►
From YouTube: 2023-03-07 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
B
C
It's
messing
with
zoom
here
as
I.
Apparently
a
new
release
went
out
and
then
just
killed
all
my
stuff.
So.
A
C
I
guess,
because
that
it
asked
me
to
download
the
packages,
so
I
can
blur
my
background
and
then
it
was
like
too
late.
I
joined
and
everybody
saw
my
messy
office
here-
that
the
fact
that
dozens
of
people
that'll
watch
this
video.
B
B
A
Which
lets
you
ship
from
one
pipeline
to
another,
so
you
can
generate
metrics
off
of
traces
and
then
not
have
to
do
like
an
H.
You
know
like
not
have
to
do
a
round
trip
back
to,
like
you
have
to
hit
the
external
internet
to
emit
metrics
off
your
traces.
You
can
just
connect
it
to
a
metrics
pipeline
or
logs
off
of
traces
or
pictures
off
of
logs
or
whatever
arbitrary
connections
between
pipelines
defined
by
these
connector
things.
A
Is
the
dream
I
don't
know
what
got
released
but
appears
to
be
something
that's
not
released
about
it.
A
A
E
C
And
then,
if
the
Spanish
processor
right
it
supports
some.
A
C
A
C
A
Yeah
I
mean
I,
we
don't
have
anything
specific
I.
It's
been
on
a
couple
PR's
recently,
so
I
figured
I
would
stop
by
I,
don't
work
on
any
other
stuff
or
any
of
my
free
time
or
any
of
my
day.
Job
at
this
point,
but
yeah
I'm
happy
to
discuss
this,
the
usual
stuff
or
I.
C
A
It
goes
yeah,
I'm.
D
D
A
A
Yeah,
my
team
has
like
he's
like
stupid
sweatshirts
but
they're
like
very
comfy,
so
I
wear
it
and
now
I'm
like
and
then
it's.
B
D
G
A
Good
cool,
yeah,
I,
don't
know
who
else
is
joining
that
guy
left.
A
D
G
Know
well
I
I
figure.
If
we
don't
have
anybody
to
wrap
to
give
us
back,
Sig
update
Tito's
could
pay
us
to
review
PR's.
Well.
I
have
this
time
set
aside
so
yeah.
A
A
Yeah
I
just
made
caveat,
is
that
I've
been
pretty
lightly
involved.
So,
like
a
lot
of
these
things,
I'll
just
have
to
be
like.
Oh
I,
ask
Francis
a
Robert
or
whatever,
but
like
I'm,
happy
to
get
my
opinion
and
try
to
approve
or
move
things
forward
as
much
as
I
can.
C
Sure
thing
it
says,
my
screen
is
loading
and
it
is
sharing
so
waiting
on
reviews
for
The,
Following
PRS.
C
E
C
B
C
As
me,
so
if
you
see
features
that
their
secret
right
should.
C
So
first
first
ball
up
is
to
add
additional
code
coverage
for
Sinatra
3.0
and
change
the
constraints
around
the
appraisals
to
allow
for
minor
bumps
and
testing,
basically
like
the
2.0
tree,
but
just
one
set
like
basically
the
latest
of
two
point:
some
you
know
of
minor
releases
same
same
goes
for
the
1.0
Branch.
G
C
Two,
not
when
you
say
Point
release.
Are
you
referring
to
bug
fix
or
are
you.
G
C
C
G
C
Get
generated
during
the
build
okay,
so
one
exception
to
this
is
for
the
3.0
I'm
doing
latest
patch
for
3.0.
B
C
A
You'll
rip
a
couple
patches
pretty
quickly,
probably
that's
usually
how
it
goes.
Yeah
I
mean
yeah.
If
they
start
doing
a
bunch
of
minor
bumps,
we
can
bump
that
we
can
just
do
it.
How
we
do
two
and
one
with
minor,
pessimistic
rather
than
patch
yeah
I,
approved.
C
Yeah
I
just
did
an
update,
so
it's
going
to
take
a
while,
because
I
I
merged
a
bunch
of
these
dependabot
things
yesterday.
So
is
that
safe
to
do
like
something
something
we
I
think
we
socialize
and
I
just
want
to
be
sure
about.
It
is
when
the
panda
bot
opens
up
a
PR,
we're
treating
it
like
it's
a
third
party
opening
a
PR
and
we're
doing
one
person
to
review
and
merge
because
dependent,
Bots,
basically,
upgrading
actions
and
gems
kind.
G
C
Past
then,
it
should
be
okay,
this
one's
a
little
bit
different,
so
this
one
adds
a
little
more
detail
to
the
build.
C
So
what
you
might
note
here
this
is
as
part
of
this
build
here-
is
that
we
used
to
do
just
a
quiet
installs
to
try
to
keep
it
like
the
verbosity
a
little
bit
low
for
when
it's
actually
doing
bundle
install
and
then
we
ran
the
test,
but
it
was
hard
to
tell
which
appraisal
file
was
running
at
a
particular
moment
because
it
doesn't
say,
like
hey,
I
I
know
what
appraisal
like
like.
You
know
it
would
fail,
but
it
wouldn't
tell
you
like
what
version
of
Ruby
failed.
C
What
version
of
the
appraisal
combination
would
fail
or
whatever
we
will
tell
you
what
version
of
Ruby?
Sorry,
because
it
was
saying
like
I'm
running
Ruby
three,
but
it
wouldn't
say:
I'm,
running
Ruby
3
with
Faraday
2,
right
appraisal,
combination
right,
so
I've
added,
just
like
basically
some
output
here
to
say:
hey,
we're
doing
an
appraisal,
install
and
testing
this
appraisal
file
and
then
saying
also
list
the
install
the
gems
once
the
installation
is
complete.
D
C
Opposed
to
doing
like
you
know,
removing
the
quiet
flag,
because
you
know
the
bundler
output
has
so
much
more
output
that
isn't
necessary.
This
is
kind
of
giving
us
a
summary
of
like
what
it's
going.
What
bundler
configuration
it's
going
to
use
when
it
runs?
So,
let's
look
at
what
that
looks
like
in
our
in
one
of
these
right
so
like.
If
we
look.
A
C
Three
two
yeah:
oh,
these
were
giving
us
a
bunch
of
Errors
right
so
anyway.
So
it's
saying
here
it
was.
It
was
giving
us
some
errors.
C
Oh
actually,
it
did
not
substitute
the
variable
because
it
was
a
syntax
error.
On
my
part,
bad
news.
E
C
C
Let's
go
ahead
and
I'm
just
gonna
go
ahead
and
commit
that
and
the
the
text
should
run
again.
This
time
will
actually
tell
us
we
can
come
back
to
this
one,
but
you
get
the
gist
of
it
right.
It's
gonna
say
like
I'm
running
this
file:
hey
cool
I,
like
this
little.
A
C
Later
and
so
look
here,
I
am
being
sloppy.
See
me
being
sloppy
here.
C
Everybody
in
the
internet,
you
can
see
it
happen,
okay,
so
this
is
the
same
deal
right
with
rack,
except
there's
a
different
Behavior
than
what
we're
using
for
the
for
Sinatra,
because
this
rack
2-1
was
separate
from
2-0.
F
A
I'm,
fine,
with
rack
being
us
being
a
little
nor
verbose
or
like
casting
more
combinations
of
things.
Just
because
it's
rack
and
everything
in
the.
A
Yeah,
all
the
yes,
the
the
comic
is
that
so
except
it's
maintained
by
a
horde
of
well
very
well
paid
co-workers,
who
might
that's
a
good
point
and
so
yeah
happy
with
any
additional
testing
there.
So.
C
C
A
C
I
think
the
problem
with
these
tests
is
that
you
know
how
you
have
like
the
initializers,
so
we
could
take
a
look
at
what
the
actual
code
is
doing
in
a
minute.
C
But
this
is
an
example
of
one
of
these
test
where
it's
like.
It
sets
the
state
of
the
instrumentation
and
when
you
install
the
instrumentation,
you
pass
in
a
config
object
and
then
it
does
some
stuff
to
the
config
object.
And
then
it
has
like
an
instance
variable
on
the
instrumentation
and
then
you
have
to
like
reset
it
right.
D
D
C
C
So
I'm
just
you
know
throwing
that
in
there.
So
if
anybody
is
interested
in
looking.
C
G
G
The
previous
ones
we
have
in
there
are
two
one
and
one
something
there
is
a
two
two
that
is
wholly
untested
by
our
Matrix,
and
it
is
the
one
that
has
had
of
the
two
series
more
recent
updates.
2-1
hasn't
had
an
update
since
20
20.,
as
well
as
2-0,
hasn't,
had
an
update
since
2012.
G
G
C
All
good
man,
it's
all
good
I-
want
to
know
actually
so
should
we
do?
Should
we
just
open
this
up
to
latest
too.
C
D
C
C
G
We
could
we
could
put
the
latest
one
minor,
the
latest
2-1
patch,
the
latest
2-2
patch
and
the
latest
3-0,
and
let
that
bake
for
a
while
and
then
at
some
point,
maybe
drop
to
one.
Oh
I
could
do
the
the
what
that
that
my
magical
query
of
what
are
people
downloading
from
ruby
gems.
D
C
D
C
C
F
C
C
C
C
C
So,
okay,
so
then
there's
the
biggie.
So
this
is
the
Biggie
Biggie
Biggie.
Can't
you
see,
sometimes
your
words
just
hypnotize
me
for
those
of
you
who
was
in
previous
episodes
of
this
show.
This
uses
the
rack
and
events
API,
which
was
introduced
in
rank
two.
C
However,
we
claim
kind
of
compatibility
with
Sinatra
one
which
relies
on
rack
one
now
I
have
no
idea
what
the
current
like
support
for
rack.
One
is:
if
it's
end
of
life,
we
don't
we
don't
see
active
development
on
rack
one.
C
So
it
looks
like
these
things
are:
gonna
have
to
coexist.
The
rack
events,
API
implementation
versus
the
middleware,
and
this
is
specifically
due
to
features
and
rack
that
are
not
available
in
rack,
one
that
make
it
so
that
we
can't
instrument
properly
right.
So
what
I've
done
is
added
a
new
option
so
that
users
can
opt
into
using
the
new
rack
events.
C
D
C
But
in
a
sense
it's,
like
you
know,
here's
a
rack
events
middleware
there
are
hooks.
You've
got
our
onstart
on
Commit
and
on
air
hooks.
There
are
some
other
things
that
Francis
had
asked
for,
which
was
like
hey,
be
nice.
To
add
these
other
events
to
inline
them
and
I'm,
like
you
know
what
I'm
gonna
hold
off
on
those
until
I
actually
get
some
usage
of
these.
C
So
the
the
one
deviation
of
feature
parity
is
going
to
be
that
there's
no
longer
going
to
be
an
HTTP
proxy
span.
Option
right,
HTTP
proxies
are,
is
represented
as
an
event
in
this
in
this
instrumentation,
but
everything
else
right
so,
like
I've
named
this
proc
this
event,
it
should
be
proxy
request
started
because,
as
the
time
stamp,
the
request
started.
C
That's
it
I
mean
outside
of
that.
Everything
else
is
about
the
same
and
yeah.
G
C
Yeah
I
don't
agree
either.
This
is
something
that
Francis
requested
he's
like
we've
already
do
this
for
sidekick.
You
know
it's
not
a
problem
that
the
events
are
represented
as
things
that
occur
outside
of
the
time
span
of
the
span
itself.
Oh
yeah,
so
I'm
like
yeah,
fine
I'll,
throw
it
in
there
as
an
option.
I'm,
not
gonna.
That's
not
not.
The
horse
I
want
to
die
on
the
one.
I
want
to
die
and
I
was
getting
my
feature
merged.
G
D
C
Stream
yeah,
it's
like
it's
gonna,
say
like
the
middleware
as
of
today,
right
exit,
like
the
span,
is
closed
before
writing
occurs.
C
This
event.
Api
has
different
levels
of
granularity
and
on
finish,
which
is
always
called
when
the
response
has
been
written
and
complete,
and
so
that's
what
we're
relying
on
here,
as
opposed
to
saying
when
the
application
has
completed
processing
and
the
response
has
been
committed.
So
there's
another
like
life
cycle
hook
here,
that's
like
the
response
was
committed.
C
Sure,
and
so
that's
what
we
are
so
this
requires
some
scrutiny,
obviously
and
I-
welcome
people
to
review
this,
but
I've
been.
This.
Is
my
one
I've
been
waiting
on
for
a
while
I
want
to
test
this
out
on
real
production
workloads
and
and
see
what.
A
C
Yeah
yeah
and
so
to.
D
C
With
this
option
of
use,
rack
events
does
is
that
you
know
it's
actually
we're
introducing
a
new
helper
method,
which
is
like
give
me
the
arguments
that
I
need
to
give
it's
a
rack
Builder.
So,
like
you
can
see,
let
me
let's
look
at
the
actual
code
here.
Let's
review
a
file
so
looking
at
like
an
example
of
the
usage,
you'd
have
to
do
something
like
this,
so
users
have
to
make
a
change
so
there's
this
use,
rack
events.
C
C
Where
in
the
world
is
it
because
there's
so
many
files
that
are
named
exactly
identical?
It's
not
here
it's
in
this
patches
right,
okay,
nope,
it's
in
the
rail
tie
there
we
go.
The
rail
tie
says
basically,
hey
look
I
want
to
insert
this
middleware,
so
I've
added
this
helper
function.
Here,
that's
like
actually
do
the
comp.
You
know
give
me
give
me
the
appropriate
middleware,
based
on
this
configuration
option,
flag
yeah.
C
Only
the
way
that
this
works
is
that
it
has
to
take
multiple
arguments
because
what's
getting
returned,
is
this
rack
of
like
rack
events?
Is
the
middleware,
but
then
you
add
an
event
handler
that
has
all
the
life
cycle
hooks
on
it?
C
A
I
I'm
happy
to
give
my
two
cents,
which.
A
0.87
cents
or
whatever,
but
yeah
I,
think
the
truth
is
I.
Don't
care
about
the
all
the
underlying
things
as
long
as
it's
like
I
think
the
goal
should
be
get
this
tested
in
production.
You
know
like
validate
that
it
is
robust
and
then
we
can
and
then
we
can
budget
over
like
the
ergonomics
of
it
and
all
that
stuff,
but
like
it's
hidden.
G
G
A
A
What's
like
the
easiest
way
to
like
allow
people,
the
flexibility,
while
still
like
you
know,
making
it
relatively
easy
to
use
and
like
those
are
all
valid
questions,
and
just
like
don't
matter
if
it
doesn't
work,
you
know
so
like,
let's
make
sure
you
know
like,
let's
just
make
sure
it
works
and
then
okay,
you
know
I
do
think
it'll
be
like
I
I've
mentioned
this
thing.
It's
a
recurring
thing
before
where
it's
like
we're
at
some
point,
like
let's
say
you
know,
you
run
it.
A
You
know
GitHub
for
the
next
month
and
it's
great
and
whatever
you
know
like
in
school
and
performance
is
better
about
100
milliseconds.
So
it's
like
there's
a
push.
We'll
have
this
weird
scenario
where,
like
suddenly,
if
we
don't,
if
we
release
it
as
the
default
on
like
a
minor
version
box
like
people's
performance,
like
people's
measurements,
will
just
like
suddenly
change
under
their
nose.
So
I
I,
like
the
I've,
been
like
Harpy
honest
for
a
little
bit,
but
just
like
the
concept
of
like
using
a
1.0
like
very
intentionally
to
be
like
look.
A
Why
did
my
you
know
all
my
graphs
suddenly
started
alerting
because
the
thing
you're
measuring
now
takes
50
milliseconds
longer
and
because
in
reality
we
were
measuring
the
wrong
thing
earlier
and
so
like
yeah
anyway,
this
is
cool.
That
was
awesome,
I
didn't
know,
racket,
that's
the
thing
I've
been
paying
attention
at
all
I'd
be
curious.
If
any
vendors
are
using
this
like
elastic
or
data
dog
or
anyone,
none.
C
And,
in
fact,
right
like
this
is
in
response
to
or
like
this
is
related
and
I
referenced
this
in
the
issue.
C
D
C
A
B
B
A
No
I
definitely
get
your
nose
either
here
since
2.0
I
think
yeah,
that's
that's
whatever
I
can
do
to
help
get
this
out.
I
haven't
deeply
reviewed.
It
so
I'm
hesitant
to
approve
it,
but
I
can.
If
you
need
an
approval.
I.
A
That's
more
like
the
that's
fine
I
think
you
know
again,
like
I
think,
there's
patterns
we
can
establish
this
won't
I.
Don't
think
this
like
something
like
AWS
SDK.
Might
this
we
might
have
similar
things
with
this.
We're
like
some,
you
know,
V2
versus
V3
is
going
to
have
a
long
Delta
of
like
some
repos,
maybe
stuck
using
V2
and
so
yeah
like
I,
do
think.
A
There's
probably
this
is
in
a
code
smell
that,
like
we
should
probably
establish
some
sort
of
pattern
for
having
like
wildly
different
versions
of
instrumentation,
depending
on
major
version.
A
That
was
a
thing
we
had
to
do
at
datadog
for
a
couple
I'm
blanking
on
the
instrumentation
so
like
it's
not
not
unheard
of
I
mean
and
yeah
I,
don't
know
again,
it's
all
second
order,
stuff
where
it's
like,
let's
just
make
sure
it
works
first
and
so
and
you're
a
candidate
to
you
know
like
normally
that's
the
hard
problem,
because
you
have
to
release
it,
get
people's
feedback
but
you're
willing
to
dog
food,
it
so
cool.
So.
G
Less
than
two
and
three
combined
yeah,
so
on
the
one
side,
instrumenting,
an
old
app
tells
you
that
your
old
app
is
old.
So,
like
would
instrumenting
the
app
help.
People
know
that
it's
the
whole
the
promise
of
observability
and
instrumenting
things
is
so
that
you
can
see
that
how
works
when
you're
operating
it
yeah
and
when
you
start
making
changes,
does
your
still
work
and
like
your
test
Suite
can
pass,
but
is
production
healthy?
G
You
don't
know
unless
you
instrument,
so
that's
an
argument
to
support
some
of
the
older
things
so
that
you
can
instrument
and
then
now
that
you
can
see
that
rack
spans
are
successful,
whether
they
end
when
the
middleware
is
done,
processing
or
whether
they
end
after
the
stream
has
been
written
fully
to
the
socket.
It
is
interesting,
but
but.
G
Like
did
it
succeed,
or
is
it
the
bed
and
and
there's
a
there's
part
of
me
that
wants
to
argue
for
supporting
old
stuff
so
that
you
can
get
your
old
stuff
instrumented?
You
could
then
have
visibility
into
whether
the
system
is
working
and
then
you
can
go
like
well.
I'm
gonna
upgrade
all
right.
The
test
Suite
passes,
but
when
we
run
it
in
production,
I
see
that
my
racks
pants
are
the
bed
like
there's
date,
because
it's
been
instrumented.
You
can
see
whether
your
migration.
C
B
C
C
Our
instrumentation
stopped
supporting
a
previous
version
if
you're
maintaining,
if
you're
running
you
know
Sinatra
one
or
whatever
right,
okay,.
C
A
Instrument
I
am
like
he
was
reading,
not
to
not
to
interrupt
too
much
I
I
would
say
the
one
like
concrete
strong,
not
even
just
like
oh
air
gap,
computer
and
somewhere,
but
like
I
was
reading
the
quinnipeg
blog
the
other
day
and
the
latest
version
of
Lambda
for
Ruby
is
is
2.7
because
AWS
is
just
like
not
I.
A
Don't
know
why
that
was
the
point
of
the
blog
to
be
like
hey
like
what
what's
going
on
here,
like
biggest
company
in
the
world
like
have
you
thought
about
like
in
modern
software,
but
there's
probably
a
huge
number
of
users
running
ruby27
just
from
Lambda,
you
know
just
from
like
those
environments
so
yeah,
we
might
wanna
special
case
this
one
or
something
I,
don't
know.
That's
just
like
an
obvious
one
that
sticks
out
to
me
of,
like
that's.
A
Enough
to
run
Ruby
three
on
there,
their
most
popular
serverless
offering
but
I
don't
want
to
get
into.
C
It
so
questions
so
so
this
this
shifted
the
conversation,
because
this
is
a
tangent.
This
is
related
right,
like
making
policies
about
Ruby
versus
individual
gems
and
where
we
do
our
cutoffs.
C
So
this
is
for
me
saying
like
there
will
be
a
cut
off
right
like
so,
if
you're,
an
AWS
and
and
you're
running
AWS
Lambda.
What
are
you
instrumenting,
these
libraries
that
don't
have
compatibility
with
ruby27
that
do
have
compatibility
with
two
seven
right?
Is
this
not
the
same
problem.
G
A
Oh
I'm
saying
is
that
I
think
if
you
looked
at
the
cncf
and
maybe
like
who
pays
for
it,
you
know
my
Amazon
might
be
up
on
that
list,
and
so
we
might
just
I
just
I'm
just
doing
it
out.
There
I
think
we
should
have
a
policy
which
stick
to
the
policy
and
if
the
world,
if
that's
difficult
for
people
like
they
should
take
it
up
right.
C
A
Agree
we're
right
we're
we're
doing
this
on
a
part-time
unpaid
basis.
We
should
do
things
that
we
can
support
realistically
and
even
at
that
part-time
basis.
It's
really
just
like
you
know
one
or
two
people,
so
yeah
I
I
was
just
I,
really
don't
have
strong
opinions.
I
was
just
throwing
out
that
one
fact
I
just
was
like
oh
I
have
a
factoid
and
I'm
gonna
insert
it
into
this
conversation,
because
I
like
to
hear
myself
talk.
C
Truth,
absolutely
absolutely
so
what
I
so
I'm
gonna
go
back
and
I'm
gonna
say
this
one
more
time
so
that
it's
very
clear
there'll
be
nothing
that
changes
about
our
instrumentation,
the
you
know.
As
of
whatever
moment
we
say
the
next
release
of
this
thing
will
no
longer
you
won't
be
able
to
install
newer
versions
of
things
going
forward
and
that's
it
I
mean
that's
all
that
we're
saying
and
our
test
Suite
will
no
longer
try
to
adhere
to
compatibility
with
two
seven.
C
G
B
B
G
G
C
So
what
I,
what
I,
don't
know?
What
that
means
is
for
the
for
the
instrumentations
themselves,
like
what
instrumentations
would
have
to
get
dropped
as
a
result
of
this,
because
they
don't
support
Ruby
three
I,
don't
think
there's
any
right,
because
we
test
everything
against
Ruby
three,
so
I'll,
be
it
I
get
you
know.
So
the
cousin
of
this
one
is
the
is
the
dropping
Sinatra
one.
C
G
A
A
A
I'm
leading
yeah
I'm,
leading
on
you
and
Matt
as
well
as
here
like,
if
that
Playbook
turns
out
to
be
ineffective,
like
I,
think
we
should
be
flexible
or
like
think
about
how
we
can
make
it
a
better
experience
for
you
know
you
guys,
but
right
now,
you're
saying
that's
fine,
so
taking
the
word
for
it.
Well,
we've
we've.
D
G
A
G
G
E
B
D
G
C
The
one
thing
that
I
was
as
we
continue.
This
is
conversation.
One
thing
I
thought
was
interesting.
What
Eric
was
saying
was
you
know?
Perhaps
maybe
what
we
need
is
like
something
has
a
little
more
explicit
or
something
I,
don't
know.
Maybe
there's
like
not
to
make
it
more
complicated,
but
an
idea
that
I
thought
I
heard
you
say
was
have
different
instrumentations
four
different
versions
of
something.
Does
that
mean
like
you're,
suggesting
that
we
have
like
a
Sinatra
1.0
gem
versus
a
Sinatra
latest.
A
Gem
or
something
like
that,
just
establishing
a
pattern
for
sort
of
like
no
I,
don't
think
we
should
start
cutting
individual
gems
per
like
that.
Just
feels
insane
and
not
just
beyond
maintainable,
but
just
having
some
way
to
strike.
You
know
so,
when
you're
going
through
like
a
repo
you
can
in
instrumentation.rb
like
there's
some
CL,
you
know
maybe
there's
some
canonical
method
that
determines
this
and
there's
some
canonical
file
structure
for
splitting.
A
You
know
different
major
version
instrumentations,
something
along
those
lines:
I,
don't
it's
not
a
well
I,
don't
have
like
a
I
can
find
the
example
in
datadog.
This
is
our
DD
Trace
RB,
where
like
this
was
occurring,
I,
don't
know
if
we
ever
established
like
a
pattern,
super
great
pattern,
I
know
I
think
in
JavaScript.
Maybe
there
was
more.
This
happened,
more
I've
been
Ruby,
but
anyway,
just
something
like
that,
so
that
we're
not
yeah
I,
don't
know
it
just
so
that
it's
not
you.
A
You
know
it's
not
a
question
every
time
of
like
do.
We
start
overloading
these
because
it
just
matters
with
like
you
know.
If
then,
you're
trying
to
test
do
unit
testing,
it
becomes
like
you
know
it's
better
to
just
have
some
clear
pattern
for
this
stuff
and
but
yeah
on
the
like
I
think
it
would
be
an
extreme
case
where
we'd
ever
want
two
gems.
For
the
same,
you
know,
Library
I,
think
that
would
really
be
that's.
Something
I
haven't
really
thought
about,
and
just
intuitively
feels
like
a
mess
agreed.
A
C
In
the
spirit
of
you
know,
use
as
many
bring
up
as
many
ideas
as
you
can
right
like
yeah
and
share
them
and
see
what
sticks
right.
Okay,
continuing
on
this
thread
here
of
this
is
that
the
channel
you
know
you
wrote
this
up
some
time
ago.
This
issue
and
I
should
have
linked
it
from
what.
D
E
D
C
It
see
if
I
can
find
it
in
here,
which
was.
C
I
could
spell
that
right?
How
about
issues
where
the
author,
let's
go
here
and
say
like
this.
A
C
C
C
Let's
do
that
so
we're
back
to
this
again,
where
it's
kind
of
like
what
do
we
do
about
this
versioning?
So,
like
one
thing,
I
know
the
pen
about
doesn't
understand
appraisals,
so
it's
not
able
to
go
through
and
say
like.
C
Let
me
go
update
these
appraisals
to
get
this
compatibility
Matrix
up
to
date
and
we
kind
of-
and
you
know
this
requires
so
then
that
requires
human
beings
going
through
and
saying,
let's
update
these
appraisal
files
so
that
we
have
code
coverage
which,
as
you
can
see,
we
already
fell
behind
and
I
missed
one
right,
Rob,
it's
like
2-2
doesn't
get
any
code
coverage
unless
we
add
it
manually
sure.
So
that's
not
I,
don't
think
that's
sustainable
or
scalable
one
anyway
right.
The.
C
Yeah,
so
it's
kind
of
like
we
need
to
come
back
to
talking
about
sort
of,
like
version
constraints,
figuring
out
better
ways
to
determine
compatibility
as
new
things
are
released.
How
do
we
add
coverage
for
those
is
appraisals,
the
best
approach
for
this?
Is
there
something
else?
Should
we
contribute
back
to
appraisals
and
make
it
more
robust?
C
C
F
C
F
A
I,
don't
have
a
good
in
my
mind.
The
answer
is
like
some
process
that
automatically
grabs
the
latest
version
of
every
instrument.
Every
Library
and
you
know
once
a
day
runs
our
test
suit
against
I.
Don't
know
if
that's
it
like.
Maybe
that
exists,
maybe
there's
an
actual
way
to
do
that.
I
still
would
would
help
us
surface.
These
incompatibilities
of,
like
hey
Dolly,
came
out
with
a
minor
version
bump
that
violated
summer
and
changed
stuff,
like
you
know,
but
I
don't
actually
know.
A
Don't
know
if
that
exists
in
the
world
or
if
there's
a
way
to
actually
do
that
programmatically,
but
that
would
help
I
think
would
be
a
nice
would
be
a
basically
a
low
friction
work
around
to
giving
us
this
visibility
rather
than
having
to
implement
this
whole,
like
min
max
version
stuff
where
we're
like
protecting
ourselves-
and
you
know,
and
just
rather
than
that,
just
sort
of
like
let
things
fail
on
a
pointless
overnight,
build
and
use
that
as
a
as
a
canary
but
I.
G
A
G
The
smaller
the
baby,
your
step
in
there
would
be
something
like
of
the
gems.
We
have
instrumented
poke
the
rubygems
API
and
point
to
see
what
are
the.
What
are
the
minor
versions
available
for
this
thing?
So,
like
get
all
versions
truncate
the
last
one
sort
unique
now
you
have
a
list
of
Minor
versions.
Are
they
all
present
in
appraisals,
if
not
send
a
slack
message
or
right?
If
that's
like
a
Cron
job,
GitHub
action
that
that.
A
G
The
babiest
of
Step
would
just
be
alert
when
there's
a
difference,
not
trying
to
make
the
change
and
run
the
test.
Suite.
A
G
A
G
A
C
A
C
Applaud
Applause
what.
C
A
C
G
Yeah
I'm
working
when
I
I'm
working
on
trying
to
add
that
to
my
standard
schedule,
all.
C
Right
mommy
I'll,
set
aside
to
I'm
gonna
ping,
you
in
the
in
the
cncf
side,
all
right
peace
out.