►
From YouTube: 2022-08-30 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
B
B
So
I
cj
wouldn't
be
joining
today
and
I
think
michael
blanchard
is
also
out
on
a
vacation.
So
cjo
asked
me
to
drive
the
meeting.
B
Cool
there's
only
one
item
on
the
agenda
from
like
I
put
this
item.
This
is
about
the
patch
that
we
are
planning
to
release
for
1.30,
so,
basically
with
diagnostic
source,
seven,
the
fields
of
some
of
the
fields
have
been
renamed
and
our
reflection,
stuff
breaks.
B
So
anyone
who's
using
1.3.01
starts
and
then
starts
to
also
bring
in
diagnostic
source,
seven
version.
They
will
see
errors
like
method
access,
exception,
errors
and
yeah.
I
think
there
is
this
issue
which
describes
it
so
yeah
blanchett
has
a
pr
out
to
fix
this
thing
like
we
are
creating
an
upper
bound
on
the
version
of
diagnostic
source
that
you
could
use
when
using
1.3.1.
B
B
And
also,
I
think
we
don't
have
any
documentation
on
how
to
create
how
to
like
push
patches
so
like
our
releasing
dock
is
always
pushing
stuff
to
main.
So
I
think
we
could
also
create
documentation
for
that
like
how
you
check
out
a
particular
tag
and
from
there
on
you,
make
fixes
and
then
push
it.
B
A
Yeah,
I
guess
the
release
on
this.
One
will
be
interesting.
I
was
gonna
ask
like
it
seems
like
I
did
a
release
once,
but
it
was
a
long
time
ago.
We
always
release
from
main
right,
like
the
process
is
add
the
tags
core
and
encore
and
then
the,
but
when
the
when
the
release
actually
goes,
it
takes
the
build
off
of
main
right.
B
B
Actually
should
be
at
that
comment.
Let's
take
a
look.
B
A
B
B
Okay,
so
what
had
happened
last
week
was
that
after
I
merged
the
updated
change
log,
we
pushed
the
other
change
of
michael
blanchard
and
then-
and
I
so
I
had
already
tagged
the
comment
after
I
come.
I
merged
the
changelog
pr
right,
so
I
had
already
typed
it
as
alpha.2
and
now
we
push
in
another,
we
merge
another
pr.
B
B
A
I
think
that'd
be
a
good
thing
to
confirm
our
understanding
of
because,
at
the
end
of
the
day,
like,
I
think
that
we
want
a
release
process
that,
basically,
we
can
just
say
like
hey,
like
release
at
this
tag,
in
which
case
you'll
never
run
into
this
situation.
Where
menver
will
add
an
additional
suffix.
B
B
Then
quickly
again
check
the
releasing
dock
see.
So
what
we
do
is.
Okay,
we
decide
the
tag
update
the
change
log
entries.
We
then
push
the
tag.
A
B
Then
we
manually
trigger
a
build
right.
This
is
the
same,
build
which
we're
looking
at
back
in
publish
to
market,
and
this
is
triggered
manually.
A
B
So
this
workflow,
because
I
had
I
mean
I
had
completed,
step
six
and
then
we
merged
another
pr,
and
this
one
just
picked
it
as
a
next
one
extra
comment:
after
the
last
latest
tags
released
the
packages
as
2.1.
A
B
This
pack
would
was
the
this
is
where
that
dot
one
got
added,
because
this
is
following
the
manual
version,
minerva
versioning.
I
think,
even
if
you
were
to
do
it
locally,
we
would
see
that
like,
if
I
add
a
tag
and
then
I
push
another
comment,
then
I
do
a
dot
net
pack.
It
will
check
whatever
the
last
tag
was
and
then
add
the
number
of
comments
as
the
suffix.
B
Yeah-
and
this
is
packing
it
for
all
of
the
projects
right
so
like
all
of
the
different
packages
like
we,
where.
B
A
B
A
B
Yeah,
I'm
not
sure
like
what
this
does,
though,
what
is
like
we
don't
like
what
is
this
even
supposed
to
do.
Build
number
is
equal
to
this.
B
A
parameter
it
might
be
used
in
one
of
the
other
property
files.
A
A
A
B
Yeah
yeah
you're
right,
I
think,
in
the
contrib
menu.
We
have
it
a
little
different,
because
here
it's
manual
push
right
like
we
manually
trigger
the
workflow,
but
right
over
here
I
think
we
set
a
condition
saying
anytime.
A
new
tag
is
pushed
like
anything
here,
yeah
just
so
yeah
when
somebody
pushes
a
tag,
a
certain
kind
of
tag.
That's
when
we
trigger
this
workflow
automatically.
B
A
It's
triggered
from
that
tag,
but
yeah,
I
guess
I'd
still
question
whether
it's
it's
taking
the
commit
at
that
tag
or
whatever
is
at
the
tip
of
main,
probably
at
that
tag
in
this
case.
But
I
don't
know.
A
Yeah
anyways,
I
think
that
would
be
nice,
because
I
mean
that
that
I
think
would
make
the
patch
release
real
easy
in
that
case,
right
just
create
the
new.
B
B
Yeah,
I
think,
yeah,
let's,
let's
take
a
look
at
that,
I
think
when
we
are
close
to
merging
the
pr,
even
before
that,
I
think
I
can
at
least
make
sure
if
this
workflow
can
even
be
used
for
the
patch
purposes,
guys
yeah.
We
might
have
to
see
if
we
need
another
workflow
or
just
have
a
manual
approach
to
patches.
A
B
A
Yeah
we
we
emit
some
il
that
effectively
accesses
an
internal
count
field
right
there.
In
order
to
you
know,
I
don't
know
what
the
benefit
here
is.
I
imagine
it's
probably
avoids
like
an
allocation
or
something
this
kid's
been
here
for
a
long
time,
but
it's
it's
the
same
kind
of
it
could
introduce
the
same
kind
of
problem.
A
Right,
it
would
be
a
change
with
the
protobuf
library
and
we
we
had
a
problem.
I
don't
know
mine
about
a
year
ago,
or
so.
With
the
otlp
exporter,
we
had
another
field
that
was
being
accessed
just
in
the
same
exact
way,
and
it
did
change
on
the
protobuf
side
and
we
experienced
the
same
same
problem.
In
that
case,
we
fixed
it.
B
So
I
think
you
mentioned
some
ways
to
deal
with
this.
A
Yeah,
I
think
I
think
option
one
is
what
makes
the
most
sense
for
sure
I.
I
know
that
michael
was
looking
into
swapping
out
a
new
protobuf
into
or
use
the
new
protobuf
library,
but
I
completely
agree
with
sijo.
It
doesn't
make
sense
to
necessarily
do
that
investigation
as
part
of
this
effort,
I
think
that
all
we
need
to
do
is
just
like
catch
catch,
an
exception.
B
B
Yeah,
so
you
want
that
to
be,
I
mean
that
should
be
included
in
this
patch
itself.
A
B
B
Yes,
he's
on
vacation
this
week,
I
I
didn't
get
a
chance
to
talk
to
cj
about
it.
I
was
gonna,
ask
him
what
this
was
about,
so
I
think
it's
good
that
I
asked
you
right
now
like
what
was
the
second
option
and
like
investigate,
protobufnet
and
consider
using
it.
What
is
it
like
is
that
is
that
a
different
liability
to
like
create
those
contracts?
B
A
A
The
comment
that
michael
has
right
above
mine
here:
how
does
that
help?
It
has
basically
okay.
Now
now
my
memory
is
kind
of
getting
refreshed
here,
so
that
count
property.
I
think
what
you
need
to
do
in
order
to
get
the
count
of
this
repeated
field
which
is
effectively.
You
know
it's
a
list
of
of
spans
that
repeated
field
thing,
that's
what
it
represents.
A
Yeah
and
the
you
can.
You
can
do
a
count
on
it,
but
you
have
to
effectively
enumerate
the
collection
to
get
the
count
and
that's
expensive.
That's
that's
the
expense
that
this
code
is
avoiding
again.
If
my
memory
serves
me,
I'm
pretty
sure
about
that.
So
the
protobuf
net
library,
apparently
has
account
property.
It
has
the
same
kind
of
thing
like
this
repeated
field
thing,
but
it
has
a
it
has
account
property,
that's
public.
B
A
B
A
So
he's
looking
at
that,
though,
I
was
also
kind
of
talking
with
him
offline
about
that.
We
could
also
entertain.
I
know
that
the
open
telemetry
java
project
has
done
something
to
optimize
their
serialization
to
protobuf
over
the
wire
and
it's
totally
hand
rolled
like
they're,
not
using
a
library
at
all,
so
that
could
also
be
a
thing
that
we
look
at.
It
might
even
be
more
performant
than
protobufnet,
but
I
think.
B
B
A
Well,
the
doing
the
first
one
I
think
is
going
to
be
super
easy.
I
just
I
just
want
to
make
sure
nobody
else
already.
B
A
It
on
their
mind
to
do
it.
I
can.
I
can
do
that
tomorrow,
but
I
was
going
to
ask
like
do
you
have
you
all
talked
about
when
you
might
push
out
this
patch
release?
Would
it
be
this
week
or
would
it
be
later.
B
Not
yet,
I
don't
think
we
haven't
talked
about
this
yet,
but
I
mean
like
I
think,
let's
see,
let's
see
you
get
it
out,
get
that
pr
out
tomorrow
or
day
after.
I
think
it
should
be
fine.
We
need
to
include
it
in
the
patch
like
I'll,
have
to
talk
to
cjo
again
like
if
he
has
any
timeline
in
mind
for
releasing
the
patch.
B
Okay,
there's
a
chat
as
well
somebody
post,
okay,.
B
Let
me
open
this,
the
other
pr
on
the
agenda.
B
A
B
A
B
B
Okay,
any
other
things
to
discuss.
A
A
A
B
B
A
B
Joe
yeah,
I
think,
yeah
we
can
leave
early
then.