►
From YouTube: 2020-09-10 meeting
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
B
B
Is
this
on
the
the
context
thing.
A
B
Okay,
yeah:
no
I've
not
looked
into
the
metrics
or
side
of
things
at
all.
A
Yeah,
the
main
concern
is
just
I
mean,
I
guess
not
a
concern.
I
think
I
know
we
would
just
not
do
it,
but
doing
something
that
dd
sketch
would
mean
having
a
c
component
here.
One
second.
A
A
A
B
Okay,
yeah,
I
I
don't
know
whether
I'm
just
like
it's
one
of
these
things
that
I
you
know
is
entirely
plausibly
like,
because
every
everybody
else
has
yeah
call
stacks
and
maybe
maybe
other
language
communities
don't
have
this
problem
that
we
have
where
we
feel
like
we've
got
to
add
our
stuff
without
making
any
modifications
to
what
we're
adding
it
to.
B
I
mean
like
like
this
this
this
this
kind
of
a
hands-off
integration
method,
and
so
then
that
maybe
that's
what's
backing
up
and
leaving
us
with
this
situation
where
we're
like
you
know,
we've
got
this
weird
telemetry
hookup,
and
that
means
we
can
get
our
first
call
and
not
our
second,
and
I
think
so,
so
we
need
to
cope
with
not
getting
our
second,
which
means
we.
You
know
that
starts
applying
back
pressure
on
the
way
that
we
keep
track
of
our
context
and
before
you
know
it,
it's
feeling
a
bit
ridiculous.
B
A
Kind
of
where
this
came
up
not
exactly
in
the
same
way
because
it
started
out,
I
think,
as
how
do
we
access
correlations
in
the
exporter
or
the
span
processor,
but
it's
the
same
fundamental
issue
of
well.
How
do
we?
How
do
we
go
back?
How
do
we
close
a
span
and
keep
the
content,
the
correlation
context
and
not
overwrite
stuff?
All
that
kind
of
oh.
A
B
Not
not
leave
a
booby
trap
right,
because
the
problem
was
that
you
know
I
would
set
my
context
back
and
then
the
the
weird
yeah
sorry
weirdest,
it's
too
strong
a
word,
the
the
behavior
that
made
sense
under
normal
operation
of
well.
Every
time
I
set
that's
a
push
and
then-
and
then,
if
you
you
end
the
span,
then
it
gets
quietly
popped.
B
You
know,
and
the
problem
was
that
if,
if
in
order
to
put
the
correct
span
context
back
because
someone
else
had
stuffed
up
that
pushed,
that
would
mean
that
the
the
next
person
back,
who
then
ended
a
span
thinking
that
they
would
end
up
with
the
parent,
would
end
up
with
my
context,
and
I
I'd
stuffed
him
up
and
so
now.
Of
course,
I'd
been
set
up
for
failure,
because
I
was
being
forced
to
clean
up
somebody
else's
failure.
B
But
it
just
struck
me
as
a
brittle
system
and
people
would
need
to
have
a
an
exquisite
map
of
exactly
how
the
whole
thing
worked
in
their
head.
Otherwise
they
couldn't
troubleshoot
it,
which
in
troubleshooting
code,
is
the
worst
thing
ever
right.
You
you
need
like
of
all
the
bloody
things
you
need
this
to
be
simple
and
predictable,
if
at
all
possible,
so
that
people
can
stay
worried
about
their
own
code,
not
ours.
B
A
Yeah,
I
think
the
original
intent
was
that
you
would
lose
it,
and
you
were
supposed
to
it's
supposed
to
be
immutable.
So
then,
in
that
case,
like
your
thing
would
be,
you
would
set
back
the
original
contact
like
you
wanted,
and
it
would
have
the
original
correlation
context
as
well,
and
everything
would
be
fine.
You'd
have
the
same
stack
of
spans
same
correlations
as
before,
and
the
children
would
go
away.
B
So
I'm
worried
I'm
going
to
be
wasting
time
by
going
through
examples
without
actually
having
read
anything
on
the
issue.
So
sorry,
if
this
is
too
much
of
a
divergence
which
case
just
go
in
there,
but
is
this
so
we
can
pull
off
tricks
like
we're
in
the
middle
of
handling
an
http
request.
As
part
of
that,
we
need
to
figure
out
who
the
user
is,
and
that
would
let
us
call
into
something
which
calls
into
something
which
goes.
B
Aha,
you
know,
and
in
its
own
span,
set
the
the
user
id
and
that
and
then
everybody
gets
to
gets
a
copy
of
that
kind
of
thing
so
that
if
then
the
you
know
and
then
the
problem
is,
if
you
crash
out
of
your
handler,
you
still
want
to
make
sure
that
the
user
id
is
recorded
on
all
the
spans.
Is
that
kind
of
the
thing
we're
after.
A
B
I've,
I
don't
know,
I
haven't-
had
this
kind
of
thing
available
to
me,
so
I've
just
gotten
really
good
at
having
things
return
the
values
high
enough,
so
we
can
put
them
in
the
spans.
We
want,
in
particular
like
the
the
high
cardinality
values
that
I'm
most
interested
in
are
all
available
very
shortly
after
taking
the
request.
B
B
Right
yeah,
I
mean
I
have
more
trouble
going
in
the
other
direction,
which
is
why
I
keep
asking
about
baggage
yeah
right,
it's.
Otherwise.
I
have
to
come
up
with
my
own
boutique.
I'd
have
to
come
up
with
my
own
mechanism,
to
you
know,
get
the
the
user
id
into
the
database
request.
A
B
Still
able
to
solve
problems,
I
mean
what
I
think
I
love
about.
The
the
environment
is
like
I'm
still
an
open
census
and
I
I've
which
is
not
even
as
capable
as
ot
and
I
took
all
of
the
ot
semantic
attributes
and
I
backboarded
them
to
my
instrumentation
libraries
and
I'm
comfy,
like
I'm
solving
problems,
I'm
productive.
This
is
good.
A
B
Right
and
especially
if
it's
across
language
communities
and
use
cases
because
yeah
like
I've,
also
got
this
very
happy
world
where
we
we
burnt
our
front
end
to
the
ground
and
read
it
read
the
back
end.
So,
instead
of
having
to
fight
our
way
from
react
through
apollo
client
through
absinthe
and
into
our
content,
you
know
phoenix
contacts,
you
know
and
and
then
to
our
records
ecto.
B
We
just
switched
to
live
view,
and
now
we've
got,
I
think,
literally
a
few
hundred
lines
of
javascript
total
for
a
very
interactive
web
app
and
we
we
skipped
all
of
the
fighting
through
those
layers
of
abstraction,
in
particular
graphql
and
yeah.
It
is.
B
We
are
moving
so
much
faster
than
we
used
to
because
we
taken
on
all
of
the
overhead
of
an
abstraction
that
was
appropriate
for
ten
times
our
employee
count,
because
that's
what
everybody
does
right
and
yeah
so
but
anyway,
so
so
then
yeah
we've
got
this
situation
where
we're
all
in
one
machine
like
it,
we're
back
in
happy
monolith,
land,
and
I
mean
we-
we
still
need
tracing,
because
you
know
I
I
haven't
had
a.
B
I
haven't
worked
on
a
single
monolith
where
I
wouldn't
have
wanted
this
technology
available
to
me
it's
just
too
bloody
handy,
but
life
is
easy,
and
I
can
imagine
that
you
know
once
you
widen
the
meeting.
You
end
up
plenty
of
people
who
are
not
having
it
quite
so
easy
and
can
come
up
with
some
really
curly
ones.
B
What
else
is
going
on?
I
kind
of
flamed
out
on
all
of
the
example
code.
I
just
can't
spare
time
for
that.
B
Did
brian
keep
going
on
that.
B
And
not
to
worry
okay,
so
we've
got
a
whole
bunch
of
spec
changes.
We've
got
just
still
trying
to
figure
it
my.
How
do
I?
How
do
we
fix
this
problem
of
dealing
with
the
span
and
the
context
separately
might
go
away
if
they're
no
longer
being
dealt
with
separately?
That
sounds
pretty
cool
to
me,
and
that
is,
do
I
think,
there's
much
else
going
on
no.
B
Yeah,
normally
we're
still
going
and
strong
for
t
plus
45
minutes,
but
yeah
I'm
good
to
go
mate
good
to
see.
You,
though,.
B
No
crazy
yeah,
I'm
trying
to
I
was
trying
to
time
the
the
end
of
the
californian
bushfire
season.
B
There's
going
to
be
like
the
yeah,
their
fires
are
going
to
start
tailing
off
and
then
there's
going
to
be
a
gap,
and
then
ours
are
going
to
come
up
and
I'm
like
that's
the
time
to
buy
an
air
particular
sensor
right
is
between
because
I'm
sure
they're
not
going
to
be
available
right
now,
because
everyone's
got
the
same
problems.
They
did
a
year
ago,
right
and
then
and
two
months
when
you're
not
going
to
be
able
to
get
them
here
either.
I
just
got
to
get
my
timing
right,
crazy,
pants,
sorry,
kids,.