►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
All
right,
yeah,
I'm
recording
this
to
the
cloud.
So
we've
gone
through
the
clustering
worker,
the
two
models
it
has
contributor
breadth
worker,
which
looks
at
for
all
the
people
who
contribute
my
project.
What
are
the
other
projects
they're
interested
in?
It's
the
contributor
repo
table
contributor
worker,
which
resolves
identity
for
all
the
different
emails
and
stuff?
That's
what
anuj
has
proposed
to
do
so
and
so
that
contributor
worker,
I
would
say
we
should
discuss
a
news
with
regards
to
the
contributor
worker.
We
should
probably
have
a
conversation
about
design
for
that.
A
So
that's
the
the
contributor
worker
we'll
we
can.
We
should
probably
set
up
some
time
to
talk
through
that,
but
I
think
we
want
to
think
about
like
I
can.
I
can
walk
you
through
the
design
for
it
as
it
stands
and
then
kind
of
what
we
would
prefer
it
to
be,
and
maybe
we
should
set
up.
Probably
I
should
set
up
one-on-one
time
with
each
of
you
this
week
just
to
make
sure
that
your
individual
projects
are
off
to
a
good
start.
Anuj
are
you
fo?
C
C
A
Okay,
I
can
go
over
that
with
you
and
then
there's
the
there's.
The
get
lab
workers
get
lab
issues,
worker
get
lab
to
issues
tables
and
then
there's
the
kit
lab
merge,
request
worker,
which
is
get
lab.
Two
pr
tables.
A
The
we
go
insights
tables
and
there's
like
this
merge
data
stuff
shouldn't
be
in
there.
A
A
And
then
there's
the
linux
badge
worker,
which
is
pretty
straightforward.
It
just
does
the
cia
badging,
which
is
very
low,
very
low
number
low
percentage
projects,
use
that,
but
it's
still
very
important
information
for
anything.
That's
infrastructure,
critical
and
then
the
message
insights,
worker.
C
D
D
A
A
A
And
the
pull
requests
analysis
worker
it
attempts
to
predict.
A
Based
on
a
priority
similarity
with
already
merged
and
projected
prs,
basically
that's
what
that
one
does.
I
think
training
of
this
model
needs
to
be
optimized,
I'm
just
like
putting
all
the
disclaimers
on
there
for
you
and
then
pull
request.
A
A
Messages
see
earlier
sketch
yeah,
it's
got
like
anything
with
pull
request
on.
It
is
probably
populated
by
that
or
one
of
the
machine
learning
workers,
and
it
should
be
obvious
14
the
release
worker.
A
Data
needs
a
git
lab
update
for
what
it's
worth
so
that,
for
example,.
A
We
use
this
to
validate
the
completeness
of
auger's
data
collection,
so
anything
that's
crawling
a
bunch
of
apis.
You
never
can
have
total
confidence.
That's
actually
got
all
the
data,
and
so
the
best
way
to
get
that
confidence
is
to
have
some
metadata
from
a
platform
that
tells
you
how
much
of
everything
you're
supposed
to
have,
and
so
we
just
essentially
run
those
tests
so
that
we
know
how
much
of
everything
we're
supposed
to
have
in
each
case.
A
A
So
with
that,
if
you
just
look
at
the
this,
is
the
repo
labor.
A
E
A
So
I'm
just
going
to
show
you
what
the
data
looks
like
in
here,
because
it
could
be
interesting
because
essentially
it
tells
you
what
repo
the
analysis
date
and
the
particular
file
in
question
file
name
gives
you
the
total
lines,
the
lines
of
code.
A
Blank
lines
move
your
faces
over
here
and
then
code
complexity
so
get
down
into
some,
not
trivial
file
sizes
here.
So
this
is
a
make
file
with
a
co
complexity
of
290,
which
is
pretty
high
stats,
dot
c
file
with
complexity
of
three
net.
So
you
can
see
that
there
is
a
substantial
variation
in
the
complexity
scores
based
on
how
the
algorithm
evaluates
the
complexity.
A
I'll
say
this
after
having
worked
with
it
for
a
year
a
few
years
now,
I
guess
maybe
three
years
that
it's
the
algorithm
is
consistent.
It's
not
right!
There's
no
canonical
truth
that
it
reveals
about
complexity
and
how
long
it
would
take
to
recreate
a
project
and
what
the
level
of
investment
might
be
estimated
at
for
a
project.
A
So
if
it's
consistently
overestimating.erl,
which
are,
I
don't
know
what
language
that
is
erlang,
so
if
it's,
you
know
if
it's
overestimating
the
complexity
of
erlang
it'll,
do
that
consistently,
but
it
does.
It
calculates
complexity
and
counts
lines
for
almost
every
language
that
I've
thrown
at
it.
A
A
And
so
that's
that
completes
the
worker
overview
of
what
we
have
in-house
right
now.
I
think
driven
yemen.
You
can
see
where
you
probably
need
to
go
for
data
anuge.
A
Probably
we
have
to
do
a
little
bit
more
discussion.
What
do
you
all
want
to
talk
about?
I
mean,
I
think,
probably
I
should
spend
a
little
time
checking
on
anuj's
setup
to
make
sure
he's
what
figure
out
what's
wrong
with
this
configuration,
but
dhruv
and
yeming
are
there
other
questions
you
have
before
I
dive
into
that
sort
of
technical
realm.
A
So
this
is,
this
is
the
so
yes,
so
there's
another
way
to
look
at
each
of
these
workers
is:
where
does
it
get
the
data.
A
A
Hublab
api
calls
to
get
activity
stream.
I
don't
know
those
for
each
contributor,
so
the
contributor
breadth
worker
is
taking
a
look
at
each
id
in
the
contributor
table
for
a
platform
and
make
an
api
call
to
get
the
activity
stream
for
that
user.
On
that
platform,
then
it's
populating
the
contributor
repo
table
and
from
that
you
can
determine
if
there's
a
major
contributor
to
your
project,
that's
starting
to
make
a
bunch
of
other
contributions
on
other
projects
and
that
sort
of
thing
which
can
be
useful,
especially
make
it
useful
for
a
lot
of
things.
A
A
Github
worker
data
sources
github
api
here
the
data
source
is
the
get
lab.
A
A
A
D
A
It
persists
it,
so
the
the
heaviest
lift
when
you
create
a
new
auger
instance
is
getting
the
initial
data
populated
and
for
facade
it
does
it
pretty
fast,
like
I've
done,
seven
thousand
repos
in
two
weeks,
like
counting
every
commit
on
a
server
for
the
initial
population
and
then
from
there
it's
just
looking
really
for
the
diffs,
so
it
goes
much
faster
once
you've
done
the
initial
population,
I've
also
had
it
where,
if
I'm
looking
at
scientific
repositories
where
they
store,
some
scientists
are
not
super
sophisticated
about
how
they
manage
their
computational
models,
and
so
there's
a
collection
of
them
that
use
github.
A
For
that
and
when
I
have
to
scan
repos
that
contain
very
large
computational
models,
you
can
imagine
that
the
diffs
and
those
diff
calculations
get
extremely
slow.
So
that's
just
like
if
you
find
facades
run.
If
you
try
to
do
facade
on
your
own
and
you
find
it's
really
slow,
is
it
like
if
it
won't,
if
it
for
less
than
a
thousand
repos
it
shouldn't?
Take
you
like,
at
the
outside,
assuming
some
kind
of
decent
computer
configuration.
A
I
would
say
the
likelihood
is
that
you
have
some
very
large
files,
some
very
probably
computational
models
that
somebody's
committed
to
the
repo,
and
I
just
think
that
get
tools
are
not
sophisticated
enough
yet
to
skip
those
or
to
know
how
to
treat
them
because
they've
been
put
into
a
version
control
system
so
shouldn't.
I
track
the
line
by
line
discrete
changes,
actually
no
they're
kind
of
binary,
but
git
doesn't
know
any
better,
so
it
counts
them.
A
A
E
F
I
was
under
some
chat
with
gonna,
go
back
and
assume
that
my
that
did
didn't
have
everything
installed
on
my
computer
already,
which
I
do
have
everything
installed,
but.
D
F
F
That
I
know
that,
as
of
last
month,
my
main
problem
was,
I
still
couldn't
get.
F
Dco
wasn't
working
for
me.
I
couldn't
do
it
on
the
web.
I
had
to
install
the
sit
on
to
the
github
onto
my
computer
and
then
still
didn't
work.
I
was
doing
a
thousand
different
things.
I
didn't
know
what
to
do.
We
did
wrong,
but
after
about
eight
hours
of
over
two
days,
I
just
gave
up
and
had
someone
else
sign
for
me.
A
It
what
what?
What
repo
you're
trying
to
make
commits
to.
B
F
F
Dissolved
now
it's,
it
was
okay,
diversity,
repo.
A
Okay,
when
stuff
like
that
happens,
if
you
get
stuck
on
sign
off-
and
this
could
happen
for
any
of
you-
you
can
message
me
and
I
can.
I
can
do
a
force
sign
off
for
you.
F
F
A
And
if
you're
the
other
thing
is,
it
works
the
same
way
as
reverting
commits
where,
if
you've
made
a
series
of
local
commits,
like
10
local
commits
or
whatever
in
the
course
of
fixing
something
it
wants
sign
off
on
every
commit,
and
so
it's
like,
then
you
have
to
go
back
and
do
the
sign
off
on
every
commit.
If
you
haven't
been
signing
off
all
the
way-
and
it
just
ends
up
being
a
lot
of
work.
A
So
sometimes
if
you've
got
a
lot
of
commits
in
something
before
you
realize
that
you
have
this
dco
issue,
I
can
go
in,
I
can
go
into
the
code.
I
can
go
into
github
and
fix
that
for
you,
you
know
the
first,
the
first
time
or
two,
because
otherwise,
with
all
these
chain
commits
that
you
have,
it
gets
to
be
this
tedious
nightmare
of
signing
off
on
each
one
of
them.
In
retrospect,.
F
So
hello,
google
center
of
code,
guys,
I
have
positive
things
to
tell
you
guys
about
google
summer
code
from
a
non-coder
perspective.
F
I'm
a
survey
guy-
and
I
just
did
a
survey
of
maintainers
and
my
survey
of
maintainers
says
that
mentoring
programs
is
the
best.
It's
the
third
best
way
to
increase
diversity
of
the
maintain
contributor
base
of
of
open
source
projects
out,
after
just
being
having
an
open,
welcoming
community
and
specifically
having
efforts
to
onboard
and
encourage
people
to
get
involved.
F
But
there's
it's
pretty
hard
to
actually
have
actual
programs
and
efforts
to
to
get
have
mentor
programs.
So
google
summer
code
is
good,
but
you
actually
have
to.
I
think
efforts
to
do
that
and
the
second
thing
is
actually
another
important
thing
is
actually
getting
non-coders
involved
and
from
that
perspective,
linux
foundation
hire
someone
to
be
in
charge
of
their
research
projects.
A
F
E
F
So
I'm
trying
to
deal
with
baby
steps
and
so
for
her
if
she
she
has
to
deal
with
multiple
things
but
yeah,
but
it's
just
just
in
terms
of
markup.
That's
going
to
be.
That
would
be
an
obstacle
for
her,
but
in
addition
to
that,
it's
the
virgin
control
and
the
other
thing.
So
what
I've
done
was
that
you
are
using
google
docs
now
for
to
track
for
for
commenting
and
then
what
I
was
doing
was
I
was
taking
the
major
changes
one
by
one.
F
E
F
Change
I
made
so
people
could
still
see
the
track
record
of
the
changes
that
were
being
made
compared
to
last
year
right,
but
that's
the
compromise
from
the
old
from
dealing
with
people
who
aren't
used
to
it
it,
but
still
having
a
track
record
and
letting
people
have
see
what's
going
on
and
and
still
using,
and
we
use
issues
to
some
extent
also
for
for
people
who
are
comfortable.
That
was
a
compromise
that
was
made.
A
That
seems
like
a
good
compromise.
It's
tough!
When
people
I
mean
I,
I
love
what
I
love
about
git
is
that
it
lets.
You
see
all
of
the
changes
line
by
line
it's
extremely
useful
and
it's
extremely
useful
for
survey
people
as
well,
because
you
can
see
what's
different
in
this
year's
survey
very
easily.
So.
F
It's
yes,
it's
and
it's
embarrassing.
When
I
have
to
make
comments
saying
this
question
didn't
work
because
I
didn't
code
it
in
surveymonkey
right,
so
we
didn't
actually
publish
the
data
because
I
didn't
do
it.
I
coded
I
miscode
didn't
enter
something
in
it's
trans,
radical
transparency.
E
A
F
You,
you
probably
you
know
this
but
yeah,
but
in
the
this
in
this
world
that
I'm
in
most
people,
you
could
hide
all
the
mistakes
you
make
if
you're,
if
you're
getting
paid
by
pr
people
or
vendors
or
et
cetera,
except
for
if
you're,
trying
to
work
by
the
same
standards
and
ideas
as
the
as
the
industry
and
the
people
that
you're,
covering
and
working
with.
F
A
No,
that's
that's
important
insight
for
this
team.
Yeah.
F
I
think
it's
it's.
A
F
Honestly,
I
could
get
over
running
today.
I
was
I'd
like
to
get
over
running.
Eventually,
I
don't
know
what
I
really.
F
Pretty
good
using
python
for
data
analysis
in
in
jupiter
workbooks
and
data
visualization
and
be
able
to
be
collaborate
with
people,
basically
be
a
manager
or
people
who
are
doing
data
analysis.
Okay.
F
Yes,
that's
honestly,
so
10
years
ago
I
used
to
be
managing
people
who
were
doing
the
analysis
for
me
using
spss,
because
I
had
that's
how
I
learned
how
to
do
data
analysis
years
ago
and
then
I
was
basically
managing
the
projects
and
I
knew
how
to
do
everything
myself,
but
then
I,
but
I
would-
and
I
would
go
in-
and
people
to
spot
check
and
fix
things
that
knew
how
to
do
everything.
F
F
And
I
and
you
have
used
all
the
google
stuff
like
get
big
query
and
everything
else,
but
yeah
I'm
not
I'm
not
good,
and
it's
not
good
enough
to
do
what
I
really
want
to
do
and
in
that
a
couple
years,
a
few
years
ago,
basically
I
said
I
was
doing
everything
and
then
I
said
well,
teamwork
and
hershey
are
doing
it
better
than
me.
So
I'm
going
to
stop
doing
it.
I'm.
F
Turkey
people
do
it
instead
of
me
yeah
and
I
so
I
I
still
think
that's
the
best
idea,
but
I
think
I
won't
be
able
to
step
in
and
be
able
to
do
things
ad
hoc
and.
F
And
it's
and
yeah
mostly
because
I
don't
want
to
rely
on
obviously
I've
got.
I
don't
want
to
rely
on
asking
everyone
else
to
answer
the
questions.
You
saw
the
big
thing
that,
like
there
was
a
big.
F
Oh
yo,
I
have
da
I
research
questions
I
want
to
answer
and,
and
I
don't
I
was
getting
tired
of
copying.
Philly
papa's
stuffed
syntax
on
big
for
bigquery,
for
example,
some.
D
A
F
Oh,
I
don't
even
you
can
think.
F
F
F
F
What
are
the
questions
that
are
coming
up
to
you
to
you
guys?
So
one
of
the
questions
that
I
heard
like
two
or
three
scout
in
the
chaos
weekly
meeting
was
music
was,
should
chaos
be
helping
with
consulting
projects.
F
Like
that's
like,
what's
the
use
case
for
chaos,
how
should
chaos
tools
being
be
used
should
be
overall
general
use,
asking
use
for
general,
very
big
questions
overall,
introductory
questions
or
should
be
used
for
as
a
for
more
in-depth
type
answer
answering
questions
and
that
more
in-depth
type
of
answers
is
what
I'm
interested
in
that's,
but
that's
also
not
that's
where
honestly,
that's
where
the
there's
consulting
money
to
be
made,
and
that's
not
so,
and
that's
where.
F
F
You
guys
present
the
findings
or
you
talk
about
the
problem
that
this
company
has
and
that's
how
you
solve
that
problem
of
you.
You
deal
with
that's
how
that
company
would
basically
get
solve
their
problem
and
give
you
guys
an
extra
fifty
thousand
dollars
to
deal
with
the
issue.
That's
that's
most
pressing
to
this
company.
It's
a
way
to
stop
to
be
able
to
to
raise
the
money
and
solve
the
problems
and
still
be
a
non-profit.
F
F
F
A
F
One
of
the
things
that
I've
been
writing
about
a
lot
is
like
where
people
are
doing
where
people
live
and
how
that
affects
their,
where
they're
contributing
and
one
of
the
biggest
and
basically
I've
been
writing
about,
are.
Are
people
based
in
china?
Are
people
based
in
south
asia,
so
literally
I've
written
about
israel?
Recently,
china,
who
are
the
top
contributors
in
china,
who
are
the
stack
overflow
just
got
bought
last
week.
F
Okay,
so
process
is
a
is
a
consumer,
is
a
invest
in
technology
companies.
Just
like
softbank
invests
in
a
bunch
of
technology
companies
and
softbank
owns
was
in
the
user
investor
in
yahoo,
japan
and
that's
the
deal
with
that.
So
process
process
isn't
owns
like
like
10
stake
in
mail.ru,
almost
30,
say
contestant,
which
is
like
mail
check,
wechat
and
a
bunch
of
other
things
qq.com.
F
I
don't
know
what
the
exact
how
to
pronounce
it
in
in
chinese
the
the
code
academy,
all
these
other
things,
but
they
bought
they
put
100
stake
in
stack
overflow.
F
What's
interesting,
there
is
that
in
december
a
indian
investor
had
put
a
head,
but
a
big
private
equity
stake
into
it,
and
so
they
got
now.
They
because
stack
overflow.
F
A
huge
percentage
of
their
users
are
south
asian,
okay
and
stock
overflow,
still
48th
largest
web
traffic
in
the
in
the
world
in
terms
of
internet
users,.
F
But
not
a
lot
of
check
users
from
china,
depending
on
which
way
you
look
at
it.
Where
am
I
going
with
this?
I
don't
really
know.
I
forgot
where
I
was
going
with
that.
A
Just
probably
answering
tool
answering
what
questions
can
chaos
answer?
It
sounds
like
certainly
stack.
Overflow
is
a
major
influencer
in
open
source,
because
we
all
search
it
to
figure
out
what
the
how
to
solve.
Surely
somebody's
encountered
this
error
before
and
inevitably
that
leads
me
back
to
stack
overflow
or
stack
exchange.
One
of
the
two.
C
D
F
It
together
in
terms
of
a
combining
the
where
the
traffic
comes
from
either
from
either
of
them,
yeah
and
and
basically,
what
that
thought
was
recently
is
that
I
don't
think
they've
actually
changed
their
api
since
2014.
F
Okay,
that's
maybe
I
was
wrong,
but
what
I
was
looking
at
over
the
weekend.
That's
the
terms
of
documentation
this
but
they've
changed
the
access
to
the
api
since
then
like
in
terms
of
what's
it
called.
A
F
So
stack
overflows,
creative
commons,
that
their
intellectual
property-
it's
supposedly
free,
but
you
can't
access
it
anymore.
So,
basically,
just
like
with
github
with
github
the
old
stuff
you
could
get
access
to,
but
the
new
stuff.
How
much
can
you
get
access
to
it's
it's
much
harder
to
get
access
to
the
old
to
the
new
stuff,
because
in
terms
of
how
fast
you
get
access
to
it,
you
know
more
than
I
do
about
this,
but.
A
I
mean
certainly
certainly
getting
the
data
downloaded
like
getting.
The
first
round
of
data,
for
a
collection
of
repositories
is
the
long
pole
in
any
auger
installation,
because
there's
just
a
lot,
that's
already
there
that
we
have
to
accumulate,
and
you
know
it
should
for
the
most
people
who
use
auger.
It
should
take
a
week
or
less,
and
you
know
I've
had
some
outlier
cases
as
I
was
mentioning.
I
think
when
you
came
on
the
call
or
just
you
know,.
F
So
stack
overflow,
it's
mostly
the
issue
is
that
most
of
the
good
data
is
only
so
most
70
76
to
80,
plus
percent
of
the
traffic
from
on
stack
overflow
is
from
search
itself.
Yeah.
Some
people
go
there
directly.
F
Not
that's,
unlike
anything
else
like
that,
and
so
that's
so
when
you
go
there
you're
not
right.
I
don't
know
the
answer
to
this
question.
I
want
to
know
it
when
you
can't
answer
this,
but
you
actually.
B
F
F
A
Yeah
yemen
drew
a
news:
why
don't
you
introduce
yourselves
and
kind
of
what
your
project
is
briefly
to
lawrence.
D
D
Dependencies,
basically,
we
can
there's
a
lot
we
can
make
out
from
dependencies
basic
it.
It
comes
with
its
own,
the
freshness
of
dependencies,
the
vulnerability
it
has,
if
you're,
using
it
the
risks
and
just
analyzing
just
making
a
resource
for
that
using
various
tools
like
scorecard
labias,
and
you
can
just
make
a
nice
a
proper
data
resource
which
could
be
like
hugely
helpful
in
terms
of
analyzing
dependencies
for
the
project.
Okay,.
F
I
I
know
that
for
analyzing
dependencies
I
would.
F
There,
I
would
basically
think
about
it
in
terms
of
I
don't
know
about
the
coding
aspect.
I
know
in
terms
of
thing
about
it
in
terms
of
what
people
could
do
with
the
information
you're
getting.
Can
they
take
action
based
off
of
it
once
you
know,
if
there's
like,
can
they
make
a
change
based
off
of
the
information
they
get
or
are
they
just
going
to
say?
F
Oh,
it's
a
risk,
I'm
scared,
or
can
they
actually
make
an
upgrade,
that's
or
change
in
license
or
those
those
are
type
of
things
that
are
important
to
be
able
to
do
just
being
able
to
pay
a
report
saying
that
there's
a
lot
of
risk
is
not
always
helpful.
A
Yeah
yeah
the
risk
working
group-
I
don't
know
if
you
followed
the
risk
working
groups
work
lawrence,
but
it
has
a
assembled
a
list
of
tools
and
minimum
viable
metrics.
F
I'll
look
at
that
I
haven't.
I
know
I
I
saw
the
people's
name
and
I
I
got
in
trouble
for
criticizing
the
the
google.
The
google
tool
that
came
out
last
week.
E
F
Mostly,
which
was
fine,
I'm
like
what
it's
basically
I
thought
it
was
what
I
wish
it
was
more.
That
was
what
my
way
was.
What
did
you
think.
A
I
just
heard
about
it
just
now,
which
means
nothing
other
than
I
just
heard
about
it.
Just
now,.
A
F
A
I
don't
know
like
I
don't
like
I
just
I've
just
been
introduced
to
this
concept
of
the
modern
repository
and
apparently
it's
it's,
not
just
google
github
uses
one
as
well,
and
I
don't
I
don't
have
a
clear.
I
don't
have
a
firm
grasp
of
what
that
might
might
be,
like
other
than
it
breaks
all
of
my
models
for
understanding
version
control
in
a
large
organization.
F
What
they
yeah,
what
I
might
take
away
is
basically
what
they're,
what
they've
created
is
pretty
good
for
a
visualization
of
the
dependencies.
F
That's
good
and
try
to
I
mean
to
debug
a
specific
application,
that's
good,
but
if
you're
an
app
developer
data
or
working
on
things
day-to-day,
if
you're
an
application
manager
day-to-day,
it's
not
something
you're
going
to
use,
but
if
you
need
something
for
a
toolkit,
that's
great
or
if
you're,
if
you're,
building
a
brand
new
application,
maybe
you
can
look
at
it
or,
if
you're
doing
audits
and
software,
maybe,
but
it's
not
something.
That's
gonna
be
used
day
to
day
the
versus
something
like
you've.
Seen
like.
F
What's
what's
integrated
into
what
linus
foundation's
integrated
into
it's
dashboards
for
his
projects,
with
sneak,
which
I
thought
that's
I
don't
sneak.
I
don't
know
how
they
got
a
deal
like
that
for
their
marketing
seems
to
be
a
little.
F
Not
there
do
you
know
what
I'm
talking
about
no
here
I'll,
be
fine.
The
link
now
you're
gonna
hear
my
ditching.
F
F
F
So,
basically,
all
the
different
projects.
He
has
different
metrics
involved
for
each
the
projects
and
you
have
to
log
into
the
project
to
review
it,
and
then
you
can
basically
and
they
scan
each
project.
Then,
if
you
log
in
use
the
sneak
software,
they
let
you
take
action
based
off
to
address
different
dependency
issues.
They
identify.
F
Interesting
and
that's
because
that's
the
state,
that's
what
they've
been
that's
their
core
business
value
prop
that
they've
been
selling
for
the
last
year
and
a
half
in
terms
of
their
software
composition,
analysis
product-
and
I
know
this
because
that's
my
job
is
doing
in
terms
of
competitive
intelligence.
It's
so.
Basically,
it's
a
little
bit
of
a
and
because
people
from
linux
foundation,
marketing
went
over
work
started
working
for
snakes.
So
I'm
like
just
how
that
happened.
C
Okay,
yeah,
like
I'm
doing
like
there
are
multiple
email
addresses
so
I'll,
have
to
identify
and
map
them
according
to
across
the
platform.
Whatever.
F
Okay
and
yemen
is
that,
can
I
pronounce
your
name
right.
B
My
my
research
lab
is
also
do
the
similar
things
like
chaos.
We
want
to
collect
some
data
from
github
or
vlab
and
we
use
this
data
to
create
some
model
and
to
analyze
this,
and
we
can
get
some
magic
to
help
companies
or
communities
yeah.
That's
what
my
lab
did
my
proposal,
something
about
my
proposal.
I
want
to
use
some
machine
learning
methods
to
do
the
similarity
and
to
do
the
similarity
prediction
and
the
the
risk
prediction
between
the
projects,
especially
I'm
I'm
so
interested
in
the
machine
learning
on
graph.
F
So
the
the
report
that
was
published
two
months
ago,
it
was
a
blog
on
the
chaos
website.
F
F
Lab,
yes-
and
I
don't
know-
is
that
a
chinese
name
but
frank
was
that,
is
he
the
guy
in
your
lab?
That
was
the
english
name
he
put
on
yeah
yeah,
that's
the
same
guy
I'm
trying
to
go
is.
E
A
F
Going
back,
but
so
to
the
the
github
site.
Yes,
here
we
go
xlab
2017,
open,
digger.
B
F
Yes,
this
is
great.
I
love
this
because.
E
F
I
more
than
I
spent
a
lot
of
time
looking
at
it.
If
I
I'll
throw
something
in.
F
E
F
F
So,
in
terms
of
questions
in
terms
of
sean,
there
was
a
lot
of
questions
here
that
I,
in
terms
of
what
they're
doing
in
terms
of
updating
things,
there's
a
lot
of
questions
that
we
could
have
answered.
They
could
have
answered
that
they
just
I
do
they
already
have
the
data
sets
to
answer
that,
but
they
had
they
had
different
questions
than
I
had
they
looked
at
china.
We
could
have
looked
at
india,
for
example,
and
they're
going
to
be
looking
at
things
next
year.
F
F
I
didn't
necessarily
like
all
the
ways
they
did
the
graph
analysis
or
actually
I
like
a
lot
of
the
ways
they
did
the
graph
analysis
and
it
was
more
math
that
I
understood
within
an
hour
of
reading,
so
I
would
have
had
to
read
a
lot
more
read
it
much
more
closely
to
understand
the
math
behind
it.
I
think-
or
maybe
I
would
never
understand
or
not,
but
I
and.
F
Yeah
so
I
mean,
but
I
unders
they
explained
it.
The
paper
explained
it
pretty
well
what
they
did.
I
just
was
not
well
enough
so
that
I
did
not
wasn't
going
to
try
to
be
a
phd
art
phd
counter
argument
against
it,
but
there
were
other
things
that
they
were
doing
that
were.
F
F
A
F
E
F
And
those
are
two
separate
issues
and
how
to
deal
with
spice,
because
a
lot
of
people
who
are
github
contributors.
F
Oh
yeah,
okay,
I
saw
it
and
with
this,
if
you've
never
seen
this
before
problem
with
this
database.
This
study
is
that
it
it
will
be
better
if
you
has
more
companies
in
it,
but
you
have
the
only
companies
that
it
looks
at
are.
Companies
that
are
are
nominated
to
go
into
this
into
their
database,
so
you
have,
to
add
more
companies,
add
their
name
into
their
gamma
file.
F
F
And
again,
this
is
why
I
like
chaos,
because
you
guys
actually
know
what
what
the
limitations
are
and
that's
why
I
keep
on
referring
everything
back
to
you
guys,
because
I
so
I
don't
have
to
be
the
person
who's,
always
shouting
about
the
limitations.
I
can
say
this
is
the
this
is
the
report.
This
is
this
is
great,
but
here's
a
limitation
and
if
you
guys
really
want
to
rely
on
something
talk
to
these
guys
because
they
spend
hours
on
dealing
with
it
all
the
time.
F
That's
that's
why
I've
been
advocating.
That's
why
I've
been
a
proponent
since
you
got
started
and
that
again
it's
awesome,
I
don't
want.
I
don't
want
to
obviously
have
to
do
the
ad-hoc
queries
to
find
all
the
mistakes
and
everyone
else's
work,
and
I
hate
seeing
the
report
saying
something's
wrong
and
then
I
know
it's
wrong.
A
No,
I
mean,
I
think
this
is
a
workshop.
Everybody
gets
to
talk
about
what
they
want
to
talk
about,
and
I've
got
a
better
idea
of
what
you're
working
on,
and
I
think
that
you
know.
I
think
that
auger
does
fill
a
nice
niche
for
the
kind
of
problem
you're
looking
to
solve
in
terms
of
gathering
a
lot
of
data
and
having
it
in
one
place
for
analysis.
A
I
think
the
questions
about
dependencies
and
the
questions
about
organizational
affiliation
are,
I
think,
going
to
be
they're
going
to
be
areas
where
we're
constantly
improving,
where
everyone
in
this
business
of
metrics
is
that's.
Building
software
is
going
to
have
to
be
constantly
improving,
because
neither
of
those
universes
will
ever
be
perfect
and
there's
a
lot
of
idiosyncrasy
and
data,
cleaning
and
mapping
that
go
into
effectively
recognizing
which
companies
are
contributing
and
effectively
recognizing.
F
I
think
that
yeah.
F
F
E
A
Yeah,
I
think
it
is,
I
think
I
don't
think
I
don't
think
apache
apache
has
not
given
up
on
their
their
source
control
apparatus.
I
forget
which,
which
one
it
is,
but
most
of
them
are
mirrored
on
on
github.
F
A
Think
get
lab
is
making
making
get
labs,
definitely
making
an
impact
right
now
I
mean
I
would
say
that
either
I'm
aware,
for
the
first
time
this
year
in
the
last
year
of
projects
that
are
moving
their
infrastructure
from
github
to
gitlab,
and
so
obviously
there's
some
kind
of
push
underway
and
that's
good.
F
I
have
statistics
I
don't
have.
I
don't
know
if
I
trust
the
statistics
in
terms
of
market
share,
so
but
it's
and
but
the
question
is,
is
it
what
are
the?
What
part
part
of
their
infrastructure
is
being
used?
Is
it
just
for
the
is
it
just
for
private
private
like
is
it
below
the
iceberg
type
of
code?
Is
it
like
just
the
public
the
code?
Is
it
like?
F
What
are
they
doing
with
it,
because
the
idea
is
basically
most
of
the
code
work.
The
government
work,
that's
done
is
not
being
done
in
public
still
so.
F
A
F
I
don't
know
what
to
do
with
it
either
I
got
it
so
the
idea
is
basically
thinking
about
in
terms
of
like
amazon
aws,
where
people
complain
about
aws
the.
So
the
question
would
be
is
this
is
related
to
a
kf
relate
to
a
chaos.
Question
is,
is
aws
contribute
back
enough
to
open
source?
That's
a
question.
F
F
Do
they
do
x,
y
z?
What
percentage
of
their
developers
are
participating
in
open
source
projects?
What
percentage
of
the
developers
time
are
is
being
involved
with
open
source
projects?
F
I,
if
you
look
at
it,
it's
like
three
percent
of
their
developers
are
contributors
to
open
source
projects.
I
I'm
making
that.
I
think
that's
what
it
was
a
couple
years
ago.
That's
pretty
damn
low
and
but
then
you
look
at
more.
You
say:
well,
that's
because
they're
spending,
most
of
their
time,
working
on
other
projects
totally
or
I
know,
containers
pretty
well.
F
Take
maybe
they're
working
on
projects
that
are
going
to
be
going
the
open
source
but
they're
working
on
them
internally
and
then
they're
going
to
release
the
whole
entire
project
to
the
public.
Eventually,
there's
different
ways
to
look
at
that
and
there's
different
ways
to
analyze
that
and
that's
when
that
exact
example
is
what
docker
did
docker
was
basically
said.
We
don't
want
to
collaborate
with
everybody
until
we're
ready
to
release
it,
but
we're
going
to
release
it
to
the
public.
A
I
don't
see
how
like
at
the
very
beginning
we
could
not
have
like
fully
opens
like.
We
were
open
source.
We
were
on
public
repo,
but
we
were
pretty
low
profile
and
just
building
it
up,
and
that
was
a
pretty
small
team.
A
All
things
considered-
and
I
think,
that's
almost
necessary
to
get
like
the
first
edition
of
something
out
the
door
like
if
you
try
to
open
source
the
thing
when
you're
still
defining
the
thing,
it's
really
hard
to
communicate
a
contribution
path,
and
I
mean
I
think
I
think
where
we
are
now,
especially
with
this
summer
google
code
we've
got
three
developers
who
I
think
understand:
github,
understand
auger
and
have
identified
some
gaps
that
we
can
close
in
the
in
the
feature.
Robustness
matrix
related
to
these
installation
networks,
risk
dependencies.
F
That's
a
good
lesson
learned
for
projects
that
are
being
launched
through
foundations
or
startups.
Even
if
you,
when
the
pr
person
makes
a
lot
of
noise
to
announce,
make
an
announcement,
there's
going
to
be
a
big
lag
time
and
there's
going
to
be
a
lot
of
expectations
and
you
just
have
to
basically
shut
those
out.
A
Yeah
now
I've
seen
I've
seen
projects
and
I
won't
say
which
foundations
but
let's
say
some
number
of
foundations
that
support
open
source
software,
have
projects
that
are
open
sourced
and
have
been
sort
of
shepherded
by
those
foundations
through
development
and
what's
ended
up
happening
in
a
couple
of
pretty
high
profile.
Cases
in
my
experience
is
that
somebody
takes
that
open
source
code
and
then
builds
pieces
on
top
of
it.
A
A
Is
there
like
actual
support
for
being
able
to
run
one
of
these
things
that
this
platform
is
supposed
to
run
for
me,
or
am
I
forever
locked
out,
because
I
don't
have
access
to
some
proprietary
service
that
gives
me
a
certificate
or
authorizes
my
my
my
tokens
and
so
yeah.
I
mean
there's
a
lot
of
there's
a
lot
of
sort
of
weird
open
source
stuff.
F
All
right,
well,
I'm
gonna
get
going.
I.
A
Yeah
this
time
is
over,
but
I
would
like
to.
I
would
like
to
set
up
some
time
with
the
google
summer
of
code
students
for
tomorrow
or
later
today,.
F
You
do
that
I'm
gonna
hang
up
I'm
going
to
try
to
catch
up
with
the
rest
of
the
chaos
people
later
I'm
and
I'm
just
and
I'm
busy
hyping
some
surveys.
I
did
because
I
did
something
I
released
my
maintainer
survey
a
couple
days
ago.