
makes colour three a bit more you know give it with the for promotion and what your appetite
00:03:13
um yeah so if you if you want you know to explain to somebody what
00:03:18
is colour three all about what are the core principles behind it you know we
00:03:22
we discussed about this if you know many times already of course and and kind
00:03:26
of this still distilled into these four points that i've i have on this slide
00:03:32
um maybe if you look at other resources the first point uh the first point your compatibility was colour too
00:03:40
that doesn't show up all that often so you know people usually talk about colour three oh about all the
00:03:46
simplifications and the new features and how great it's going to be no doubt
00:03:52
but in reality compatibility with colour to is really at the core of everything that's happening
00:03:58
um in the development of scot free and this point i put it
00:04:02
on the top now on this slide because by now it is released
00:04:06
be on the top of the priority list maybe it was a
00:04:10
bit further down in the beginning but it's slowly started bubbling up
00:04:13
the more we start we we discussed about you know what will always this actually going to happen
00:04:19
so by now everything that's happening us colours three will it is always a keeps the compatibility of scholar to in mind
00:04:27
other than that the main principles plans colour three are simplifications so
00:04:32
this language features that are maybe not too complex are we want to kind of them
00:04:39
make its something's easier to use for the programmers and maybe also forty one language implement there's
00:04:48
one thing that has really changed between scholar through to whence colour three is um
00:04:54
the fact that it's colours becoming more opinionated i will talk about this a bit more in the next slide
00:05:00
that's coloured free really tries to embrace idioms in praise patterns that are common installer and make them
00:05:08
um easier to write for programmers and the last point is a consistency so in force colours strength
00:05:16
that means um there are many aspects of colour that were considered like where
00:05:21
you know the the it's got three team looked at school i'm maybe missing pieces all send and swear you know skull has
00:05:29
s. all the baseline but you know maybe some inconsistency is here
00:05:33
and there and thinking most of the time type system in this area
00:05:38
so yeah let me talk a bit more about this paradigm shift about you know it's colour becoming more opinionated
00:05:46
so the best example to to talk about this i think is is this the uh is in place it's
00:05:53
it's got to win place it's can be used for a lot of things um
00:06:00
impact these these are the most common um you know programming
00:06:03
patent leather expressed using in place it's type classes of course
00:06:08
are they were discussed in various talks already in this conference extension methods
00:06:15
contextual abstraction it which means you just want to um you know to pass parameter along
00:06:21
that them is a very common pattern i mean this colour three compiler do weights implemented
00:06:27
um and also type level computation it so if i i'm scott that when places
00:06:34
are very very low level feature which allow you to um implement all of these things
00:06:40
but as a user as an average user if you see a type class or if you see some type level computation
00:06:48
you see implicit keywords and you start wondering okay you know you have to kind
00:06:52
of know the pattern how these things are expressed in order to understand what's going on
00:06:57
in order to understand even but this is about so in order to understand all this is a type class and you know
00:07:03
here i'm using a type less you have to know this is one way of using in place it's so it's very in direct
00:07:11
and it's got three um all of these features get um
00:07:17
specific support in the language so the employees it's keyword will
00:07:23
um we'll slowly slowly be phased out so you will see it less and less when you look at colour three code and
00:07:31
the idea is really to allow programmers to harm to to our express their intent so i
00:07:38
want to integrate the type plus i want to use like high cost of their syntax for that
00:07:45
and as somebody who has been working with martin for ten
00:07:48
years now i could say um it's it's really not only with
00:07:52
uh you know ah it's so maybe i should so ten years ago i
00:07:57
i remember in places were considered really cool because it was such a powerful feature
00:08:02
that would allow you to do many things um you know
00:08:05
without complicating to language meaning without adding stuff to the language
00:08:10
and it was considered a strength of scar that we have one feature to express many different
00:08:15
things and that's really a change of perspective no that happened now i'll be look at carlisle
00:08:22
and a slightly different way and say hey you know people want to do this let's give them the best way to achieve it
00:08:30
okay so i'm going cricket run through a few examples you've seen
00:08:35
this already in martin's keynote so i can go fast but when you
00:08:38
define an extension method is got to you have to be really careful
00:08:42
not to mess things up so one one now interesting example is that
00:08:47
what you can define imply implicate class to create an extension method but you have to be really
00:08:52
uh i'll careful to make this value your private because it happens otherwise otherwise every
00:08:58
string was suddenly um get an s. extension method or field which returns itself and
00:09:06
uh the avoiding this kind of you know corner cases is is
00:09:10
yeah it's really a annoying and it comes up again and again
00:09:16
so yeah you see how it looks it's got
00:09:21
a second example again i would say there's a few more times you've seen it in the keynote already um
00:09:28
yeah you know is a a new feature to express at um abstract data types
00:09:35
also to replace coloured enumeration has got it and told enumeration is defined in the library and it uses hack
00:09:41
with the reflection and it's also difficult if you want to add a feature to it it's really
00:09:45
not easy often too you know so this thing has yeah that has not been loft very much
00:09:52
so you know of course is a replacement for that but also for a d. t.'s um again when you define
00:09:57
an eighty teens colour to there's just boilerplate associated with
00:10:01
it and you have to be careful to get everything right
00:10:04
and um it's got a three this very common pattern just become simpler to to to write down
00:10:14
okay last example is a type person coding is the same type
00:10:17
last that the miles used in his presentation just in precision um
00:10:24
here's how you do it it's got also uh it's got too so you have to
00:10:28
be again quite careful we have to you know you have to you make sure to use
00:10:32
any place it while or an object here in it or could get f. as well but
00:10:36
anyway that would be not very efficient you have to make sure you could use a uh
00:10:41
you you you know you crate apprised method here to create
00:10:45
um show type for option that itself takes an implicit show and
00:10:51
um yeah this is just a good example where you have in place it the keyword showing up
00:10:56
in various places and you have to get it right it's not very easy and as a
00:11:01
as a user you have to learn to see through this pattern to understand what's going on
00:11:07
so yeah here's the new syntax for that basically everything i every place where i'm in place it shows up
00:11:13
in the code will be replaced by something new so
00:11:16
uh you can define implicit instances using the delegate keyword
00:11:22
the out the name of them instance can be are given or not so it can be anonymous
00:11:29
and uh the the this like four permit priced
00:11:34
instances uh the syntax is more closely related let's see
00:11:39
it's important to see that that they've been if definition of the type class
00:11:43
is still a trade or napster class so that didn't change and also using
00:11:47
a type class on the bottom on the last line is the same so
00:11:51
here's the same example in as it looks in a is colour three um yeah
00:12:00
it's worth mentioning that it's about the the the
00:12:04
this colour three is providing more than just syntax um
00:12:09
the interesting feature in this area is type restoration verse got three gives you
00:12:14
primitives to um so uh as there if you write a type plus you can actually um
00:12:21
like define how instances of your type was should look like using a very
00:12:26
generic our representation and then uh if somebody wants to extend your show type class
00:12:33
if it has the right structure you can just use our your generator
00:12:37
presentation and doesn't have to writing code so there's also a lot of
00:12:41
in mention going on again not miles talked about this a lot in his tall in just previous session
00:12:53
so this one goes in the category of enhancing scholars strength i would say of
00:12:57
course the scallop type system is one of its you know core features is one of
00:13:01
the distinguishing features of scott that has very strong type system and there's quite a
00:13:06
lot of evolution going on in this area so it's gotta forgets union and intersection types
00:13:12
on it's worth mentioning that they are not tactic meaning types are not packed which means of um
00:13:19
if uh the so there's the uh the uh if you have a value of a union type
00:13:24
it's just either or devalue um of that
00:13:28
that that matches but there's nothing no information at
00:13:31
run time telling a out which of the two cases it um is actually represented right now
00:13:38
so there are corner cases when at run time you couldn't uh tell apart you know which of the two types your in but they're pretty rare
00:13:46
other um features in the type system i type when does um martin mentioned
00:13:52
in keynote oh the various new kinds of function types that's cause colour three supports
00:13:58
and there's also a new type inference algorithm which um
00:14:03
is you know i it's more powerful so it can
00:14:07
sometimes give you a better solutions at at um uh the um inferring the type for complex inspections
00:14:16
okay so i'll that was my quick overview of a few scholar three features that
00:14:22
are coming but uh uh now we are going to put them into three categories
00:14:27
by how they affect migration from scarlet to create that i created a four categories here one is
00:14:34
breaking changes since colour three second is new features so stuff that you can start using once migrated
00:14:42
number three is features that continue to be supported so access things colour two features that
00:14:46
are still the same it's got a three but de emphasise twitch so features that you know like in place it's that are
00:14:53
i'm slowly being phased out in favour of new things
00:14:58
and number four is stuff that didn't change so
00:15:02
breaking changes um for some so general existential types
00:15:08
are no longer supported in um in source code in scott three source code so this colour three compiler
00:15:14
doesn't know that keyword for some white cards are the same as they used to be inspired to so they still work
00:15:23
um another feature that's not supporting it's got three is early initialise it's it's a
00:15:28
bit of a corner case features that um if you act happened to use it
00:15:33
i guess most people wouldn't uh might have never come across it but if you happen to use it their work arounds
00:15:40
i'm already today that you can use and colour to and got three um provide
00:15:45
straight parameters which um which are really out a a good replacement for early initialised
00:15:52
there are a number of features that's got a three supports in compatibility mode
00:15:57
um so these are not very complex mostly syntax stuff so procedures syntax
00:16:04
has been deprecated already i think into thirteen simple materials um
00:16:10
other applications so if you have a function that takes an empty parameter list uh all you have to actually
00:16:16
provide the ante up anti a parameter list at at a call side
00:16:22
um okay the late in it is uh
00:16:26
i heard going to be supported installer three in the end it's not implemented at this point um
00:16:33
oh so every packages in in places got so that's the
00:16:37
there's a a small change an implicit resolution that the packages um
00:16:43
okay so so uh when you when an object always
00:16:47
in in place it's coped and its parents are as well
00:16:51
so it's in closing score as well but it's got to be exclude packages there
00:16:56
and but in that again in in this compatibility mode um it is going to
00:17:00
support and one small thing that i'm a bit sad about this this nail operated
00:17:07
um it comes up in pattern matching with wire x.
00:17:11
maybe if you use it i think i've used it once so far i
00:17:15
just find it cute but unfortunately has new syntax and got three i think
00:17:23
okay
00:17:25
one big breaking change that i didn't mention of the previous slide is macros and programming
00:17:31
so my cousins got three are going to be different than they are today it's
00:17:34
got too so there's going to be a new a. p. i. to implement macros
00:17:39
can you help yeah it's more principal so it's um in scarlet we literally but we
00:17:44
basically exposed in terms of the compiler to the user that writes a macro and are
00:17:51
you know if you want to get full access to all the compiler intern is you
00:17:54
can you can do basically a lot of complicated things that conflict with countries that type jacket
00:18:01
you can also should yourself in the foot very easily by um creating
00:18:05
trees in the wrong shape or trees if the wrong types and things like
00:18:09
so the the new a. p. i. for macros ins colour three is is more principal it's more
00:18:14
um so it has it has like better foundations with in line in quotes
00:18:19
yeah with cotton supplies uh sorry and the a. p. i. that you use to write um
00:18:24
your macros twin allies the syntax trees and build new syntax trees is based on tasty which means oops
00:18:31
the the um which means the a. p. i. is supposed to
00:18:34
be stable over a along time over many versions of scott three
00:18:42
so there was a talk to dave any cause to key i think i missed it yeah i think
00:18:47
tomorrow okay so it took more than a where you will show how much the programming in topic in scholar three works
00:18:57
some macros are no longer needed and that's actually big one
00:19:01
again as talked about is just in the previous session but especially
00:19:05
everything that is related to type restoration can now be handled by
00:19:08
the language without macros and i think that's a really big improvement
00:19:15
good ah last rating change especially station so there this colour three team is still
00:19:23
um you know that's one of the things that still on the drawing board still being
00:19:28
considered scott three i was told will support at least oops
00:19:35
specially station for core pipes so it was that support special station for
00:19:39
functions into paul's which is really needed for performance if using scotland standard library
00:19:46
um you know how difficult is it to implement specially
00:19:50
stations so special icing methods apparent yep but is relatively straightforward
00:19:55
just duplicating code specialise in classes um gets a
00:20:00
bit more complicated special lighting super classes and trade
00:20:03
that's where where it gets really hard so um it's again it's not yet clear what
00:20:09
will happen exactly it's colour three with the at specialised annotation and definitely are the people that
00:20:16
if you have a that that the lab are interested to hear you
00:20:20
know about the weather to important use cases so reach out of your effected
00:20:27
a good second category of a change like if a change is it's got three is new features so here's
00:20:35
the shortlist um trade parameters opaque types
00:20:40
were mentioned already top level definitions in arms okay
00:20:44
extension methods change the type system and then like you
00:20:48
met the programming facilities match types in line matches um yeah i'm not going into details on the slide
00:20:55
i put the few when green and these uh green boxes are
00:21:00
things that are basically on the road map for back according to to fourteen so that means you will be able to
00:21:06
start using these features already in this colour too serious and
00:21:09
they will be available for cross building between scott one score three
00:21:16
so yeah once you migrated just got three of course you can
00:21:18
start gradually move migrating your coat using start using the new feature but
00:21:23
again crossed building will be ah a limiting factor for a while so projects that to
00:21:28
cross built between is that it's got two and three there would be limited to the concepts
00:21:36
but now next categories colour two features that will continue to work on style of three
00:21:43
i actually most people when i you know talk in the hallways and as people around
00:21:49
i i hear that a lot of people are not really aware that's colour three this colour three compiled as it is today
00:21:56
it supports almost all of scarlet to there's very very
00:21:59
little code that actually doesn't work when using discover three compiler
00:22:04
so here's a list of things that work exactly in the same way as the work
00:22:09
and it's got three well okay maybe they're small differences in type inference find but um
00:22:14
in general this colour three compiler compiles all of these features
00:22:18
the same as this colour to compile some in place it's
00:22:21
there's no change house colour three handles place it's it continues
00:22:25
to supporting basic parameters implicit conversions implicit classes also value classes
00:22:31
are here like some of the thrills compound types are still there
00:22:35
package up sick objects are still there so this is just to
00:22:39
assure you when you start cross building your existing project is colour three
00:22:43
it's not that you don't have to limit yourself to a
00:22:46
very small subset of scholar but you can basically continued use scarlet
00:22:52
the way you do it today and with uh for many project is we'll cross built quite easily
00:23:01
that's our goal fox uh yeah things that didn't changes colour three
00:23:06
um the standard library including collections of course so the way
00:23:11
this works today is um when at this colour three itself
00:23:17
well the preview releases that are being published does not have its own
00:23:21
collection uh it's nonstandard library so just got three release uses colour to binary
00:23:28
um standard library it uses some additional classes so there's an additional charge file it uses some tricks
00:23:35
to replace some classes on the class about that i i'm not aware of what he tells but
00:23:40
it's important to know that the standard library is the same and it will
00:23:43
be the same between scholar three the final resent this colour to fourteen release
00:23:49
the tooling we will invest of course to make sure that all the tooling oh
00:23:55
that is important for for for the the users is going to continue to work
00:24:01
and then of course a big um uh well a very big
00:24:07
a step for every may just gotta realise is getting the ecosystem ready
00:24:12
so we just released colour to thirteen and now you know people are slowly waiting and to to be able to
00:24:17
pop state library waiting for it depends to become available and it's got a three is going to have the same
00:24:24
i'm the same prude put step procedure to some degree there's
00:24:28
going to uh i will talk about binary compatibility know what
00:24:32
and we are we are really committed to how um the maintains of of libraries
00:24:39
to cross build their libraries for tots colour twenty three and make
00:24:42
sure that the ecosystem is available will be available for so colours three
00:24:47
so as a user if you migrate from two to three you don't
00:24:50
have to think about changing you know all your dependencies at the same time
00:24:55
so the goal here really is that as a user i can choose you know
00:25:01
to update my dependencies get ready get my project in shape and then i
00:25:05
can move this got three with the same dependencies that i'm already you see
00:25:15
i assume yes i assume that's true fourteen yeah so i will talk about
00:25:19
two fourteen mineral map and timelines stealing a bit but yes the goal for
00:25:25
yeah so our the way we see it is people would would migrate
00:25:29
from scholar to thirteen to fourteen and from there to scott of three
00:25:36
okay and just you know random selection of features that are
00:25:40
still the same as colour three so classes and objects functions
00:25:44
pattern matching all these things didn't change are the platforms
00:25:48
joint probability we continue to do use the same test it except or okay
00:25:55
good that was it for part one
00:25:59
maybe have to speed up the little if i okay part two working together um
00:26:07
this slide is i'm i'm i'm still going to spend some time on it because i think it's
00:26:12
important maybe people are not so much a you know aware of all that your graphics behinds got um
00:26:18
this red circle with the big red box that's e. p. f. l. so who is
00:26:22
located that e. p. f. l. lamb is the research lab where martin others case delete 'em
00:26:30
that's the lap that develops i'm scott of three the kids got three compiler of
00:26:34
course it has other responsibilities also the p. f. l. is located this colour centre
00:26:40
so the way uh from sebastien is the the leaders discuss enter the
00:26:45
ask us into focuses on tooling we still metals today um as
00:26:50
for the focus on a education and in general um fostering discover community
00:26:55
and then there's this got team at like and which maintains its got to um
00:27:02
this colour standard library like by adrian and you see it's very distributed so we have set
00:27:08
a nature in in san francisco you have huge in in new york stefan and me in
00:27:15
middle of europe and jason in australia that's how we work together
00:27:19
and um i in today's nice eighteen something very important happened namely eight when he moved from s. f.
00:27:28
into the red circle and um so that's a year and a half ago right
00:27:36
and in this one year and has the communication between you know the scarlet team
00:27:42
and the lamp team the communication between it's got two aunts colour three teams it has really uh
00:27:48
it hasn't accelerated and improve and we've been working much more closely together since then
00:27:54
and um i think this is one of the one a big drive for
00:27:58
about uh uh behind making its colour to anne's colour three finally coming together
00:28:06
so where does discuss three you you know where to all these discussions happened where is colour three being
00:28:11
developed most of the things are happening in public so there's a lot of discussion going on at the
00:28:17
top the issue tracker and ample requests also on the discussion forum
00:28:23
so these things are really open to everybody the discussions are um are numerous and
00:28:28
they are long and a lot of people are try me and so i think this
00:28:32
is working quite well the community is um is is very active in the development
00:28:37
of colour three us well there's the subcommittee we're all the new features are being discussed
00:28:43
the subcommittee has members of e. p. f. l. does people from
00:28:46
scott centre people from my pen but also people from the community
00:28:51
so that's ultimately what features are being accepted or rejected and
00:28:56
the nodes meeting notes of the sea sick meetings are being published something
00:29:00
to this mall late i'm also for me in many cases there are um
00:29:05
there are videos online of uh like i think to meeting some sometimes trained on a life and you two went
00:29:10
the videos are available so again this is this is pretty open of course it's off line discussions over coffee and beer
00:29:18
um three times for your to like vince colour team but delighted engineering
00:29:22
people need somewhere in europe or in the us and martin is there
00:29:26
as well so these meetings of course important to synchronise to discuss new stuff
00:29:31
there's also weekly meetings at e. p. f. l. re can often try and since he's on campus page
00:29:39
okay so that's working together teams working together on the technical
00:29:44
level like you know how how are this colour is is going
00:29:47
to work together that's basically what defines the road map for two fourteen zero two fourteen is all about preparing for scott free
00:29:55
um we're going to backward some features um on the list
00:29:59
is typed them thus opaque types strip parameters and top level definitions
00:30:04
we didn't publish this colour to fourteen group that uh yet we will do that soon it will open
00:30:09
it for discussions and um uh and with of course would happy for everybody to to which i mean
00:30:17
um we will continue with applications of features that are
00:30:21
that are uh you know slowly being phased out installer three
00:30:25
so for some existential so our application package object inheritance and early initialise there's
00:30:32
are things on the list that that that might get deprecated and we're also
00:30:38
going to and we plan to remove some of the you know lighter
00:30:41
feel like like the syntax um changes already that have indicated into thirteen
00:30:52
so if you look at the development of sky major releases to twelve to thirteen to fourteen
00:30:58
this always happens one after the other okay so discovered life and team
00:31:03
has now finished to thirteen the finals out okay there will be two thirteen one of course maybe
00:31:08
people that find a buck or two but basically by the time tools to thirteen zero is released
00:31:14
the team starts working on two forty so that's that's thought that um are moles of working on major
00:31:20
releases now first got to fourteen and colour three this is different because they're being developed at the same time
00:31:28
by different teams of it um
00:31:34
timeline yep there will be a slider oh about time like later but um so we will work on
00:31:39
this to realise is at the same time to to realise is will share the stand the same standard library
00:31:44
how it works technically in detail that's me on there might be a few changes they'll but it's
00:31:50
oh it's going to be the same standard labour into two releases we will invest
00:31:55
um between the two co yeah teams to invest a lot of time in sharing more
00:32:00
code so we want to make sure that the two compilers use the same test suite
00:32:05
we will look into uh possibilities of sharing components of the
00:32:09
compiler so already today discovery compiler uses basically the same back
00:32:14
and this colour to except that it's a fork so it's a copy of the source code we we want to proof
00:32:19
the situation of course um we will try we will share code m.
00:32:25
in the way it's colour to will emit the tasty um output there's also code that
00:32:29
can be shared care and so we will start also merging the two code bases into one
00:32:37
and again one of the goals for our release is to have the possibility for maintain is
00:32:43
to cross bill to fourteen and three that's really the main the main goal of two fourteen
00:32:49
yeah so inter operability at the binary level so two days colour three can use
00:32:54
um libraries that are compile was colour okay to twelve currently but it will be to fourteen at some point
00:33:02
what this allows is basically people to operate as colour
00:33:06
three early so pop people cut start using scott three when
00:33:10
uh before all of their dependencies has been cross compiled so that's a very powerful
00:33:15
feature of course to to you know to go ahead and try things out um
00:33:21
if it's not yet clear in every detail how this will work like if if if
00:33:25
we really want to support is at a like a a a professional level let's say
00:33:30
we need to make sure that the two compilers are are the same in terms of
00:33:34
code generation and then that will need a lot of testing and and still some investment
00:33:40
there are some carry out so one of them for example ah you
00:33:43
know if you have just got to library that uh defines a macro
00:33:47
this colour three compare is not going to be able to execute this colour to
00:33:51
macro because that just depends entirely on the and pearls of this colour to comply either
00:33:56
so this on interoperate build he's not going to be the like the
00:34:00
final solution for all the cost building um problem so therefore we will
00:34:05
help and we will in a um um encourage people to cross build the libraries between the two releases it
00:34:13
um testing yeah so already mentioned if we really want
00:34:17
to achieve this binary compatibility will have to invest a lot
00:34:20
into testing to make sure the two compilers generate the same byte code
00:34:24
for um for the same source code um there's this a continuous so
00:34:31
running joke very say we want to build a frankenstein compiler which is kind of
00:34:35
a mix between this got two and the three compiler so here the idea is
00:34:39
it's got to would you know a parse the source code and write out the tasty
00:34:45
um to like the taste and then we would use this colour three compilers to do the rest of the compilation pipeline and
00:34:53
no one uh this would basically be like an integration test for tasty
00:34:56
and integration test to show that's colour to um generates correct takes the output
00:35:05
still on the testing front one of the things that we are relying on a lot these days is
00:35:10
the community both so every scholar releases are tested by
00:35:13
the community build which tools to scully consistent from source
00:35:17
it's roughly three million lines of code into twelve were
00:35:20
slowly getting there and two thirteen actually already pretty far and
00:35:25
come you build force colour three is only getting started these days um so it's still pretty
00:35:30
small but this will be of court so very important aspects in testing has got the three days
00:35:36
and it's important to know that it's not about only about testing are if things works
00:35:42
but it's also about like testing how much that we actually break in terms of yeah you
00:35:47
know uh changing stuff around so when we when we receive redesign the collections into thirteen seeing
00:35:55
which kind of things would break in the community bold gave us a good estimate of you know how how bad our
00:36:01
our breaking changes and we could quantify that so the communicable
00:36:05
is not only for correctness testing but also for like feature validation
00:36:11
alright let's part is uh about time line and you know maybe few
00:36:15
practical aspects about getting there so we just released got two thirteen zero um
00:36:21
martin said in his keys not keynote that's colour three m. one which will be the feature freeze
00:36:27
force colours three is going to be in the in the fall in the end of this year
00:36:32
and yeah from there on it's from there or it's you know it's a
00:36:36
forecast and estimations which are always a bit difficult especially if it's about the future
00:36:42
so so uh yeah it is oh so yeah um
00:36:51
somewhere you know in the future there will be a scarlet to fourteen
00:36:55
realise that will be discussed release within the next one to two years um
00:37:02
there's a lot of things that need to be figured out and told and of course you know which releases going have is if there is one
00:37:08
realise that has to come first probably to fourteen has to come first
00:37:12
because colour three values it's standard library all these things have to be discussed
00:37:19
how do you migrate discover three well you move discover to fourteen first this well um
00:37:24
this will allow you to start using some of the new features already
00:37:27
which you know it will allow you it will simplify cross building as well
00:37:32
um so that's really what this to fourteen releases all about it's it's supposed
00:37:36
to pay for the way just got of three um yeah we have a tool
00:37:44
a call scott affix a um which can rewrite source code it can it can fix 'em syntax
00:37:52
well we can fix 'em syntactic you know changes that are required by um um one example is the
00:38:00
application to exclude d. v. d. r. application off
00:38:04
uh in the parameter lists and things like this so we can also be write a simple the
00:38:10
thrills uh that don't know deprecated things like this so we are we started working on our repository
00:38:16
because colour re writes recently where we want to collect migration rule still
00:38:20
rewrite rules first colour fix and definitely we will spend more time um
00:38:27
on this on a on these rules to make sure the small tedious things can be
00:38:33
automated and you know people have to only intervene with the stuff that is actually breaking
00:38:41
yeah we will absorb some of the shop let's say into fourteen so we
00:38:44
weren't already to we want people to migrate off certain things already into fourteen
00:38:51
like the existential types orderly initialise there's um this is basically
00:38:56
to you know to make the transition to scott of three
00:39:00
a bit easier and these things are coming anyway so we decided it's it's good to spread out a bit
00:39:05
over the time of the migration some fifty some states breaking changes
00:39:11
and the macro so if if you define macros in your library
00:39:15
or in your application um yeah this will have to be written
00:39:19
once you start moving just got three if you cross killed you have to most like you
00:39:23
have to you know have separate source files that's my next like check slide actually yeah so ah
00:39:29
the goal is really to clock cross build the ecosystem between scarlet once colour three
00:39:35
um i was very pleased to see miles saying that he wants to
00:39:39
backwards got that his new shapeless three that is building on top
00:39:43
of colours when he wants to back work that has got to fourteen
00:39:46
because it will be really important that we have the same ecosystem onto fourteen available
00:39:51
so users don't have to to uh you know the upgrade of the language and the pens is at the same time
00:40:00
yeah so if you're defining macros you will probably have to have separate source files for the two versions
00:40:07
there is a possibility of recording on the the new mac re p.
00:40:11
i. does or does the tasty based mac right p. i. just got to
00:40:15
um but this thing is just on a on the drawing board right now
00:40:19
we haven't we haven't started any of this work or or committed to anything
00:40:25
ah one thing that might come one day is uh some kind of preprocessor first colour with um
00:40:35
we have been talking about is already a few times and there's just situations
00:40:39
where you know this would simplify life's so maybe this will come at some point
00:40:47
okay so
00:40:50
just in terms of uh uh um responsibilities from our side from this colour team at light and
00:40:57
um we are going to be the main developers of stolid to fourteen we will
00:41:02
maintains got to fourteen for a very long time that's what we expect
00:41:06
it's we expected to be the last release of this colour too serious
00:41:11
so that means it will have to have on the maintenance stretched and then then other releases that people
00:41:18
once to fourteen is released the like pants got him
00:41:21
will start working maintaining and supporting us to discover three compiler
00:41:29
and the team at p. p. f. l. um yeah it's mostly a
00:41:33
a is is the core behind the development of the scroll through comp
00:41:38
right so that's what i have my main message for this talk that is that's got a three is really colour to plus
00:41:46
one so it's it's not the new language is the same
00:41:51
language but there's plus one so there's new stuff in it and
00:41:56
we're really working together we are you know we are committed to ensure that the migration will be smooth if we're
00:42:02
also committed to spend time on health in the community to make sure the eh 'cause it ecosystem can move ahead
00:42:09
and uh yeah we hope you're all as excited about the new stuff as we are
00:42:15
thank you
00:42:24
for one quick question please raise your hand if you have
00:42:29
maybe he actually cannot then we can do
00:42:36
uh_huh okay uh_huh so many companies use go out because uh_huh sparks so
00:42:44
i'm just wondering if there is any collaboration will communication between no like band
00:42:49
old two thousand and ah good debris solaris sparc people to make use
00:42:56
of migration to tell us we only have all been entered smooth as possible
00:43:03
yeah so definitely there is communication there was actually today after the keynote and yes um
00:43:09
we are yeah so we we we are aware that the scale at spartan to eleven was
00:43:13
kind of a a a big you know big lack for many people and um that's
00:43:18
also why we got and you know we we invested some time and resource from like bend
00:43:23
to helping data bricks and well but sparkling the let's say to to migrate has got to twelve
00:43:28
so it's got to twelve will be the default for spark three which is supposed
00:43:32
to become coming out in a few months i heard like six months maybe um
00:43:37
and yes we talked about it's got three as well today and yes that's definitely on the radar um
00:43:43
i just thought for example today's got as part doesn't have any macros which
00:43:46
is great you know so the yeah thank you uh_huh okay quick question then
00:43:55
so no two thirteen forty sixty two fifteen to sixteen the ah
00:44:02
okay yes we we don't plan to have a two fifteen release
00:44:09
yeah but you can slowly against his tooth foreseen turn out to be a little that's what you have like no
00:44:15
plans pretty nice i mean our goal of course is is you know is the success of scholar and not too
00:44:22
make people suffer from mistakes that we did or what not so you know if if there are
00:44:27
good reasons for us got to fifteen realise that you know it's not that uh it's excluded by definition
Conference Program
A Tour of Scala 3
Martin Odersky, Professor EPFL, Co-founder Lightbend
June 11, 2019 · 5:15 p.m.
8337 views
A story of unification: from Apache Spark to MLflow
Reynold Xin, Databricks
June 12, 2019 · 9:15 a.m.
1267 views
In Types We Trust
Bill Venners, Artima, Inc
June 12, 2019 · 10:15 a.m.
1569 views
Creating Native iOS and Android Apps in Scala without tears
Zahari Dichev, Bullet.io
June 12, 2019 · 10:16 a.m.
2232 views
Techniques for Teaching Scala
Noel Welsh, Inner Product and Underscore
June 12, 2019 · 10:17 a.m.
1296 views
Future-proofing Scala: the TASTY intermediate representation
Guillaume Martres, student at EPFL
June 12, 2019 · 10:18 a.m.
1157 views
Metals: rich code editing for Scala in VS Code, Vim, Emacs and beyond
Ólafur Páll Geirsson, Scala Center
June 12, 2019 · 11:15 a.m.
4695 views
Akka Streams to the Extreme
Heiko Seeberger, independent consultant
June 12, 2019 · 11:16 a.m.
1552 views
Scala First: Lessons from 3 student generations
Bjorn Regnell, Lund Univ., Sweden.
June 12, 2019 · 11:17 a.m.
577 views
Cellular Automata: How to become an artist with a few lines
Maciej Gorywoda, Wire, Berlin
June 12, 2019 · 11:18 a.m.
386 views
Why Netflix ❤'s Scala for Machine Learning
Jeremy Smith & Aish, Netflix
June 12, 2019 · 12:15 p.m.
5026 views
Massively Parallel Distributed Scala Compilation... And You!
Stu Hood, Twitter
June 12, 2019 · 12:16 p.m.
958 views
Polymorphism in Scala
Petra Bierleutgeb
June 12, 2019 · 12:17 p.m.
1113 views
sbt core concepts
Eugene Yokota, Scala Team at Lightbend
June 12, 2019 · 12:18 p.m.
1656 views
Double your performance: Scala's missing optimizing compiler
Li Haoyi, author Ammonite, Mill, FastParse, uPickle, and many more.
June 12, 2019 · 2:30 p.m.
837 views
Making Our Future Better
Viktor Klang, Lightbend
June 12, 2019 · 2:31 p.m.
1682 views
Testing in the postapocalyptic future
Daniel Westheide, INNOQ
June 12, 2019 · 2:32 p.m.
498 views
Context Buddy: the tool that knows your code better than you
Krzysztof Romanowski, sphere.it conference
June 12, 2019 · 2:33 p.m.
394 views
The Shape(less) of Type Class Derivation in Scala 3
Miles Sabin, Underscore Consulting
June 12, 2019 · 3:30 p.m.
2321 views
Refactor all the things!
Daniela Sfregola, organizer of the London Scala User Group meetup
June 12, 2019 · 3:31 p.m.
514 views
Integrating Developer Experiences - Build Server Protocol
Justin Kaeser, IntelliJ Scala
June 12, 2019 · 3:32 p.m.
551 views
Managing an Akka Cluster on Kubernetes
Markus Jura, MOIA
June 12, 2019 · 3:33 p.m.
735 views
Serverless Scala - Functions as SuperDuperMicroServices
Josh Suereth, Donna Malayeri & James Ward, Author of Scala In Depth; Google ; Google
June 12, 2019 · 4:45 p.m.
936 views
How are we going to migrate to Scala 3.0, aka Dotty?
Lukas Rytz, Lightbend
June 12, 2019 · 4:46 p.m.
709 views
Concurrent programming in 2019: Akka, Monix or ZIO?
Adam Warski, co-founders of SoftwareMill
June 12, 2019 · 4:47 p.m.
1974 views
ScalaJS and Typescript: an unlikely romance
Jeremy Hughes, Lightbend
June 12, 2019 · 4:48 p.m.
1377 views
Pure Functional Database Programming‚ without JDBC
Rob Norris
June 12, 2019 · 5:45 p.m.
6375 views
Why you need to be reviewing open source code
Gris Cuevas Zambrano & Holden Karau, Google Cloud;
June 12, 2019 · 5:46 p.m.
484 views
Develop seamless web services with Mu
Oli Makhasoeva, 47 Degrees
June 12, 2019 · 5:47 p.m.
785 views
Implementing the Scala 2.13 collections
Stefan Zeiger, Lightbend
June 12, 2019 · 5:48 p.m.
811 views
Introduction to day 2
June 13, 2019 · 9:10 a.m.
250 views
Sustaining open source digital infrastructure
Bogdan Vasilescu, Assistant Professor at Carnegie Mellon University's School of Computer Science, USA
June 13, 2019 · 9:16 a.m.
375 views
Building a Better Scala Community
Kelley Robinson, Developer Evangelist at Twilio
June 13, 2019 · 10:15 a.m.
245 views
Run Scala Faster with GraalVM on any Platform
Vojin Jovanovic, Oracle
June 13, 2019 · 10:16 a.m.
1342 views
ScalaClean - full program static analysis at scale
Rory Graves
June 13, 2019 · 10:17 a.m.
463 views
Flare & Lantern: Accelerators for Spark and Deep Learning
Tiark Rompf, Assistant Professor at Purdue University
June 13, 2019 · 10:18 a.m.
380 views
Metaprogramming in Dotty
Nicolas Stucki, Ph.D. student at LAMP
June 13, 2019 · 11:15 a.m.
1250 views
Fast, Simple Concurrency with Scala Native
Richard Whaling, data engineer based in Chicago
June 13, 2019 · 11:16 a.m.
624 views
Pick your number type with Spire
Denis Rosset, postdoctoral researcher at Perimeter Institute
June 13, 2019 · 11:17 a.m.
245 views
Scala.js and WebAssembly, a tale of the dangers of the sea
Sébastien Doeraene, Executive director of the Scala Center
June 13, 2019 · 11:18 a.m.
661 views
Performance tuning Twitter services with Graal and ML
Chris Thalinger, Twitter
June 13, 2019 · 12:15 p.m.
2003 views
Supporting the Scala Ecosystem: Stories from the Line
Justin Pihony, Lightbend
June 13, 2019 · 12:16 p.m.
163 views
Compiling to preserve our privacy
Manohar Jonnalagedda and Jakob Odersky, Inpher
June 13, 2019 · 12:17 p.m.
302 views
Building Scala with Bazel
Natan Silnitsky, wix.com
June 13, 2019 · 12:18 p.m.
565 views
Diversity and Inclusion Panel: Bring the Thunder!
Panel
June 13, 2019 · 2:30 p.m.
245 views
Asynchronous streams in direct style with and without macros
Philipp Haller, KTH Royal Institute of Technology in Stockholm
June 13, 2019 · 3:45 p.m.
304 views
Interactive Computing with Jupyter and Almond
Sören Brunk, USU Software AG
June 13, 2019 · 3:46 p.m.
681 views
Scala best practices I wish someone'd told me about
Nicolas Rinaudo, CTO of Besedo
June 13, 2019 · 3:47 p.m.
2707 views
High performance Privacy By Design using Matryoshka & Spark
Wiem Zine El Abidine and Olivier Girardot, Scala Backend Developer at MOIA / co-founder of Lateral Thoughts
June 13, 2019 · 3:48 p.m.
754 views
Immutable Sequential Maps – Keeping order while hashed
Odd Möller
June 13, 2019 · 4:45 p.m.
277 views
All the fancy things flexible dependency management can do
Alexandre Archambault, engineer at the Scala Center
June 13, 2019 · 4:46 p.m.
389 views
ScalaWebTest - integration testing made easy
Dani Rey, Unic AG
June 13, 2019 · 4:47 p.m.
468 views
Mellite: An Integrated Development Environment for Sound
Hanns Holger Rutz, Institute of Electronic Music and Acoustics (IEM), Graz
June 13, 2019 · 4:48 p.m.
213 views
Closing panel
Panel
June 13, 2019 · 5:54 p.m.
400 views