►
From YouTube: 2021-01-29 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
C
A
A
A
A
B
D
B
A
G
Well,
yeah
nikita
wants
to
talk
about
benchmarks,
so
I
I
think
I
showed
some
people
the
spring
rest
pet
clinic
benchmarking
thing
that
I
put
together
and
it's
proven
pretty
helpful
in
a
few
cases,
most
recently
with
matash's
new
flexor,
we
could
demonstrate
pretty
clearly
some
of
the
performance
gains
and
I
think
nikita
has
this
idea
that
we
might
be.
That
might
be
something
that
would
help
the
community
if
we're
into
contributing
that
to
the
the
main
instrumentation
repo,
so
that
was
kind
of
the
topic
at
hand.
G
Right,
yeah,
I'm
and
I'm
into
it.
I
don't
want
to
yeah.
I
definitely
I
mean
it's
already
open,
it's
not
opened,
but
I
did
it's
just
under
my
name
in
github,
and
certainly
anybody
could
use
or
fork
that
so
I'm
not
like
trying
to
keep
it
protected
and
I
would
love
for
other
people
to
be
able
to
use
it
and
contribute
it.
But
I
also
don't
want
to
muddy
the
waters
in
the
main
gigantic
mono
repo
too
much.
B
E
B
E
G
Sure,
and
with
the
smoke
tests,
I
guess
they're
run
on
a
schedule
or
periodically
or
with
prs
or
whatever,
and
so
they
get
some
attention.
If
there's
a
breaking
change
that
causes
the
smoke
tests
to
fail,
we
at
least
have
a
way
of
figuring
it
out
or
finding
out
about
it.
We
would
probably
want
to
do
something
similar
with
benchmarks.
I
guess.
B
Yeah-
and
we
have
even
a
task
for
that-
I
think
just
to
run
them
to
verify
they're
still
working,
that's
easy,
yeah.
The
problem
is
is
if
we
want
to
run
them
periodically
to
get
meaningful
numbers
from
that,
then
we
need
more
of
a
stable
infrastructure,
but
I
think
that's
a
separate
issue.
I
mean,
if
you're
already
using
them
manually,
to
verify
the
effect
of
the
pull
requests.
G
Okay,
are
there
other
thoughts?
I
want
to
hear
from
other
people
too.
A
In
I
totally
agree
with
this
benchmarks
should
run
on
ci
just
to
make
sure
that
they're,
compiling
and
running,
and
that
if
people
want
to
know
how
to
run
them,
they
can
always
look
at
the
ci.
A
A
little
bit
even
better
would
be
to
have
a
readme
yep
under
here,
and
we
could
split
out
if
we
have
multiple
benchmarks.
We
could
split
them
out
under
this
folder
for
sure,
and
my
only
request
would
be
to
have
the
same
infra
for
all
of
the
benchmarks
sort
of
how
the
docker
containers
or
how
how
the
info
works,
whether
it
uses
work
or
some
other
j-meter
or
whatever
it
uses.
I
don't
have
a
preference
for
website
what
it
uses.
Just
that
they're
consistent.
G
A
I
would
prefer
that
just
so
that
it's
easier
for
us
to
one
less
thing
to.
G
G
The
one
that
I
used
allows
these
like
workflows
to
kind
of
be
defined,
and
I
know
that
this
one's
using
work
or
work
too
right.
So
it's
probably
easily
converted
to
the
other
one.
G
A
So
one
one,
I
think,
would
discuss
this
like
last
one
of
the
previous,
maybe
evening
meetings
but
one
of
my
feedback
on
the
benchmarks
I
tended
I
tend
to
like
the
the
work
style
thing,
not
the
flows,
I
mean
the
flow
is
awesome,
and
maybe
maybe
we
do
need
something
separate
for
a
flow
like
that,
a
more
realistic,
but
in
general
for
the
benchmarks,
the
more
sort
of
narrow
you
know
just
hitting
one
page,
and
we
can
parametrize
that
over
you
know
five
pages,
but
keep
hitting
the
same
page
over
and
over
gives
more
reliable
results.
A
G
A
Would
the
with
the
the
work
file
single
page
have
found
the?
Do
you
think
that
wouldn't
have
found
the
same
issues
that
you
found
with
the
flow.
G
G
A
So
yeah
I
mean
it,
it
would
make
a
makes
a
good
case
for
having
you
know
two
I
mean
having
that
flow
as
a
separate,
but
maybe
being
clear
about
that.
This
is
a
flow
versus
our
other
sort
of
end-to-ends,
and
even
if
we
could
have
an
end-to-end
on
that
pet
clinic,
I'm
not
sure
what's
have
to
review
what's
in
this
one,
but
if
it's
not
using
pet
clinic
pet
clinic
is
an
awesome
app
for
benchmarking.
F
A
I
can
start
with
a
just
update
on
one
o
discussions
that
we
had
on
tuesday
night
wednesday
morning,
depending
on
where
you
are.
A
So
the
java
john
everything
still
best
guess:
mid
february
feb
15
for
java
sdk,
one
zero.
D
We
were,
we
were
told
we
should
not
release
one
zero
until
the
spec
has
been
released,
officially
cut
at
one
zero,
so
I'm
still
hoping
to
do
to
publish
o.15
this
week,
onorak
broke
the
build
last
night
broke
the
publishing
last
night
again,
so
we
got
to
get
that
sorted
out.
D
A
Are
you
going
to
call
it
zero
fifteen
zero
or
one
zero
rt.
D
A
Makes
sense
so
on
the
java
instrumentation
side,
so
for
what
one
zero
means?
Api
stability
is
the
main
what
it,
what
one
zero
is
meaning
for
open
telemetry
in
general.
What
does
abi
stand
for.
D
D
A
Cool,
so
that's
one
component
for
us
to
consider
with
java
instrumentation
one
zero
and
the
other
is
telemetry
stability
and
so
there's
two
different
specs.
A
I'm
not
going
to
find
them
quickly
discussing
so
the
the
main
one
zero
for
sdks
was
all
focused
about
api
stability
and
then
telemetry
stability
is
an
ongoing
discussion
and
there
were
some
proposals
of
like
that.
Breaking
the
telemetry,
like
changing
a
span
name
or
on
changing
an
attribute
value
or
removing
an
attribute
or
even
adding
an
attribute,
would
be
potentially
a
breaking
change
as
far
as
telemetry
stability,
and
so
there
was.
A
We
were
concerned
where
that
was
going
because
for
the
java
agent,
which
bundled
so
much
instrumentation
it
felt
like
that
was
going
to,
we
were
going
to
end
up.
You
know
breaking
potentially
breaking
having
a
major
breaking
version
like
with
every
release
or
every
couple
months,
and
so
luckily
it
looks
like
the
way
that's
going.
Is
that
that's
not
going
to
be
required
for
1
0?
A
A
There's
still
a
lot
of
sort
of
fine
combing
that
we
need
to
do
on
our
apis.
Like
instrumentation
api
java
agent,
api.
A
Java
agent
spi
is
better
than
the
other
closer.
Probably
the
others
need
more
more
work.
A
A
And
so
the
remaining
kind
of
part
of
our
contract
with
users
would
be
configuration,
and
so
what
we
discussed
on
tuesday
is
environment
variables
from
the
spec.
A
You
know
that
those
would
we
would
keep
as
backwards
compatible,
not
make
any
changes,
but
other
things
we
wouldn't
necessarily
were
not
totally
sure
about
as
far
as
a
stability
guarantee,
but
there
there's
nothing
in
the
spec
at
this
point
saying
that
we
have
to
have
a
stability
for
those
configuration
properties
and
there's
a
couple
of
different
yeah
go
ahead.
A
A
So
I
just
dumped
a
lot
there
was
that
did
that
make
sense?
Are
there
questions
about
one
zero.
I
I
have
one
question
or
comment,
because
just
today,
I've
pushed
the
pr
that
publishes
testing
common,
and
I
was
wondering
whether
this
module
should
also
be
on
the
list
right.
It's
just
for
vendor
distribution
users
right
so
that
we
can
write
unit
tests
basically,
but
yeah
should
we
try
to
make
it
have
some
sort
of
stable
api.
B
I
B
A
My
rough
kind
of
thought
is
library,
instrumentation
and
instrumentation.
Api
are
p1
because
spring
cloud
sleuth
is
depending
on
them,
so
these
would
be
our
our
top
priorities
for
stabilizing
and
then
definitely
we
want
to
stabilize
everything
else.
A
A
A
A
And
so
there's
not
really
because
we're
reducing
the
scope
and
the
the
purpose,
the
point
of
reducing
the
scope.
So
much
here
is
because
we
would
like
to
release
the
java
agent
shortly
after
the
java
sdk
one
zero.
A
So
say.
If
somehow
the
spec
gets
one
zero
and
sdk
is
released
mid
fab.
Then
we
were
just
throwing
around
end
of
feb
for
java
agent,
one
zero
and
potentially
potentially
sooner
because
there
were
only
four
issues
here
that
we
thought
were
really
required
for
closing
out
one
zero
again,
because
we
just
don't.
We
have
such
a
reduced
surface
area
from
a
stability
guarantee
and
three
of
those
are
documentation.
A
A
A
A
Well,
I
will
share
something
fun.
Let's
look
at
the
dev
stats
just
to
show
how
how
active
both
of
the
java
projects
are
is
pretty
amazing,
I'll
look
at.
F
Up
up
iso
yeah.
F
A
That's
not
what
I
was
looking
for
at
all,
I'm
like
what
yeah
so
last
six
months,
you
know
both
of
the
java
repos
are
just
super
active,
so
you
know
good
work.
Everybody.
B
A
Yeah,
so
here
we're
back
to
six
month,
eight
one
yeah.
This
was
what
I
I
was
kept.
I
used
to
do
this
over
here
and
I
was
like.
Why
does
oh
month
does
work
here
because
it
doesn't
work
and
when
you're
on
the
default
over
here,
because
it's
too
big
for
the
last
so
months-
that's
how
wide
the
bars
are
by
month,
yeah
yeah,
it
doesn't
make
sense,
but.
A
Cool
all
right:
well,
if
there's
no
other
topics,
then
let's
get
half
an
hour
back.