►
From YouTube: 2022-07-20 meeting
Description
OpenTelemetry PHP SIG Meeting
B
All
right
we're
five
past
the
hour.
We
can
get
rolling
good
morning
good
afternoon
good
evening
to
both
you,
let's
see
what
we
have
on
our
agenda
today,
we'll
we
can
start
with
open
pr's.
B
I
know
you
have
an
open
pr
right
now.
You
have
one
it's
probably
not
worth
walking
through
the
rest
of
these.
Why
don't
we
just
talk
about
each
of
your
pr's
that
you
have
open
if
there's
anything
that
you
want
to
discuss
with
them.
Maybe
we
can
start
with
yukishan.
A
Sure
so,
basically
my
pr
is
about
v3
propagators.
I
have
implemented
a
multiheader
propagator.
This
pr
is
for
single
header
propagators
and
according
to
open
telemetry
specs,
we
don't
have
all
the
matching
fields
between
b3
propagators,
the
keys
in
the
breathy
propagators
and
our
open,
telemetry
specs.
So,
based
on
that,
I
have
implemented
both
of
them.
The
configuration
also
requires
an
extra
additional
class
that
handles
both
of
these
propagators.
So
basically
you
have
to
create
one
class
by
default.
A
It
should
send
the
headers
in
b3
single
format
unless
otherwise
given
in
the
configuration.
A
So
that's
what
I
implemented
in
that
a
new
class
for
b3,
single
adder,
propagator
and
another
class
that
would
handle
both
of
them.
The
issue
with
that
is,
if
you
look
at
the
propagators,
they
are
implemented
as
a
singleton
classes.
A
So
it's
difficult
for
me
to
test
that
composer
composite
class
that
handles
both
v3
single
and
v3
multi,
because
once
you
create
an
instance
of
that,
you
cannot
change
it.
A
Yeah,
so
what
happens
is
I
want
to
test
it?
I
want
to
see
if,
let's
say
first,
it
picks
up
b3
by
default
or
not,
but
as
soon
as
I
do
get
instance
for
that,
in
that
test
case,
it
will
set
it
as
b3
single
only
throughout
the
entire
test
case.
So
whatever
test
cases
I
write
next,
they
are
going
to
fail.
So
I
just
wanted
to
know
why
they
are
implemented.
As
singleton.
I
looked
at
open,
telemetry,
js
and
open
telemetry
python
in
js.
They
don't
have
a
static
class.
A
I
mean
whenever
they
want
to
use
this
propagators.
They
just
create
a
new
object.
Out
of
that,
I
am
not
sure.
What's
the
reason
behind
the
static
class
implementation
for
this
propagators,
I
know
yeah,
but
once
we
initialize
them,
I
mean
we
just
want
to
send
them
right,
so
maybe
it
would
be
redundant
to
create
new
copies
of
this
classes.
Every
time
we
want
to
send
some
header,
but
that
is
causing
the
issues
with
the
unit
testing.
A
So
if
there
is
any
any
other
way
around,
I
don't
know
so
that's
what
I
wanted
to
discuss.
B
B
Oh
okay,
so
there's,
if
we
go
and
look,
can
you
can
you
tell
me
on
what
file
you're,
seeing
the
singleton
pattern.
A
I
mean
it's
implemented
across
all
the
propagators,
so
if
you
pick
up
any
file
that
I
have
changed
in
that
pr
right,
we
are
implementing
yeah
singleton
class
for
all
the
propagators
we
yeah.
We
trace
context
everything
so
I
mean
initially,
there
was
only
this
trace
context,
propagator
and
that
was
implemented
by
a
singleton
class.
So
that's
what
I
carry
forward
with
this
b3
propagators.
B
B
It
was
teemo
that
did
that,
so
we
should
probably
ask
him
he's
not
here
today,
but
I
saw
you
tagged
him
in
the
pr.
I
can
I'll
follow
up
with
him
and
ask
him
about
that
and
see
if
he
can
get
you
an
answer,
because
that's
unfair
for
you
to
be
waiting
around
for
this
pr,
because
he's
yeah
I'll
follow
up
with
him.
A
Yeah
I
mean
it's
required
this
attack
it's
required
as
well,
so.
B
Yeah
yeah,
it's
very,
I
agree
with
you.
This
is
very
important
so,
but
the
good
news
is,
I
think,
we're
very
close
to
this
pr
and
your
test
coverage
is
absolutely
fine
right
now
so
we'll
I
might
not
be
as
big
of
a
deal
as
you
think
it
is
well
I'll
follow
up
with
you
sure.
B
Take
your
plane
amber,
are
you?
Did
you
have
a
pr
that
you
wanted
to
review.
C
Yeah,
so
this
is
the
issue
number
415
use
of
need
it
across
the
repo.
So
in
this
issue,
basically
in
few
cases
across
the
repo,
we
are
checking
whether
an
iterable
is
empty
or
not.
So,
basically,
in
most
cases
when
the
I
triple
is
in
the
form
of
array
and
this,
if
there,
if
it
is
empty,
the
anti-function
will
get
true,
but
in
case
of
exhausted
generators,
basically
it
will
give
false.
So
that
was
the
bug
that
was
given
this
issue.
C
Basically,
so
I've
made
the
changes
and
for
review
so-
and
I
have
also
means
correct-
used
my
function
in
which
they
are
created
in
few
of
the
occasions
in
which
it
was
used
in
the
repo.
So
you
can
take
a
look
and
let's
means
when
I
get
the
review
comments,
I
will
also
write
the
update
the
unit
test
cases.
I
have
not
updated
the
unit
test
cases
for
this.
Okay
once
it
is
reviewed.
B
B
So
you
said
you
just
you
said
that
your
next
goals
are
to
update
the
other
cases
where
we're
using
iterables
in
our
repo
and
then
this
pr
should
be
ready
for
review.
Or
are
you
sure.
C
B
All
right
that
sounds
like
that
sounds
like
a
plan
to
me.
I
think
that
this
looks
good
to
me
by
default.
I'm
happy
to.
We
could
also
yeah.
We
could
also
just
merge
this,
as
is,
and
then,
if
you
want
to
go
back
and
check
for
more
intervals,
or
would
you
rather
just
add
these
things
to
this
pr.
B
All
right
cool
I'll,
I'm
merging
it
now
so
don't
forget
about
it
and
then
yeah
that
if
you
could
make
those
updates,
that
would
be
great
sure
thanks
all
right.
B
A
Yes,
I
have
just
one
point,
so
I
think
we
have
made
good
progress
with
instrumentation
regards
in
symphony
framework,
at
least
now.
A
At
least
we
could
have,
we
could
have
instrumentation
for
a
couple
of
more
mvc
frameworks,
at
least.
B
B
I
think
it's
sort
of
a
chicken
and
egg
problem
to
me,
because,
if
we're,
if
our
library
isn't
in
beta,
then
these-
I
think,
let
me
say
this
a
different
way
when
our
library
becomes
data.
I
think
a
lot
more
people
will
take
a
vested
interest
in
having
their
libraries
or
frameworks,
or
you
know
what
have
you
instrumented
with
open
telemetry?
I
think
a
lot
of
people
see
free
beta
as
no
thanks
for
not
touching
that
yet
because
it's
not
ready.
B
So
I
think
that
our
goal
is
to
create
the
beta
so
that
we
get
more
community
involvement
and
get
the
people
that
are
engaged
in
specific
frameworks
and
libraries
and
stuff
to
implement
open
cemetery.
Now
that
we
are
a
little
bit
more
confident
in
its
functionality.
A
Okay,
okay,
yeah.
The
reason
I
asked
is
I'm
planning
to
experiment
something
with
regards
to
the
auto
instrumentation.
So
if
that
works,
fine,
maybe
I'll
raise
some
proposal
or
I'll
present
it
here
in
the
couple
of
next
segments.
D
Yeah
just
to
add
what
crystal
is
talking
definitely
auto.
Instrumentation
is
something
which
is
important
for
any
open
telemetry
as
a
php,
the
python
and
other
php.
Definitely
that's
the
something
which
is.
We
are
finding
the
gap
internally.
It
is
a
number
and
also
some
case
who
work
from
my
team.
We
are
discussing
this
thing
internally.
D
We
do
have
some
proposal,
so
that's
what
person
is
mentioning
there
might
be
some
additional
ask
when
we
plan
for
the
auto
instrumentation
one
of
the
things
and
if
you
can
mention
about
that
c,
plus,
plus
sdk
or
maybe
c,
plus
plus
dependency
right.
So
all
those
things
we
will
be
talking
but
about
the
whole
idea
is
come
up
with
the
proposal.
Seek
the
input
from
you
guys
and
whatever
agreeable
thing
should
start
doing
this
auto
instrumentation
as
quickly
as
possible.
D
B
That's
that's
great.
I
think
that
I
think
that's
a
good
mindset
too.
I
think
I
have
a
feeling
that
when
we
go
to
beta
we'll
get
a
lot
of
really
great
feedback
and
I
have
a
feeling
that
auto
instrumentation
will
be
towards
the
top
of
that
feedback
chain.
So
I
think
the
more
that
we
can
be
prepared
for
that.
The
better
off
we're
gonna
be,
and
I'm
gonna
be
very
happy
to
read
your
proposition
and
probably
I'm
sure
that
we'll
have
a
lot
of
really
great
back
and
forth
with
them.
D
Absolutely
that's
the
thought
we
have
so
yeah
come
up
with
a
proposal
and
as
soon
as
you
are
ready,
bring
it
to
here
in
the
meeting,
and
we
will
discuss
and
bob
you
before
this
you
were
saying
definitely
about
beta,
express
and
we
want
to
complete
hbp
as
possible
and
last
time
I
remember
we
were
talking
about
the
timeline,
though
we
didn't
have
commitment.
Do
you
have
any
sense?
When
can
we
reach
at
the
beta
scale?
What
is
the
current
status.
B
I
think
we
have
two
open
issues
right
now,
the
one
that
kashan
is
finishing
up
and
then
one
that's
called
configuration
interface
and
default
configuration
that
brett
and
chima
are
working
on
and
then
I
think,
we'll
be
ready
for
a
beta
release.
D
Oh,
that's
great,
it's
only
two
things
so
once
these
things
are
done,
then
we
will
have
our
beta
and
before
even
after
completion
of
these
some
of
the
pr
is
there
any
other
plan
to
have
the
formal
integration,
testing
or
those
kind
of
thing
or
those
things
are
always
happening
as
a
part
of
each
individual.
Pr.
B
Those
things
are
yeah,
those
are
part
of
our
cicd
process,
they
happen
the
the
and
the
union
integration
tests
are
part
of
our
csd
pipeline,
and
so
we
we
are
confident
that
we
can
push
this
debate
and
it's
a
beta
right.
There
will
be
bugs,
we
will
fix
them,
but
we
can't
be
timid
to
release
it.
D
Yeah
so
let's
have
even
further
discussion,
I'm
sure
in
next
week
or
probably
in
a
two
week,
we
will
have
more
progress
and
I
would
be
definitely
more
than
excited
as
soon
as
we
have
the
beta
announcement,
because
sometimes
it
becomes
the
checkpoint.
Also,
though
community
you
guys
are
putting
so
much
effort,
but
whenever
people
say
oh
php
is
still
not
in
beta
state.