►
From YouTube: GitHub Satellite India 2021 - Building the Julia programming language and its community
Description
Presented by Dr. Viral B. Shah, Co-Founder & CEO, Julia Computing
Julia is one of the popular programming languages for technical computing, widely used in domains like pharmaceutical development, space missions, financial modeling, and others. Julia is being actively developed on GitHub—with more than a thousand contributors from across the world. In this session, the co-creator of JuliaLang, Viral Shah, will speak about the journey of building Julia in the open and the challenges the team faced, and share some insights from their journey in open source.
As always, feel free to leave us a comment below and don't forget to subscribe: http://bit.ly/subgithub
Thanks!
Connect with us.
Facebook: http://fb.com/github
Twitter: http://twitter.com/github
LinkedIn: http://linkedin.com/company/github
About GitHub
GitHub is the best place to share code with friends, co-workers, classmates, and complete strangers. Millions of people use GitHub to build amazing things together. For more info, go to http://github.com
A
A
A
lot
of
it
mostly
on
github
and
and
everything
I'm
going
to
talk
about
is
is
not
just
about
me,
but
about
what
the
broader
julia
community
has
built.
So
julia
has
been
a
long
journey.
It's
been
around
for
almost
11
years
now,
and
we
started
this
journey
in
2009
with
an
urge
to
solve
the
two
language
problem.
A
The
two
language
problem,
very
simply
put
is,
is
a
phrase
that
we
coined
actually
after
after
starting
the
julia
project
and
the
reason
why
we
had
to
come
up
with
it
was
because
it
was
hard
to
explain
what
why
why
the
world
needed
a
new
programming
language.
Why
did
the
world
need
julia
and
the
reason
it?
The
reason
is
as
follows:
right
you,
everyone's
used
to
programming
with
python,
with
r,
with
matlab
with
all
of
these
interpreted
languages.
That
are,
you
know,
not
not
really
great
for
performance,
but
then
you
always
end
up.
A
Writing
your
programs
in
c
c,
plus
plus
fortran
java,
in
order
to
get
higher
performance.
So
people
like
to
have
a
certain
kind
of
language
for
data
exploration
for
interactive
exploration,
for
you
know
for
for
just
sort
of
interacting
with
your
algorithms
and
then
another
language
for
getting
performance,
and
it's
really
odd
to
write
the
same
program
twice,
and
this
is
why
we
started
out
by
creating
julia,
to
solve
this
two
language
problems
so
that
you
can
have
the
ease
of
python
and
the
speed
of
c
all
in
one
package
in
one
language.
A
So
in
2009
we
wrote
the
first
line
of
code
and
in
2012
we
actually
published
a
post.
A
blog
post
called
why
we
created
julia.
If
you,
if
you
google,
that
term
you'll
land
up
right
there
and
you'll
see
what
what
it
was
about
and
a
lot
of
what
we
wrote
back
then
still
holds
true,
and
this
is
how
julia
got
started.
It
was
the
four
of
us
with
jeff
bizanson,
stefan
karpinsky
and
alan
edelman.
A
A
I
was
actually
based
out
of
bangalore
when
I
started
this
project
and
and
three
of
my
collaborators
were
based
out
of
the
us,
but
all
of
us
were
in
very
different
geographies
and
it
was
a
completely
distributed
setup
in
in
where
we
started.
You
know
talking
by
email,
set
up
a
git
server.
This
was
this
was
not
yet
on
github
back
then
we
actually
had
a
git
server
at
the
university
of
california,
santa
barbara
that
stefan
used
to
maintain
and
very
early
on
in
the
project's
life.
A
A
For
those
of
you
who
do
not
know
this
jupiter
notebooks
are
extremely
common
for
data
science
and
they
stand
for
julia
python
and
r
and
so
julius
j
in
jupiter.
In
2015.
We
actually
founded
julia
computing
because
of
all
the
interest
around
the
julia
programming
language
and
some
of
the
first
companies
started
using
julia
commercially.
A
We
hit
1
million
downloads
in
2017.,
1.0
was
released
in
2018
in
julia
con
in
london,
and
2019
saw
10
million
julia
downloads.
We
were
extremely
thrilled
to
receive
the
the
wilkinson
prize
for
our
contributions
to
julia
and
also
the
sydney
fernbark
prize
last
year
was
a
couple
of
years
ago
now,
and
we
are
now
close
to
25
million
downloads
and
even
though
my
slide
says
julia
1.5
released.
In
fact,
yesterday
we
just
released
julia
1.6.
It
has
been
getting
rave
reviews.
It
was
trending
on
hacker
news
just
about
yesterday
and
it's
it's.
A
So,
while
julia
as
a
programming
language
solved
the
two
language
problem
for
scientists
and
engineers,
you
know
it,
it
really
is
making
impact
in
the
real
world.
People
are
using
julia
for
faster
drug
development.
You
know,
companies
like
paiser
and
moderna
are
using
julia
to
come
up
with
new
drugs
and
and
take
them
through
the
drug.
The
the
drug
pipeline
process.
A
A
So
these
are
some
really
difficult
scientific
problems
that
julia
is
addressing
today
and
what
I'm
going
to
talk
about
is
the
journey
from
developing
the
julia
language,
developing
a
community.
You
know,
building
out
julia
computing
and
then
being
able
to
actually
work
on
some
of
these
cutting
edge
problems.
A
So
here's
the
the
julia
project
on
github-
it's
you
know
just
julia
lang,
slash
julia
julia
lang
is
the
organization
julia's
the
project.
It
has
33
000
stars
and
tons
and
tons
of
comments
yeah
there
you
go
close
to
50
000
comments,
and
you
know
all
these
many
forks,
so
it
is
a
living
breathing
project
with
over
over
1
100
contributors
to
the
main
julia
project.
A
A
A
One
of
the
things
I
wanted
to
do
for
this
talk
was
that
the
dig
back
into
my
own
first
commit
out
of
you
know
into
julian
turns
out.
That
was
in
september
of
2009,
where
we
had
where
my
first
commit
was
actually
to
add
some
of
the
array
array
operations
into
julia,
and
it's
it's
just
all
looks
so
different.
I
mean
we
didn't
even
have
jl
extensions
back,
then
they
were
dot
j
and
then
you
can
sort
of
see
how
the
project's
going.
A
We
added
the
plus
method
right
there
implemented
edition
for
double
scalars
and-
and
you
know,
we're
off
to
the
races.
One
of
the
things
I
like
to
highlight
about
julia
is
that
we
started
out.
You
know.
I
personally
started
out
my
work
and
my
contributions
to
julia
in
india,
and
I
I
wrote
a
majority
of
my
julia
code
out
of
india
for
the
first
several
years.
This
is
a
picture
of
many
of
the
early
contributors
to
julia
from
julia
corn.
A
Indian
indian,
open
source
contributors
have
had
an
outsize
impact
on
the
julia
language.
Apart
from
some
of
the
people
I
have
here,
I
should
also
mention
that
contributors
from
india
are
are
a
part
of
the
google
summer
of
core
project
every
year.
At
least
you
know
anywhere
between
one
third
and
one
half
of
the
contributors
that
we
that
we
mentor
in
the
google
summer.
A
Of
course,
program
are
actually
out
of
indian
universities,
and
so
it's
not
only
sort
of
people
contributing
to
julia
from
the
industry,
but
we
have
a
lot
of
people
from
indian
universities
undergraduates,
especially
that
are
contributing
to
julia,
and
the
community
has
seen
exponential
growth.
It
has
been.
You
know:
we've
seen,
25
million
downloads,
so
far,
they've
been
compounding
year
on
year,
tons
and
tons
of
youtube
views
check
out
alan
edelman's,
computational
thinking,
class
computational
thinking.mit.edu,
which
I
think
is
the
best
way
to
learn
julia
and
some
of
the
scientific
machine.
A
Learning
examples
that
I
showed
before
we
have
a
forum
views
on.
We
have.
We
use
this
course
for
our
forums.
You
know
the
number
of
news
articles,
the
number
of
julia
packages,
5000
julia
packages
per
day,
and
it
has
been
a
phenomenal
journey.
Not
only
can
you
call
these
5000
packages
from
julia,
but
you
can
also
call
every
python
and
our
package
and
every
cn
fortran
library,
every
java
library
ever
written.
A
You
pretty
much
do
not
have
to
do
any
boilerplate
to
call
them.
Julia
has
now
entered
the
mainstream,
we
were
number
19
on
ieee
spectrum
and
in
2020
we
went
from
47
to
23
on
the
tyobi
index.
Some
of
these
indexes
are
very
noisy,
but
you
know
directionally
julia
has
been
doing
really
well
on
many
measures
of
popularity
of
programming
languages
as
flawed
as
they
all
are.
I
think
they
do
pitch,
they
do
sort
of
paint
a
long-term
picture.
A
I
think
compared
to
many
open
source
languages,
because
julia
was
developed
on
github
from
a
very
early
stage,
and
so
the
github
community
structure
match
matches
and
maps
to
github
organizations.
It's
just
something
that
has
been
built
up
organically
over
time,
and
you
can
look
at
this
right.
There's
a
climate
org
there's
a
eco
julia,
which
is
for
ecology,
flux
ml
is
our
machine
learning
organization.
A
Julia
actions
is
where
a
lot
of
our
ci
work,
you
know,
and
the
github
actions
are
held
or
julia
bury
for
raspberry,
pi
or
turing
lang
here,
for
example,
for
probabilistic
programming,
julia
images
for
image
processing.
So
these
these
are
just
fantastic
organizations
that
that
we've
been
seeing
people
put
up,
and
I
just
love
all
the
all-
the
logos
that
people
have
come
up
with
which
are
sort
of
based
on
the
julia
logo
itself
and
the
colors
and
the
theme.
A
How
fast
is
julia?
It
is
really
fast.
This
is
the
benchmark
that
is
on
the
the
computer.
Language
benchmarks
came
on.
Debian
julia
is
in
the
top
five
with
c
rust
in
fortran.
You
have
julia
right
up
there
and
you
know
really
fast
with
graph
processing.
You
know
with
with
basic
data
science
machine
learning
loading
csv
files
turns
out
to
be
really
really
fast
in
julia
because
of
multi-threading,
and
some
of
our
data
frames.
Benchmarks
also
show
julia,
comparing
very
well
with
some
very
you
know
some
very
optimized
tools
like
spark
and
pandas.
A
So
so
julia
is
fast,
but
don't
take
my
word
for
it.
Try
it
out
yourself,
there's
a
growing
collection
of
julia
books
out
there
that
that
have
been
all
contributed
by
the
community.
Many
of
these
are
from
professors
at
stanford,
especially
linear,
algebra
and
optimization
are
two
of
my
favorite
books.
I
would
highly
recommend
looking
at
some
of
these
books.
A
B
A
One
of
the
most
interesting
things
about
julia,
I
think,
is
in
terms
of
user
experiences,
is
pluto
notebooks.
Of
course,
visual
studio
code
has
now
become
the
ide
of
choice
for
julia
users,
but
pluto
notebooks
have
been
developed
by
by
fonz
van
der
pla
from
amsterdam,
and-
and
this
is
a
reactive
lightweight
notebook.
A
So
jupyter
is
a
general
purpose.
Notebook
environment
that
is
a
language
agnostic,
whereas
pluto
is
a
julia,
specific
notebook,
which
means
that
it
is
reactive.
When
you
update
a
cell
out
there,
it
will
automatically
update
the
calculations
in
all
the
other
cells.
I
would
highly
recommend
checking
out
the
juliacon
2020
talk
by
fonz.
I
will
also
point
out
that
juliacon
2021
is
coming
up
very
soon.
It's
going
to
be
at
the
end
of
july.
A
Sign
up
at
juliacon.org,
sign
ups
are
free
and
the
conference
is
online
this
year.
So
all
this
is
great.
Now,
how
do
we
go
from
you
know,
building
a
community
building
out
julia
to
sort
of
you
know
solving
all
those
interesting
problems
that
I
talked
about
right
problems
that
the
world
is
facing
problems
in
you
know
in
climate
and
problems
you
know
in
in
combating
a
pandemic,
or
you
know
how?
How
do
you
cool
your
houses
when
the
weather
keeps
heating
up
or
the
climate
keeps
heating
up?
A
But
native
gpu
programming
in
julia
is
essential
for
the
julia
machine
learning
ecosystem
and
the
scientific
machine.
Learning
ecosystem
in
julia,
it's
been
used
for
tons
of
interesting
projects.
I
would
love
to
call
out
a
project
by
jonathan
jonathan,
laura
from
carnegie
mellon
university
or
called
alphazero.jl,
which
is
alpha
zero
implemented
in
just
a
few
hundred,
maybe
a
couple
of
thousands
of
lines
of
julia
and
it
you
know,
hear
more
about
it
at
juliacon
or
download
the
package
and
try
it
out.
A
So
all
of
this
comes
together.
The
gpu
capabilities,
the
flux,
machine,
learning,
capabilities
and
just
about
everything
in
julia
comes
together
in
you
know,
with
what
we
call
differentiable
programming
and
and
what
julia
offers
is
language-wide
differentiable
programming,
which
means
that
not
only
with
your
neural
networks,
but
any
arbitrary
julia
code
is
amenable
to
differentiation,
which
means
it's
amenable
to
be
put
into
a
machine
learning
loop.
A
Libraries
together,
so
that
you
can
apply
some
of
these
ideas
into
into
just
about
every
julia
program
ever
written
and
combined
your
data-driven
machine
learning
that
comes
out
of
deep
learning,
with
model
driven
scientific
learning
and
sort
of
get
the
best
of
both
worlds.
I
would
highly
recommend
looking
at
christopher
caucus's
talks
online,
where
he
talks
about.
You
know,
he's
the
author
of
julius
scientific
machine
learning
ecosystem
and
he
talks
about
all
these
different
applications.
A
If
you
just
look
at
look
him
up
on
youtube
applications
in
pharmaceuticals
in
engineering
and
chemistry
in
manufacturing,
in
batteries
in
climate,
this
is
this
is
how
the
ecosystem
comes
together,
and
this
is
why
I
personally
believe
that
a
programming
language
can
really
change
the
world.
It's
it's
all
built
by
the
julia
community.
It
builds
up
one
foundation,
you
know,
builds
up
from
the
foundation
one
step
at
a
time
and
goes
up
all
the
way
to
solving
amazing
and
interesting
problems
out
there.
A
All
the
way
up
to
you
know
solving
some
of
these
problems.
These
entire
ecosystems
are
being
put
together
on
julia
hub
so
that
they're
available
to
everyone
from
students
to
scientists
at
some
of
the
best
companies
out
there
at
you
know
just
just
put
it
all
on
their
fingertips.
That's
what
julia
hub
enables.
So
I
would
urge
you
all
to
try
it
out.
You
know,
try
it
out
at
juliahub.com
and
let
us
know
what
your
experience
is
out.
A
B
Wow,
okay,
that
was
you,
know
viraja,
who
has
gave
such
a
nice
overview
of
julia,
hey
viral.
You
know
we're
just
having
a
little
bit
of
you
know
an
early
holi
celebration
out
here.
So
all
of
these
colors
and
confetti
is
what
you
can
see,
but
thanks
a
lot
that
was,
that
was
a
really
nice
session
about
julia,
and
we
have
quite
some
questions
coming
in.
B
A
That's
a
very
that's,
a
very
good
question
actually
and
one
that
I
often
get
that
you
know:
what's:
what's
open
source
and
what's
not
and
and
how
does
an
open
source
company
make
money?
And
you
know
we
are,
we
are
committed
to
the
open
source
community.
That's
what
created
julia
and
the
company
exists,
to
sort
of
leverage
the
community
add
to
the
community
and-
and
just
sort
of
you
know,
grow
the
community
going
forward.
A
A
What
we
like
to
do
for
our
business
for
our
customers
is
build
a
cloud-based
software
as
a
service
product
which
is
julia
hub,
which
is
what
was
my
last
slide,
and
we
expect
that
by
providing
julia
hub
on
the
cloud
to
all
the
julia
users
out
there
in
in
enterprises
and
around
the
world
is
how
we'll
build
a
successful
business,
and
these
are.
These
are
very
well
aligned
incentives,
because
this
means
that
there
is
not
going
to
it's
not
an
open
core
model.
A
C
A
So
there's,
if
you
go
to
julia
lang.org,
which
is
the
julia
website,
or
if
you
just
google
julia
you,
you
know,
instead
of
getting
you
know,
actresses
named
julia,
you
nowadays
get
the
julia
language,
especially
if
you're
a
programmer
on
the
google
top
hit.
If
you
go
to
the
julia
website,
there's
a
learning
section
and
we
also
have
juliacademy.com,
which
are
a
free
online
courses
for
getting
started
with
julia.
B
That
is,
that
is
viral.
That's
gonna
be
probably
my
favorite
sentence
for
the
day
that
you
said
you
know
like
when
you
go
on
the
search
engine
and
search
for
julia
instead
of
actresses
you'll
find
the
open
source
programming
language.
That's
that's!
That's
really
my
best
code
for
today
you
know
so
that's
really
great.
So
coming
back,
there's
one
more
question.
So,
if
you
had
to
do
you
know
one
thing
differently:
building
the
julia
programming
language
today.
What
would
you
have
done.
A
Yeah,
that's
that's
a
very
hard
question.
Actually
what
what
would
we
have
done
differently
if,
if
we
were
to
build
julia
today
right-
and
I
I
I
mean
there-
are
it's
it's
hard
to
say
right
because
had
we
taken
a
slightly
different
path?
Would
things
have
been
been
different
in
some
way?
I,
to
be
honest,
you
know
it's
all
about
the
community
and
the
journey,
and
you
know
that's
sort
of
far
more
interesting
than
any
sort
of
technical
decision
around
the
language
itself.
A
If
anything,
I
think
you
know
what
what
I
would
love
to
see
happen
in
julia
is
you
know
some
of
the
decisions
that
if,
if
you
were
to
start
out
now,
there
are
just
so
many
more
mature,
open
source
language
frameworks
that
we
can
use
today
that
were
not
available
when
we
started
2009
llvm
had
just
come
out
back
then,
and
it
was.
It
was
not
what
it
is
today,
but
it
has.
A
B
Nice,
you
know
so
so
you
mentioned
I
like
how
you
mentioned.
You
know
it
was
it's
more
about
the
community
and
the
journey.
So
where
can?
Where
can
someone
go
and
then
you
know,
interact
with
the
julia
community,
find
peers
or
find
someone
really
to
help
them
grow
in
their
journey,
in
learning
the
language
and
kind
of
like
exploring
it
to
its
depth?.
A
Yeah,
if
so
so
again
on
julialang.org
you
have
the
community
section,
which
has
all
the
community
pointers.
We
use
discourse
for
our
mailing
list
and
discussion
forum.
So
discourse.julialang.org
is
a
very
vibrant
community
out
there.
We
also
have
the
julia
lang
slack,
which
has
over
5000
people
on
it,
and
it's
just
insane
so
that
that's
that's
a
good
way
for
interactive
communication,
but
a
lot
of
our
community
interactions
also
happen
on
github,
because
you
know
you
know
if
you
post
in
a
github
comment
and
someone
replies
to
you,
it's
almost
like
chat.
A
It
happens
in
real
time
and
a
lot
of
a
lot
of
those
commun.
A
lot
of
those
communications
actually
get.
You
know
archived
and
are
crawlable
by
google,
and
someone
else
can
come
a
few
months
later,
a
year
later,
join
into
the
discussion.
So
there's
there's
many
different
forums
for
four
people.
There's
chat,
there's
mailing
lists,
there's
github
and
all
of
them
are
are
phenomenal.
A
Also,
julia
packages
are
all
100
on
github.
Well,
maybe
not
100,
but
definitely
99
on
github
and
all
of
them
have
their
own
issue
trackers
on
github
as
well.
So
I
would
say
that
you
know
if
you're
engaging
with
the
community
there's
the
language,
but
then
there's
also
all
the
packages
and
and
a
lot
of
people
end
up
often
working
on
the
packages.