►
From YouTube: CDEvents Working Group (EMEA/AMERICAS) - March 21, 2023
Description
For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/
B
A
B
A
D
A
D
Yeah
not
too
shabby
you've
been
been
pretty
busy,
so
yeah
I
saw
I
saw
quite
a
few
updates
that
link
proposal
so
excited
to
get
to
discuss
that.
E
A
All
right:
well,
it's
free
pass.
Let's
get
started
so
welcome
everyone
to
the
City
events
working
group.
My
name
is
Andrea
Frito
I
work
for
IBM
I'm
in
UTC
and,
as
usual,
feel
free
to
add
yourself
to
the
for
this
event.
List.
A
Right
so
the
first
thing
I
want
to
to
bring
up.
They
have
an
agenda.
Today
is
the
released
0.2,
hi
Evan
we're
just
starting.
A
Yeah
so
that
that
was
due
yesterday
actually,
but
it's
yeah,
almost
everything
is
almost
ready.
I
would
say
so
all
the
PRS
that
we
had
planned
for
the
specific
repo
are
merged.
There's
a
new
one,
I
added
today,
which
would
be
nice
to
have
I
think
and
it's
about
adding
example.
Files
for
all
the
events
do
I
use
the
go
SDK
to
to
help
generate
this
and
basically
yeah.
F
A
A
A
B
A
Yes,
yes,
so
in
the
in
the
SDK
I
started
using
the
the
schemas
directly
from
the
spec
repo
and
the
plan
I've
not
got
to
it
yet,
but
I
plan
to
start
consuming
those
examples
directly
as
well.
B
Yeah
I
was
not
thinking
that
we
run
since
you
know,
generated
examples
from
the
go
SDK
should
we
continue
doing
that
if
we
want
to
update
the
examples
so
that
we,
we
cannot
update
examples
here
explicitly.
We
need
to
go
to
the
go
SDK,
to
update
the
examples
or,
or
is
it
okay
that
they
somehow
damage.
B
If
you
guess
what
I
mean
well
actually,
what
I'm
after
I
guess,
if,
if
we
should
have
them
in
sync,
and
we
should
take
it
from
the
go
SDK,
we
should
probably
write
some
documentation
about
that
here.
That
please
don't
update
the
examples
explicitly,
but
instead
regenerate
them
from
an
updated,
go
SDK
test
case
or.
A
Whatever
right
yeah,
my
my
plan
was
actually
to
use
the
one
from
this
package
for
the
test
cases
in
the
SDK,
because
then
we
can
use
them
for
all
the
sdks.
As.
A
A
Okay,
yeah,
so,
but
apart
from
apart
from
from
this
PR,
everything
else
is
in
ready
for
the
0.2.
A
So
all
the
schema
IDs
point
to
the
0.2
version
yeah,
so
everything
else
is
ready
and
for
the
website,
I
have
a
PR
App,
which
is
also
ready
and
I've,
been
reviewed,
thanks
ml
for
the
review
there
as
well,
and
so
once
these
two
were
merged.
A
A
So
in
terms
of
calling
SDK
I,
what
I
did
I
added
support
for
multiple
versions
of
videos?
The
way
we
discussed
last
time,
so
the
SDK
now
always
produces
the
last
version.
That
is
so.
If
I'm
using
SDK
version
tier
one,
it
will
produce
at
zero
one
if
it's
zero,
two,
it
repeats
zero,
two
and
so
forth
and
when
consuming
event,
it's
able
to
parse
both
future
and
old
versions,
as
long
as
they
are
compatible
so
different
patch
level
or
different
minor
level
in
case
of
different
minor
level.
A
A
F
A
F
B
We
missed
the
tagging
operation
in
the
checklist
right
now,
so
I
was
thinking.
If
I
should
just
add
it.
Is
there
some
other
things?
The
issue
one-on-one?
If
you
would
open
up.
B
B
Yeah
and
I
will
stay
to
prepare
release.
Announcements
should
be
maybe
I
mean
we
don't
have
a
a
process
document
where
we
State
how
to
release
the
protocol.
So
I
guess
this
is
more
or
less
the
process
we
follow
and
we
copy
this
from
every
time.
So
we
should
probably
also
include
a
statement
on
announcing
it
on
slack
and
on
mail
mail
lists.
A
F
B
F
B
B
F
A
0.1
because
it
was
the
first
release,
yeah,
yeah
and
I-
think
we
plan
to
announce
0.3
a
silicon,
probably.
C
A
We
will
do
it
for
for
that
one
as
well.
We
might
do
it
for
this
one
as
well,
but
I
don't
know.
B
G
May
I
ask:
have
you
all
thought
about
doing
GitHub
releases,
so
I
was
trying
to
research
how
to
automate
the
change
log
and
the
Jenkins
CI
Helm
charts
I've
been
kind
of
just
slowly
trying
to
help
out
there
very
slowly,
but
they
do
GitHub
releases
and
I
was
looking
at
their
GitHub
actions
trying
to
automate
this
through
the
python
SDK
I.
Don't
know
if
y'all
thought
about
doing
that
where
people
can
basically
sign
up
for
notifications
on
GitHub
releases.
A
Yeah,
we
definitely
one
of
them
to
use
skid-up
reviews,
so
we
we
use
them
so
we
we
have
exist.
The
current
releases
on
that.
So
we
plan
to
to
continue
in
that
for
the
go
SDK
What,
I've
Done.
A
A
All
right,
so
what
I've
been
doing
for
for
the
SDK
is
basically
I.
I.
Do
the
tag
manually
because
I
want
to
I
prefer
to
do
it
from
the
command
line,
and
then,
when
you
go
in
in
releases
in
the
draft
a
new
release,
you
can
pick
a
tag
and
you
can
pick
the
one
that
you
created.
A
A
B
G
A
A
Basically,
once
all
this
is
done,
move
main
onto
0.3
draft
so
that
anything
new
that
goes
in
terms
of
PR
is
considered
part
of
the
new
release.
A
A
All
right!
Well,
if
we,
if
we
can
merge
this,
the
the
spec
PR
today
tomorrow,
once
that's
merged
I,
can't
go
through
there.
The
rest
of
the
other
process
I'm
happy
to
do
that.
Unless
someone
wants
to
a.
A
Okay,
so
what
I
can
do
once
I
go
for
the
process,
then
I'll
make
sure
if
there
is
any
extra
step
that
I
eat
here
that
I
find
I.
Let
them
to
the
list
of
the
pair
document,
then
for
that.
G
Yeah,
if
you
do
need
help,
let
me
know
I've
just
been
trying
to
focus
on
the
python
SDK
so
I.
If
you
need
help
on
that
and
that's
high
priority.
Just
let
me
know
I
can
spend
I
have
about
an
hour
or
so
each
evening
I'm
trying
to
commit,
but
I've
just
been
focusing
on
python
SDK.
So
let
me
know
what
needs
to
be
moved
in
priority
level.
A
Cool
thanks,
yeah
I,
think
well,
python
is
the
case.
The
next
stop
being
an
agenda.
Maybe
I
think
it
will
be
great
actually
to
to
move
along
on
the
python.
Sdk
I
mean
if
we,
if
you
think
we
can
get
in
a
state
where
we
could
release
the
the
python
SDK
on
the
0.2
as
well.
In
some
time
not
to
that
will
be
great
because
that'll
be
able
to
because.
G
G
So
I
have
the
CI
linteen
working
in
my
local
I
want
to
get
some
I'm
doing
everything
through
like
Fidelity
to
get
input
from
people
to
see,
but
I
have
the
CI
Lin
team
working
using
flake
8,
basically
just
checking,
and
if
there
is
a
format
issue,
so
I
ignored
it's
like
eight
once
it
over
79
characters
it
doesn't
like
and
then
Imports,
but
other
than
that
it
is
running
per
push
and
purple
request.
G
So
I've
got
the
CI
Island
team
good
to
go
for
the
releasing
this
to
the
python
packaging
it
releasing
it.
So
people
can
install
it
through
pip
someone,
maybe
just
do
a
a
Oktoberfest
already
has
the
framework
there,
but
it's
it
I
think
it
was
maybe
copied
from
a
generic
template.
So
you
need
to
go
in
customize
that
and
then
we'll
have
through
pull
requests
the
ability
to
have
action
to
automatically
deploy
that
we
need
one
or
two
files.
G
One's
called
the
toml
file,
which
is
just
kind
of
like
a
pom
file
for
Java,
but
it's
kind
of
metadata
about
the
package
and
dependencies
that
need
to
be
involved.
So
I've
started
working
on
that
I
got
from
Eric
the
contributors
list,
so
there's
three
of
us
in
total,
so
I've
started
outlining
that
just
haven't
pushed
it
yet,
because
I
also
wanted
to
look
at
changelog
automation.
G
So
that's
kind
of
where
my
time
has
been
hope
that
helps
so
I'll
have
those
pull
requests,
probably
in
the
hopefully
the
coming
week,
I've
kind
of
been
all
over
the
place
reading
through
stuff
had
to
fix
one
issue,
one
of
the
CI
GitHub
checks
or
GitHub
action.
Sorry,
it
was
failing,
so
I
had
to
go
fix
that
that
was
fun.
I
got
to
learn
about
GitHub
actions.
So
that's
where
I've
been
putting
my
time
for
the
python
SDK.
A
Okay,
yeah,
let
me
know
if
there
is
anything
that
you
need
for
me,
happy
to
travel,
look
as
well:
okay,
yeah.
G
A
And
in
terms
of
the
what
the
SDK
does,
do
you
think
it's
it's
in
a
good
State,
we
could
do
like
a
0.1
release
once
did
all
these
things
are.
G
Ready
or
absolutely
I
think
Eric
did
an
amazing
job.
I
need
to
review
its
comparison
to
other
sdks,
but
it
looks
really
good.
He
did
he's
a
great
coder,
so
I
think
it
would
be
good.
I
just
want
to
get
some
of
the
Automation
and
get
it
cleaned
up.
A
A
Or
moving
on
no
okay
next
day,
I
just
put
a
placeholder
here
for
testing
event.
I
will
ensure
whether
all
about
John
today
I
guess
the
existing
PR
on
testing
events
will
need
to
be
some
rebased
based
on
the
0.2
updates
but
yeah.
Otherwise,
hopefully
we
can
get
that
in
put
0.3.
A
Then
I
add
tools,
Integrations,
so
Spinnaker
and
Jenkins
I,
don't
know.
If
anyone
has
any
debate
on
either
of
these
two,
they
would
like
to
share.
D
Nothing
from
my
end,
I'll
I'll
reach
out
to
the
Spinnaker
folks
in
terms
of
the
pr
for
me,
it
looks
good
I
just
need
to
get
one
last
set
of
eyes
on
it
and
then
once
preferably
like
Cameron
or
or
Jason
once
they
get.
D
You
know
once
that
gets
approved,
we'll
get
two
thumbs
up
and
and
that
can
be
merged
in
I'll
ping
them
after
after
this
Sig
meeting
and
then
and
then
jillander
I
know,
was
asking
about
implementing
the
the
actual
event
type
in
in
Spinnaker
and
I
I
believe
you
can
get
started
on
it.
It's
I,
don't
think
that
RFC
is
going
to
change
too
much
based
based
off
the
last
couple
weeks
here
from
feedback.
C
Yeah
bank's
been
like,
if
you
let
me
know
like
if
any
approval
is
there
from
other,
like
other
folks
from
Spinnaker,
so
that'll
be
great,
like
meanwhile
I'm,
also
starting
on
the
actual
implementation,
like
like
the
APA
and
back
and
part
so
that'll
that'll
helpful
to
approve
the
pr
yeah.
Okay.
D
Cool
cool
yeah,
oh,
like
I,
said
I'll
I'll
message:
Cameron
he's
he's
the
best
person
to
get
someone
or
something
merged
in
and
then,
if
there's
any
last
issues
you
know
like
that
will
be
visible
in
the
GitHub.
D
Well
PR
itself,
but
yeah
like
I,
said
I
I,
think
I
think
we're
at
a
pretty
good
place:
I
I,
wanna,
I
wanna,
you
know
I,
wouldn't
hesitate
just
to
get
started
on
it
and
then,
if
you
need
any
help
outside
you
know,
just
like
you
know,
with
Spinnaker
or
whatnot
I,
don't
know
how
familiar
you
are
with
Spinnaker.
I
can
definitely
help
you.
There.
A
Great
thanks,
Ben
and
jalander.
So
do
you
guys
plan
on
being
in
Vancouver
by
the
way
in
the
city
hall.
D
It's
still
up
in
the
air
for
me:
travel's
restricted
for
for
Apple,
so
it's
still
up
in
the
air
I
might
I
might
try
to
just
go
myself.
Just
you
know
out
of
my
own
pocket,
but
but
we'll
see
we'll
see
I'll
I'll,
let
you
know.
F
C
No
you're
not
planning
to
travel,
so
some
restriction
on
travel
again
here
not
be
yeah.
Any
plans
to
travel
yeah.
C
Yeah
so
we
had
actually
accepted
dog,
wrote,
CD
events
but
yeah
so
due
to
travel
restrictions,
we're
not
able
to
make
that
happen.
So.
F
B
A
No
I
I
asked
whether
it
would
be
possible
to
do
like
remote
presentations
and
I
was
told.
No,
that's
not
an
option.
I.
B
B
A
Like
yeah
I,
don't
know,
let
me
let
me
go
and
check
with
fatty
yeah.
A
Right
yeah,
so
if
it's
great
that
this
Pinnacle
work
is
moving
along,
I
mean
if
it's
ready
in
time,
it
would
be
great
to
you
know
to
announce
it
in
the
at
the
event
of
silicon
in
some
format,
and
that's
why
I
was
asking
well,
you
guys
were
planning
on
being
there,
because
you
said
making
this
happen,
it
would
be
would
be
great
yeah.
So,
let's
see
what
happens
again,.
D
Yeah,
let
me
I'll
I'll
I'll
get
an
answer
for
you
by
the
end
of
this
week,
Andrea
and
then
we
could
figure
out
what
we
can
do
from
there.
If
that
works,
for
you,
yeah.
A
All
right,
Evan,
sorry
to
put
you
on
the
spot,
I
was
wondering:
do
you
have
any
update
or
do
you
know
if
anything
happened
on
the
Jenkins
plugin
by
any
chance,
yeah.
G
So
I
started
I
think
it
was
a
few
weeks
ago.
Maybe
two
weeks
started
with
the
initial
kind
of
plug-in
framework
for
Jenkins
started
with
they
have
a
really
nice
job
listener.
For
the
you
know,
the
one
to
start
first
off
with
the
pipeline
run,
started
queued
and
finished.
So
I
wrote
the
class
to
collect
the
pipeline
started
kind
of
referencing
the
cloud
events
plugin
for
Jenkins.
G
It
was
very
well
written,
wrote
that
we
have
for
our
kind
of
first
iteration
I
did
two
sync
connectors
one
two
Kinesis
one
to
just
an
HTTP
endpoint,
so
kind
of
got
that
framework
working
and
then
wanted
to
have
my
awesome.
G
Teammate
Rory
work
on
the
pipeline
run
queued
and
finished
basically
collecting
the
data
and
have
him
go
through
it
because
I'm
more
of
a
devops
person
he's
a
job
application
person
very
awesome
at
that
stuff
and
he
reading
through
the
classes,
kind
of
making
changes,
but
he's
been
working
on
the
queued
and
the
pipeline
finished
and
doing
great.
A
So
is
there,
do
you
know,
is
there
any
probably
repo
where
we
can
follow
progress
on
these,
or
is
that
worked
in
on
internally
instability
for.
G
Now
it
it's
internal
Jamie's,
gonna
lead
the
release
in
the
presentation.
I
hope
that's,
okay,
I'm
happy
to
give
updates,
but
that
that
whole
part,
Jamie
and
Jerry
are
leading
so
yeah.
F
A
A
All
right,
the
last
item
that
I
have
an
agenda
is
the
conducting
events,
so
I
wasn't
sure
whether
European
and
ml
we're
going
to
join
today
but
you're
here.
So
we
can
discuss
about
this.
F
E
D
Yeah
I
think
I
think
so.
I
was
reading
Emil
your
your
section
that
you
wrote
for
the
the
tracing
context.
D
So
if
we
go
up
to
that
section,
if
you
have
right
here,
if
you
go
to
the
cons
yeah,
if
you
go
to
the
cons,
real
quick
I
really
like
this,
so
the
second
bullet
point,
it
says:
let's
less
flexibility
applied
on
on
events
levels,
but
it
could
be
instead
be
applied
on
a
CD
events
level
with
our
own
syntax
I.
Think
if
we're
gonna
go
with
tracing,
we
should
go
with
that
approach.
I
think
that
makes
more
sense,
because
the
reason
why
I
say
that
is
one
tracing
again.
D
Is
you
know
it's
it's
it's
a
huge,
it's
a
huge
project
so
and
it's
kind
of
its
own
thing.
So
if
we
make
CD
events
like
connecting
events,
I
want
connecting
events
to
also
be
its
own
thing.
Inside
of
CD
events,
I
don't
want
to
make
it
look
like
we,
we
kind
of
like
disbandaged
on
a
solution
or
you
know
like
a
different.
You
know
a
just.
D
You
know
piggybacking
on
onto
tracing,
which
is
completely
fine,
but,
like
I
think
you
know,
we
can
go
with
our
own
syntax
and
then
potentially
even
simplify
it,
because
I
don't
think
we'll
need
everything
from
tracing
I.
Think
we'll
probably
need
a
good.
Maybe,
like
you
know
a
good
portion
of
it,
but
I
definitely
don't
think
we'll
need
like
everything
like
one
we'll,
probably
not
need.
D
You
know
like
there's
this
concept
of
like
joining
spans
and-
and
you
know
like
that's
in
the
spec
and
and
there's
things
of
like
you
know
how
to
deal
with
like
128-bit
IDs
like
we
can
get
rid
of
all
that
right
and
just
focus
on
the
on
the
pieces
that
we
need
so
I
I.
Think
if,
if
we
go
with
the
tracing
approach,
I
I
think
I
think
going
with
a
simplified.
D
Our
own
syntax
solution
is
probably
the
the
the
best
approach
in
my
opinion,
so
yeah
I
wanted
to
get
your
thoughts
on
that
Emil
and
Andrea
four
of
five.
B
Worries
so
I
think
we
need
at
least
the
parent
child
relationship
if
we
should
go
for
the
tracing
tracing
part.
It
doesn't
make
too
much
sense
that
this
to
me,
if
we
just
have
a
global
ID,
where
all
events
use
the
same,
with
no
possibility
of
hierarchy
in
the
tracing
but
I
I'm.
Sorry
I'm
not
too
used
to
using
this
tracing
Concepts
but
I
might
be
wrong,
but.
D
Yeah
yeah,
we
definitely
would
and
I
think
that's
so
in
the
in
this
regard.
That's
what
spans
so
like
you
would
have
this
Trace
ID
and
then
you
would
have
span
IDs
and
span.
Ids
is
kind
of
like
how
it
references
cut,
no
kind
of
kind
of
parrot
yeah.
Well,
there's
a
parent,
ID,
I,
think
field,
also
in
in
tracing
so
yeah
yeah
I,
agree.
A
Yeah,
so
there
definitely
is
a
way
to
get
this
relationship,
and
you
can
have
I
mean
the
the
root
level.
Is
the
The
Craze
ideas
which
is
shared
across
all?
The
racing
eventually
goes
to
this
real
pressing
and
then
you
have
spans
and
you
can
have
child
span.
So
you
can
have
this
kind
of
three
type
of
structure.
F
D
And
there's
also
a
lot
of
Trace
form
formats
as
well
or
specifications,
so
there's
open
tracing
and
then
there's
W3
tracing,
there's
Zipkin
tracing.
So
if,
if
we
choose
like
well
I
mean
like
we're,
gonna
put
our
own
flavor
on
it
but,
like
you
know,
like
I,
said
just
to
be
aware,
there's
different
definitely
different
different
ways
of
handling
tracing
parent
and
child
relationships.
Depending
on
the
specification
that
you
go
with.
D
A
D
Yeah
Zipkin
has
their
own
format,
but
it
does
also
have
compatibility
with
open
tracing.
Most
things
have
compatibility
with
open
tracing.
A
A
A
We
can
have
this
parent
child
relationship
and
that's
what
could
be
used
for
use
cases
where
you
want
to
see
kind
of
like
a
distributed
transaction
or
workflow,
and
then
we
could
also
have
in
parallel
to
that
or
on
top
of
that
links,
because
links
then
could
be,
would
have
semantic
Associated.
So
we
could
have
different
type
of
things
similar
to
what
iPhone
does
and
those
could
be
used
also
to
link
events
across.
A
You
know
like
longer
running
or
not
wider,
spanning
workflows
like
supply
chain
scenario,
for
instance,
where
I
have
some
packages
that
are
filled
and
generates
events,
and
then
I
have
some
other
packages
which
depend
on
those.
In
my
they
might
preference
back
to
the
events
of
the
original
packages,
while
even
if
those
happen
in
different
places
different
times.
F
B
B
Say
a
bit
hard
to
know
which
which
which
way,
which
kind
should
you
use
in
different
scenarios?
Maybe,
but
maybe
we
can
have
examples
of
that
as
well
too
exemplify
when
when
one
is
preferred
over
the
other,
if
we
go
for
the
hybrid
approach,
but
maybe
before
we
go
with
I,
also
added
a
Concepts
section
to
this
document,
just
above
the
Alternatives
here.
F
F
B
Me
to
open
that
yeah
you
can
open
it.
So
there's
a
table
there,
for
example,
comparing
what
they
call
event.
Logging
and
software
tracing
in
this
and
I
think
that
that,
to
me
at
least
makes
quite
a
lot
of
sense
when
it
comes
to
events
versus
traces.
F
D
Is
very
granular
and
and
what
we're
looking
for
is
not
granular
we're
looking
for
at
an
event
level
right
and
that's
why
I
was
thinking
like.
Maybe
the
hybrid
approach
would
work
better
than
tracing
like
like
I
said,
we'll,
we'll
still
need
to
figure
out
tracing
but
tracing
is
not
a
simple
solution,
it's
very
complex
and
so
and
that's
why
I'm
I'm
leaning
towards
not
using
it
one.
It's
very
granular,
you
know,
that's
not
to
say
we
can
make
it
less.
D
You
know
less
granular,
but
but
I
think
that's
trying
to
beat
something
into
some
be
a
solution
into.
You
know
something
it
isn't
right
and
that's
what
I
was
thinking
like.
Maybe
if
we
go
with
the
either
a
hybrid
approach
or
a
different
approach
altogether
might
might
be
a
better
solution,
just
kind
of
starting
fresh,
starting
with
the
actual
use
cases
that
we
care
about,
as
opposed
to
you
know
saying:
oh,
we
have
all
these
granular
events.
D
How
do
we
make
this
into
a
single
or
sorry
granular
traces
to
a
match,
a
single
event?
So
that's
that's
kind
of
like
where
my
thinking
was
you
know
with
with
tracing,
but,
like
I
said,
it's
still
we're
still
in
the
early
stages
of
diving
into
how
tracing
can
be
used.
I
still
don't
know
yet
whether
or
not
I'm
against
it.
But
my
intuition
says
it's
probably
not
going
to
be
the
best
solution
for
us.
B
Yeah
I
think
there
might
be
certain
scenarios
where
it
couldn't
make
sense
to
have
the
tracing
object
in
the
event,
maybe
additionally
or
so,
but
I
I
think
that
the
The
Links,
at
least
to
me,
makes
more
sense
in
many
scenarios,
but
I
think.
As
you
say,
it's
good
to
look
into
the
use
cases
or
scenarios.
B
But
these
events
and
the
the
correlated
events
or
connected
events
to
understand
what
is
best
where
but
I
think
the
stable
area
is
at
least
start
some
thinking,
in
my
mind,
on
where
we
should
use
either
or
and
I've
actually
looked
into
it
to
flash
this
table
out
as
well
too,
to
more
distinguish
on,
because
we
have
the
same
problem
ourselves
within
Eric.
So
now,
but
I
mean
we
want
to
use
open,
Telemetry
traces
for
certain
areas.
Certain
data
objects
that
we
sent
on
our
buses
and
we
want
to
use
events
for
others.
B
But
we
haven't
really
come
up
with
a
good
way
to
distinguish
between
those
two
when
to
use
what's
what
kind
of
the
data
object
or
messaging
system,
because
for
sure
we
will
use
open
Telemetry
for
for,
like
more
server
health,
related
things
and
more
getting
the
status
of
the
system
so
say.
But
when
it
comes
to
the
pipeline
execution,
we
will
use
events
which
we
then,
in
our
case
at
least
use
links
for
today,
all
over.
But
still
we
want
to
reference
between
these
as
well.
B
D
Yeah
I
think
I
think
they
solve
similar
problems,
but
you
know
they
are.
They
are
very
different
in
that
all
so,
like
I
said,
tracing
is
more
granular
right,
so
tracing
you're,
gonna
care
about
you
know
like
method
to
Method
calls
you
know
class
to
class
calls.
You
know
things
of
that
nature,
as
as
events
are
more
of
a
higher
level.
D
Grouping
of
of
these
calls
right,
where
what
you
care
about
is
a
pipeline
run,
for
example,
or
a
test
being
started,
which
can
be,
you
know,
obviously
shown
or
traced
with
with
tracing
right,
or
you
know
like
that
that
flow,
but
we
only
need
that
that
kind
of
that
entry
point
and
that
the
the
closing
of
that
entry
point
if
that
right
or
yeah,
we
will
only
need
the
start
and
the
end
for
for
the
event
side
of
things.
D
So
so
yeah,
like
like
I,
said
in
my
opinion,
like
I,
think
a
lot
of
the
times.
You're,
probably
gonna
see
tracing
and
Eiffel
in
this
case,
pre-exists
in
the
in
the
same
code
or
like
in,
in
the
same
same
way
of
dealing
with
things,
I
think,
but
in
my
opinion,
if
you
find
yourself
with
iPhone,
Alexa
I'm,
not
I'm,
not
a
maintainer
or
anything
but
like
when
I
look
at
Eiffel.
This
is
this
is
how
I
Envision
it
being
used.
D
If,
if,
if
you
find
yourself
going
more
granular
with
Eiffel,
then
you
probably
want
to
start
looking
at
tracing?
That's
probably
that's
probably
how
I
would
think
about
it.
At
least.
F
B
D
That
makes
sense
yeah
and
then
one
thing
that
might
be
good,
though,
is
if
there
was.
It
was
a
way
to
relate
Eiffel
events.
Two
tracing
events
like
that
might
be
useful,
and
maybe
that
could
be
thought
about
some,
but
because
that
way,
if
you
have
a
Eiffel
event,
you
could
be
like.
Oh
I
can
easily
get
that
Trace
from
from
this
Eiffel
event,
somehow
right
and-
and
that
might
be
a
good
way
of
solving
solving
that
problem.
A
Yeah
yeah,
so
that's
so
thanks.
Thanks
for
adding
these
for
the
discussion
about
tracing
I,
I
also
agree
with
it.
I
don't
think
we
at
least
I
didn't
mean
to
to
propose
when
including
Chinese
context,
type
of
approach,
that
we
would
go
for
like
a
full-blown
racing
system
based
on
CD
events,
rather
to
take
inspiration
of
that
that
one
of
the
existing
standards
for
the
format
but
yeah
yeah.
So
this
has
been
there's
probably
more
features
than
than
we
need.
A
I'm
trying
to
think
how
we
we
move
forward
again
from
these
I
was
thinking.
Maybe
if
we
go
back
to
the
to
the
Dora
Matrix
example,
type
of
example
that
we
started
building,
we
could
for
that,
that
those
four
metrics
identify.
What
are
the
events
and
what
are
the
sources
of
those
events
and
what
kind
of
information
a
good
event
producer
have
and
see?
A
What
kind
of
connecting
would
help
us
better
when
I
know
that
use
case,
and
we
could
start
you
know,
maybe
10,
we
have
something
concrete
and
a
great
example
to
to
see
which,
which
approach
helps
us
to
reconnect
those
events
better
and
build
the
durametrics.
That
could
be
one
example.
That's.
B
A
good
example
of
your
case,
I
think,
yeah
and
I
think
as
well
that
we
we
could
then
quite
quickly
create
a
pull
request
for
adding
the
hybrid
solution.
I
mean
by
adding
both
the
possibilities,
type
Trace
context
and
links
in
the
event
for
now,
but
they
should
both
be
optional.
Then,
and
then
we,
when
we
go,
look
into
the
use
cases
and
do
hands-on
experience
with
them.
I
think
we
can
eventually-
or
maybe
they
do
both
variants.
B
For
some
of
the
use
cases
there
are
Dora,
for
example,
and
then
we
can
compare
it,
but
what
would
it
look
like
and
what
what
aspects
do
we
cover?
What
aspects
can
we
not
cover
with
this
type
of
connecting
events
or
references,
so
I
I
would
think
that
we
would
need
some
some
hands-on
experience
with
with
the
different
ways
of
or
connecting
events
and
the
way
to
do.
B
D
Yeah
and
with
the
hybrid
approach
like
I,
wanted
to
explain
like
the
the
difference
between
like
the
global
ID
and
the
links,
so
the
global
ID.
The
whole
purpose
of
it
is
just
knowing
like
how
to
query,
or
basically
just
allow
for
easy
querying
of
of
the
database
yeah.
D
So
it
allows
you
to
grab
like
basically
it
grabs
you
like
a
map
right
and
then
the
iPhone
links
is
like
the
interstates
and
the
highways
that
connects
like
all
these.
You
know
cities
and
whatnot.
So,
like
you
know,
the
global
ID
is
just
giving
you
that
big
picture,
and
then
you
need
to
to
understand
the
relationship
between,
like
all
the
nodes
is
done
with
iPhone.
So
Global
ID
gives
you
all
the
nodes,
but
gives
you
no
idea
of
like
how
they're
connected
then
you
use
Eiffel
to
then
get
those
connections.
F
B
D
B
D
F
A
A
Okay,
one
question
so
when
you
say
ml
that
we
should,
we
could
make
a
PR
with
the
hybrid
approach.
A
Do
you
make?
Do
you
mean
actually
extending
the
context
to
others,
Fields
or
sorry?
It
was
just
thinking.
We
could
also
use,
try
and
use
custom
data
and
maybe
some
temporary
kind
of
specification
of
like
specific
Fields
within
custom
data
that
we
could
use
so
sure.
F
D
Course,
yeah
I
think
yeah
I
definitely
think
it
should
live
in
custom
data
just
because,
like
it's
kind
of
an
up
in
the
air
solution
right
now,
and
then
also
on
top
of
that,
we
can
just
have
it
like.
Like
the
custom
data
just
be
like
you
know,
links
or
content
or
trace
or
spam
context,
whatever
you
want
to
call
it
right
and
then
just
have
everything
in
there
and
then
once
we
feel
like
we
can
move
span
context
out
of
custom
or
if
we
ever
do
decide
to
move
the
span.
D
A
Yeah,
okay,
so
basically
say
in
this
in
this
fact
that
we
have
this
experimental
fields
that
we
can
use
in
custom
data
to
just
give
them
a
name
so
that
we
can
work
with
those
and
yeah
okay.
A
B
D
It
might
be
good
just
to
have
like
some
sort
of
reference
to
it.
Just
so,
people
can
at
least
know
what
to
expect
when
they
see
the
span
context,
it's
just
more
for
the
end
user.
That
might
be
interested
getting
more
information
because,
like
you
said
since
it's
custom
data,
is
this
Loosely
defined
bag
of
stuff?
If
there's
no
documentation
anywhere,
it's
going
to
be
really
hard
for
people
to
understand
like
what
the
thing
they're
looking
at
is
right.
D
So
I
I
would
vote
like
to
have
a
PR
like
at
least
giving
some
guidance
to
users.
What
what
this
span
context
or
whatever
we
want
to
call.
It
is.
B
I'm
thinking,
if
it
is
really
custom
date
that
we
talk
about
the
life
is
it's
some
other
like
experimental
data,
but
maybe
we
could
have
any
custom
data
for
now.
I,
don't
know,
I,
don't
want
it
to
do
disturb.
B
So
I,
don't
think
that
at
least
long
term
the
protocol
shouldn't
declare
anything
within
custom
data,
but
for
sure,
for
now
we,
since
we
have
custom
data,
we
could
of
course
add
something
there
for
now,
but
I'll
remove
it
later
or
we
had
another
optional
object
next
to
custom
data,
which
we
call
like
experimental
data
or
something
all
right.
That
makes
sense.
Yeah.
D
C
B
A
Yeah,
okay,
yeah,
maybe
that's
and
the
simplest
approach,
because
we
are
still
anyways
in
zero
dot.
Something
version.
F
A
To
to
move
things
around
too
much,
I
guess:
yeah,
okay,
so
I'll
scratch
the
line
of
using
custom
data.
A
D
A
D
Think
so,
I
guess
what
my
question
is
now
so
we're
asking
for
a
separate
PR
now
for
the
spec
right.
Is
that
what
you're
asking
Andrea.
A
D
A
We
need
to
actually
make
a
proposal
a
concrete
proposal
for,
but
it
would
look
like
so
we
can
yeah,
we
can
discuss
it.
Then
details.
D
So
I
can
have
my
hand
like
halfway
raised
up.
I
need
to
first
get
confirmation
that
I
can
I
can
do
this
since,
since
it's
not
code,
I
should
be
able
to
do
it,
but
I
just
need
to
make
sure
with
legal
that
I'm
allowed
to
actually
write
like
a
full-on
proposal,
but
I
don't
think
it's
going
to
be
an
issue
but
I
just
I
just
need
to
double.
D
You
know
cross
the
t's
and
Dot
my
eyes
to
just
make
sure,
and
if
that,
if
they
say
yeah,
it's
not
an
issue,
then
I
can
I
can
start
constructing
a
PR.
I
can
let
you
know
by
by
tonight
Andrea
whether
or
not
I
can
I
can
create
the
pr.