►
From YouTube: Release Generation from API
Description
Meeting with Runner & Release team to discuss how to generate a release in the YAML.
Issue: https://gitlab.com/gitlab-org/gitlab/issues/39097
A
B
A
The
idea
that
we
had
before
the
break
was
that
we
would
create
a
different
mechanism
for
generating
a
release
that
wasn't
from
the
rail
side
in
the
ammo
file,
but
it
also
wasn't
from
the
runner
side.
So
we've
created
the
sub
issue
of
our
parent
issue,
about
release
generation
from
the
API,
so
I
think,
where
we
left
off
last
time
was
Camille
added
a
bunch
of
feedback
with
some
code
snippets
about
what
we
should
be
instrumenting
and
Steve.
A
You
added
some
commentary
and
helped
filled
what
we,
what
we
should
be
implementing
as
our
MVP,
so
when
I
wanted
to
get
out
of
this
meeting
was
understanding
what
parts
should
runner
be
completing?
What
parts
should
the
release
team
be
completing
and
what's
realistic
for
things
in
12.8
and
12.9
sound
good.
C
C
Mean
you
sure
that
it's
at
this
comment?
Okay,
so
yes,
we've
got.
The
the
first
part
is
done
where
we
you
know
this
job
token.
That's
it's
all
completely
clear
to
me
and
so
then
all
right
so
now
extend
the
runner.
So
there's
a
there's,
a
template
to
add
an
additional
helpful
command.
So
I
guess
my
first
part
is
I.
Don't
really
understand
why
it's
called
a
helper
command,
and
why
is
it?
Is
this
somehow
it
outside
of
the
Rhino
reserve
yeah.
D
So
I
think
you
bring
up
a
really
good
find
out.
It
gives
me
really
good
how
to
say
ammunition
why
I
don't
think
it
should
be
in
the
runner,
because
it's
very
confusing
right,
like
you're
saying
what
is
this
helper
command
like
and
it's
not
just
you
who's
kind
of
asked
that
it's
most
people
are
gonna.
Ask
that
because
the
helper
command
is
just
a
command
we
have
for
the
runner.
That's
like
a
skinned
down
runner
binary
that
we
use
to
distribute
to
docker
images
and
things
like
that.
D
So
we
can
upload,
artifacts
and
and
doing
some
other
stuff,
so,
for
example,
using
the
docker
executors,
we
don't
just
use
one
container.
We
use
the
container
that
you
specify,
but
then
we
load
a
bunch
of
containers
around
it
that
are
used
to
have
get
installed
and
like
upload,
artifacts
and
things
like
that,
because
in
that
way
we
can't
assume
that
the
user
has
all
the
software
needed
to
upload
an
artifact.
So
we
control
that
and
that's
why
there's
ticket
LeBron
or
helper
command
that
that's
does
that.
D
Basically,
so
that's
the
idea
and
I'd
like
to
keep
that
like
behind
the
scenes,
or
only
technical
people
who
know
what
they're
doing
so.
That's
why,
if
the
comment
below,
if
you
look
at
the
comment
below
I
suggested
like
the
drawbacks,
if
we
have
it
and
the
hyper
image
and
if
that
one
or
an
aunt
abandon
CLI,
so
if
we
have
it
in
the
independent
CLI,
we
will
not
have.
Questions
like
you
just
are
like
okay,.
C
D
Is
this
helper
commands
doing
like
do
I
need
to
know
any
every
command,
because
if
we
do
release
upload,
for
example,
if
we
name
it
release
upload,
do
I
need
to
just
know
about
that
or
do
I
need
to
know
about
upload,
artifacts,
compress,
zip
files,
and
things
like
that.
Do
I
need
to
know
about
all
that
stuff.
So
it's
like
a
bit
of
leaking
abstraction
on
my
opinion,
but
that's
one
of
my
suggestions
on
ways
like
I.
Don't
know
what
you
think
about
no.
C
Well,
I
mean,
first
of
all,
when
you
guys
are
the
experts
first
of
all,
and
so
you
know
entirely
happy
to
defer
to
you,
although
I
really
would
like
to
understand
how
it
works,
but
I
guess
I
guess
sometimes
really
there's
two
things
that
are
really
completely
unclear
to
me.
Aside
from
that,
so
first
of
all,
I
mean
we've
got
this.
C
D
C
D
D
C
D
D
D
D
D
C
D
C
D
D
C
D
C
That'll
be
awesome,
so
so
so
we
do
this
transformation
of
the
of
the
mo
the
runner.
So
the
process
would
be
then
to
run
all
the
runner
CLI
or
whatever
it
ends
up
being
goes
and
picks
it
up
picks
up
the
configuration.
Then
it
actually
processes-
and
it
makes
another
set
of
API-
calls
using
an
existing
API
public
API
to
create
the
release
exactly.
D
So
the
way
it
would
work
is
going
more
into
technical
details.
You
define
a
step
right.
You
know
that
will
run
on
the
image
that
you
define
because
it
to
find
an
image
first,
and
then
we
will
take
that
step
definition.
The
runner
will
understand
it
and
say:
okay,
I'm
gonna
spawn
up
a
container,
for
example,
if
we're
using
the
docker
container,
docker
executors,
we're
gonna
container
and
run
this
command
in
that
container.
Now
this
command
will
pick
up
the
CI
job
token
automatically
and
upload
it
to
the
container.
C
C
C
D
Sense
yep
so
from
the
runners
side,
what
we
would
have
to
do,
as
we
do
not
understand,
understand
the
steps
fully
regarding
the
image
part,
so
we
would
have
to
do
that,
but
I
think
it
might
be
fine
without
the
MPC
I
would
have
to
investigate
fully
how
the
steps
are
made
to
like
give
your
salat
answer.
So
I
can
do
that.
Aysen.
D
A
D
C
D
D
A
D
We
will
have
some
stuff
already
out
of
the
box
like
picking
up
the
a
CI
job
token
and
things
like
that,
but
we
can
do
that
fairly
easily
as
well
right
now
by
extending
that
into
a
library
and
have
get
lap,
Runner
and
you're
still
CLI
to
import
that
library
through,
and
this
gives
you
another
benefit
right,
you're
completely
responsible
of
the
CLI
tool
and
not
well
here
and
abandoned
from
us.
Like
you,
don't
know
to
ban.
D
D
A
D
D
D
C
D
D
C
D
D
C
B
A
Go
and
maybe
this
could
be
a
go.
You
know
project
for
him
since
its
brand
new
anyways
right,
and
it
would
give
our
team
an
opportunity
to
kind
of
Komine
goal
with
with
Steve
to
understand
what,
like
some
of
the
learnings
but
I,
think
the
biggest
risk
that
I
see
right
now
is,
if
we
let
run
or
do
it,
we
don't
really
have
control
over.
C
A
A
The
protectability
of
our
scope,
I,
would
I
would
be
interested
in
seeing
what
we
can
do
on
the
release
side.
I
am
a
little
nervous
as
far
as
timeline,
though,
because
that
means
that
we
just
need
more
time
to
figure
out
like
what
the
heck
we're
doing,
because
we
don't
have
experience
necessarily
doing
like
a
CLI
creation
or
any
of
that
work.
C
C
Just
onboarding
at
the
moment
Frank,
and
so
you
know
so,
the
the
other
people
who
work
with
him
see
he's
got
great,
go
experience.
I,
don't
know
how
to
judge
that
myself,
I'd
certainly
be
able
to.
You
know:
I'd,
be
very
comfortable
working
through
the
rails
side
and
so
I
guess
this
con
log
free
pass
right.
C
C
But
maybe
if
we
could
ask
you
to
help
with
the
extraction
of
the
of
the
share
code
as
well
as
kind
of
just
maybe
just
be
available
to
consoled
a
bit
with
the
code-
and
you
know
you
know
know
that
we're
going
in
the
right
direction,
because
I'm
not
ago
developers
yet
and
and
mr.
Billy
just
started.
So
you.
D
Know
I
mean
at
the
end
of
the
day,
I
guess
it's
up
to
Darren
for
me
to
like
put
my
time,
but
I
can
act
as
kind
of
a
maintainer
for
the
whole
thing
like
just
guide
you
to
the
right
direction
and
tell
you
that
you're
going
well
or
like.
We
need
to
do
this
and
start
and
start
up
like
being
like
down
the
trenches
with
you.
Okay,
I
can't
talk
to
Darren
like
hey.
A
B
Gucci
yeah
yeah
good
time
for
full
on
questions
previously
say
say:
nothing
else
is
on
your
plate
right
now,
no
community
and
nothing
else,
and
you
were
doing
this
CLI
five
days
ten
days.
We
had
something
that
solid.
But
do
you
think
realistically
I
knew
an
expert,
but
the
runner
code
and
someone
could
go
as
well.
I
would.
B
Yeah,
we
definitely
needed
do
the
other
model,
then,
and
then
the
question
we
could
how
much
we'll
figure
out
the
timing
of
the
CT.
It
seemed
in
terms
of
your
resources
but
but
Jackie
and
show
on
to
cease
point.
What's
what's
gonna
happen
next
well,
my
Monday
is
that
we're
probably
gonna
be
clearing
the
runner
release
boards,
twelve
at
9:02
focus
on
the
next
big
thing,
so
we're
gonna
be
stretched
pretty
thin.
So
obviously
we
do
our
best
to
help.
B
A
D
B
A
That
sounds
good,
so
Sean
I
think
we
should
talk
to
John
Hampton
about
having
high,
maybe
in
this
I
I.
Think
from
up
from
a
go
perspective.
I
like
the
idea
of
having
someone,
who's,
natively
a
go
developer,
and
then
you
can
get
your
go
stuff
up
to
speed,
but
then
you
can
also
have
your
ownerships
like
a
pair
programming.
In
that
sense,
yeah.
A
D
A
Again
from
John's
view,
like
that's,
he
manages
the
team
so
we'll
talk
to
him
about
that.
I
do
feel
like
that
makes
me
a
little
bit
more
confident
in
you
know,
just
our
team
being
able
to
support
all
these
different
things
like
we
have
redundancy,
then
yeah.
You
and
I
can
both
support.
This
new
project
that
we
would
be
owning
as
a
part
of
a
maintenance
stream
is
the
last
thing.
I'd
want
would
be
you're
the
only
person
that's
responsible
for
everything,
with
release
generation,
which
is
what
we
have
right
now
pretty.
C
B
C
C
D
I
can
give
you
some
pointers.
I
was
just
looking
at
the
code
now
to
see
what
actually
needs
to
be
able
to
extract
it,
and
like
one
of
the
most
complex
stuff
would
be
I
would
say,
handling
the
error
codes
right
like
the
status
codes
from
the
API
like
right
now,
we
just
introduced
like
responding
to
rate
limits
and
things
like
that.
So
those
kind
of
stuff
kind
of
our
rudimentary
bus
needed
for
github.com
ski.
C
D
And
looking
things
they
don't
seem
very,
very
complicated.
I
mean
all
you
need
to
do
like
an
hight
side
is
just
pick
up
the
CI
job
token
from
the
environment,
which
is
just
one
method
and
have
dads,
send
its
to
an
end
point
and
then
responds
to
the
are
like
the
status
ecology,
get
from
the
response
and
I'm
defined
to
see
a
live
flex
to
be
properly.
Okay,.
C
D
C
Think
that
would
be,
if
that's
possible,
OD
an
awesome
user
table,
and
you
know
if
we
get
me
on
the
same
call
and
you
know
kind
of
all:
do
it
together
and
I.
B
C
A
Sean
you
and
I
can
talk
a
little
bit
about
the
direction
of
the
product
and
how
we
feel
like
this
should
be
timed
out.
I
think
there's
a
lot
of
risk
right
now
in
it.
So
I
don't
feel
very
comfortable
like
saying
that
this
is
well
that's
tied
to
my
maturity
plans,
but
I.
Just
don't
feel
like
saying
that
we
should
commit
to
delivering
it
in
the
next
two
milestones.
Given
that
I
don't
know
how
yeah.