►
From YouTube: GMT 2018-05-29 API WG
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).
B
D
D
D
D
So
initially,
in
a
previous
record,
in
your
face,
I
hope
we
have
this
one
thing
called
obviously
results
which,
when
we
call
major
Music,
Hall
and
or
weaken
a
future
in
your
return,
we're
in
a
future
of
obviously
result
of
the
response
product
and
if
you
look
at
obviously
results
like
this
contains
a
status,
and
there
is
some
probably
self
but
harder
after
I
went
through
some
discussion
in
the
job.
We
teach
these
repo
there's
an
assessment
about
whether
we
should
make
so
in
the
state
so
currently
the
job
he
drunkenly
paternity.
D
D
If
the
service
is
now
okay,
because
the
server
returns
and
focuses
the
enemy
response
may
or
may
not
be
about
so
to
ensure
staff
safety
instead
of
having
this
proxy
result
that
have
a
status
with
potentially
dangling
response
rule
of
I
removed,
introducing
the
Costas
error
and
make
the
call
returned
the
future
of
track
of
respond
with
specialized
error
class.
So
this
will
ensure
as
to
have
type
of
safety
the
reason
I
don't
have
a
Fisher
like
a
specialized.
D
D
We
use
future
of
try
of
something
with
specialized
error,
so
we
can
programmatically
handle
whatever
of
the
error
and
still
provide
a
stimulus
response
and
been
journeying
and
I
agree,
and
then
anyone
created
a
ticket
for
for
the
view
for
in
the
future.
We
pro
things
we
only
consider
how
to
design
future
with
special
hypothesis
another
another
issue.
D
Thank
you
example
to
use
your
PC.
What
we
do
is
like
we
need
to
do
is
the
when
you
call
with
a
macro
that
let's
try
to
retrieve
the
function
name
of
the
general
generated
cold,
very
cold
from
Pro,
see
well,
this
number
this
recent
mean
may
in
the
future.
If
we
only
get
the
cherry
service
support,
listening,
I
feel
listening
me.
We
will
have
a
difficulty
in
reducing
a
new
macro.
E
D
Doesn't
best
specifics
like
specialist
for
Java
server
support,
so
I
rename
it
to
trans
client
method?
So
so
we
don't
have
any
name
completely
unnatural
boundary
at
serviceable
in
the
future,
and
the
next
thing
I
make
is
I.
Do
some
remaining
remain
a
thing
called
Java
channel,
which
is
a
wrapper
of
a
global
channel
class
and
I,
feel
it's
a
little
bit
convenient
to
have
our
own
wrapper
that
over
a
circle
name
and
also
I,
look
at
how
go
wrap
around
the
job
you
channel,
please.
D
D
B
D
Currently,
these
are
exposed
in
the
Blue
Cross,
as
slash
includes.
Oh
sorry,
sorry
is
turning.
The
processing
is
exposing
process,
/j
bczp,
noting
that
file,
and
now
a
nice
which
is
used
by
CSI
clients.
This
is
only
thing
we
we
have
right
now,
but
the
reason
I'm
editing
is
in
the.
If
you
really
want
to
have
half
a
pc-based
module
and
if
so,
that's
why
I
want
to
like
first
clean
up
the
interface
first,
then
we
can
in
feature
we
can
have
if
introduce
me,
like
the
service
based
model.
F
F
A
C
C
They
worked
fine,
but
one
of
the
drawbacks
is
that
container
metrics
aren't
Lim
process,
metrics
they're,
just
an
ad
hoc
JSON.
In
fact,
some
other
things
that
are
exposed
in
I
kinda.
What
the
proto
buffers
call.
Actually,
it's
not
metrics,
it's
something
different,
but
some
of
them
throw
things
in.
There
aren't
really
metrics
they're
difficult
to
kind
of
map
to
a
metrics
mindset,
which
is
a
little
bit
wonky.
C
So
I've
been
thinking
about
this
a
little
bit
I'm,
not
an
OG,
was
quite
strongly
in
favor
of
creating
some
sort
of
enough
abstractions
in
the
lib
process.
Metrics
API
is
that
you
would
be
able
to
represent
all
the
information
you
want.
You
would
want
to
expose
in
Prometheus
I'm,
not
really
convinced,
that's
the
right
approach.
We
already
have
Mises
exporter.
It
already
has
people
every
a
better
approach
would
be
to
talk
to
those
guys
and
see
about
see
what
they
think
about
trying
to
fold
that
project
into
the
larger
me
sauce
project.
C
So
it's
more
so
we're
more
closely
aligned.
It
seems
to
me
that
if
we,
if
we
take
the
Prometheus
approach
and
like
bundle
into
the
process,
we're
probably
significantly
narrowing
the
audience
of
developers
who
would
make
contributions
to
that
so
I
think
there's
a
little
bit
of
tenets.
That's
area
that
I
think
G
was
intending,
but
it's
sort
of
an
open
discussion
area
right
now.
That's
why
I
know.
B
C
A
A
You
know,
I
have
several
issues
where
people
have
asked
for
tags:
I,
don't
know
what
they
ended
up,
adding
them
yet,
but
it
looks
like
they're
going
to
add
them.
I
didn't
look
at
other
libraries
but
I'm,
assuming
they
probably
had
tags
given
the
given
the
presence
of
them
and
things
like
Prometheus
and
a
dog
and
the.
A
C
I
mean
there's:
lots
of
dots
of
traditional
metrics
representations,
have
like
two-dimensional
representations
of
things,
so
in
SNMP,
and
also
the
performance
copilot.
They
have
like
the
notion
of
two-dimensional
tables
and
I.
Don't
remember
what
the
SNMP
terminology
is,
but
in
PCP
is
calling
instance
domain.
So
you
have
a
set
of
metrics
and
you
have
this
and
each
instance
in
your
instance
and
has
each
of
these
metrics.
That's
representing
like
two-dimensional
data,
but
tags
doesn't
really.
C
A
C
It
Tech's
on
some
metadata,
depending
I,
think
for
the
container
one
it
it
because
it's
manually,
transforming
the
JSON
it
texts
on
the
container
ID
metadata
it'd,
be
when
it
has
two
separate
modes
for
master
and
agent,
and
when
it
knows
it's
working
in
one
of
the
either
of
those
modes,
it
tacks
on.
You
know
specific
metadata
around
the
master
or
agent,
so
it
no.
C
A
Another
thing
that
came
up
was
like:
should
this
be
configurable
in
some
way
like
for
those
that
don't
have
tags?
Should
they
should
we
be
able
to
specify
the
key
structure
if
they
want
to
change
it
like
if
they
want
to
say
I,
want
my
favorite
metrics
to
include
just
the
framework
name,
because
I
know
that
it's
unique
for
me
and
I
couldn't
read
those
and
I
understand
them.
We
could
let
them
do
that,
but
by
default
the
project
would
wouldn't
pick
framework
name
on
its
own,
because
it's
not
unique
right.
A
People
have
been
like
the
framework
ID
and
therefore
DCOs,
because
in
these
shows
the
framework
names
are
unique
and
the
framework
IDs.
You
can't
predict
ahead
of
time,
so
was
it
was
deemed
difficult
to
set
up
dashboards,
pre-existing
dashboards,
because
we
wouldn't
know
what
the
framework
IDs
would
be.
Yeah.
C
A
C
A
C
It's
it
is
I
feel
like
it
is
sort
of
a
hack,
but
necessarily
I
can
in
in
a
bad
sense.
I
mean
we
can
change.
The
thing
is
we
can
change
me
sauce
exporter,
much
faster,
you
can
change
and
deploy
mr6
water,
probably
much
faster
than
you
can
change
and
deploy
these
sauce
agent
I.
Think
it's
really
hard
to
design
to
boil
to
distill
the
common
essence
of
off
tags
into
a
general-purpose
API,
and
you
know
the
audience
of
people
who
are
going
to
make
changes
to.
A
D
A
A
E
A
A
Probably
because
whatever
they
use
there
is
at
lift,
probably
has
tags
in
it
or
someone
who
uses
envoy
asked
for
tags
or
something
like
that
sure.
A
E
A
A
C
A
C
B
A
A
D
Taking
that
for
the
tech
support,
we
probably
can
just
add
it,
and
many
unanswered
have
to
way
to
output
like
you
can
have
a
switch
to
say
we
on
the
conventional
key
value,
style,
output
and
we'll
pull
the
tag
as
part
of
the
kings
and
or
we
can
enable
some
tag
out
of
the
format
then
em
like
like
like
then
out
goes
something
just
follow
them.
Give
it
up
here.
F
D
Bringing
that
it
really
has
like
taxable
reaper
vu.
Another
thing
we
do
we
can
do,
is
we
just
add
text
and
in
our
existing
matrix
objects
and
for
each
metric
we
need
to
define
like
a
two-way
to
output
them
like
one
way
to
conventional
eBay
way.
The
other
way
is
like
richer
for
mantle
and
come
in
stacks
in
the
conventional
readers
and
somehow
have
a
convention
to
put
tax
into
the
teens,
and
then
let
the
user
to
have
a
some
switch
to
say.