►
From YouTube: 2023-02-15 meeting
Description
OpenTelemetry PHP SIG Meeting
A
A
A
A
I'm
wondering
do
you
work
on
this
open,
Telemetry
and
things
during
your
work
time
or
during
your
let's
say,
slack
time
or
free
time,
but.
B
A
Okay;
okay,
because
you
know
I
got
time
from
my
companies,
so
I'm,
currently
working
at
my
work
time
on
on
this
on
these
things,
yeah.
B
I
I'm
very
happy
to
have
you.
Have
you
working
I
didn't
realize
that
you
worked
at
Sumo
logic.
A
Yeah
I'm
at
simulating
since
last
year-
oh
cool.
Where
were
you
before
that?
Before
that
I
was
working
for
three
plus
years
at
TomTom,
so
TomTom
is
I,
don't
know
if
you
know
the
company
is
it's
companies
that
working
on
navigation
systems.
B
A
Yes,
that
that's
right
and
and
before
Tom
tomay
was
also
working
for
more
more
than
decade
at
the
diner
Trace.
So
I,
don't
know
if
you
know
dinotrace
yeah
yeah,
it's
also
APM
company,
so.
A
Okay,
okay,
so
the
to
it
is
an
owner
of
of
your
company.
That's.
B
A
B
That's
that
changed
yeah
that
changed
in
November
of
last
year.
So,
like
you
know,
14
months
ago,.
B
C
B
Fixes
and
texts
from
metric
exporter,
Factory
Brad.
This
is
you.
C
Yeah,
yes,
this
replaces
that
one
of
the
base
generator
for
about
a
month.
So
it
speaks
to
me
and
you
see
it,
okay
for
it
as
well.
So.
B
This
is,
this
is
ready
for
review.
Yes,
it
is
okay,
I'll
review
this
after
today.
After
this
meeting.
C
B
Cool,
so
we
got
two
two
ready
for
review
for
man
put
them
in
the
queue
all
right:
contribs
get
close
to
100
100
contributions;
that's
cool,
nothing
open
to
instrumentation
throughout
any
new
open.
Bugs
me
at
test
casing
for
power,
PC.
B
A
A
So
so
I'm
wondering
if
we
need
to
hook
any
other
methods
regarding
PDL,
because
currently
we
are
supporting
nine
methods,
I
think
the
most
important
ones,
but
I
don't
know
if
you
think
that
we
need
to
support
any
other
methods.
B
B
Maybe
maybe
I
guess
my
perspective
here
is
if
we
feel
like
these
are
the
most
important
methods
to
hook.
Currently,
maybe
we
should
just
run
with
that
until
somebody
requests
additional
methods
to
be
hooked
because
I've
they
seem
like
it's,
the
nlpd
or
exec
is
like
that's
a
huge,
a
huge
part
of
it
and
the
construction
and
the
pr
and
the
preparation
and
the
preparation
like
all
of
these
are
like
the
normal.
B
A
C
B
A
And
so
in
the
meantime,
I
started
on
another
tool
developer
tool,
because
my
vision
is
that
I
don't
want
to
spend
hours
in
debugger,
trying
to
figure
out
what
functions
are
called
so
I
read:
I
written
simple
extension
that
that
dumps
each
every
function,
that
is,
that
is
called
and
present
that
in
the
kind
of
Stack
trace
or
or
contrast
flow
graph,
so
you
can
see
all
the
methods
that
are
called
and
there
are
specific.
There
are
relationships
between
parent
and
child
methods.
A
A
Stack
Trace
that
there
is
a
in
fact
a
lot
of
stacked
traces
here
in
this
file
because
because
I'm
generating
each
stacked
race,
every
time
method
is
called
but
and
now
I
I
would
like
to
merge
them
and
to
to
build
one
huge
structure.
So,
as
you
can
see
here
and
the
bottom,
we
have
handle
that
the
main
method
and
then
at
the
top
we
have
full,
which
is
the
last
one
that
is
called
here
and
I
was
wondering.
A
I
was
thinking
about
some
kind
of
tool
that
where
we
can,
for
instance,
select
specific
methods
here
or
functions
from
from
this
stacked
trace
and
generate
a
kind
of
template,
hooking
template
and
then
a
developer
or
user
can
get
it
an
extent
and
so
on
it.
It
would
be
I
think
useful
for
some
kind
of
Rapid
development
and
trying
to
to
to
look.
You
know
what
how
this
spans
will
look
like.
A
So
so,
I'm
wondering
what's
your
opinion
or
on
something
like
that.
I
I
did
it
also
for
Symphony
for
Symphony,
laravel
and
and
Drupal,
and
it
works.
So
the
next
step
would
be.
You
know
generate
this
template
that
I
mentioned.
C
B
A
B
B
A
C
The
code
to
generate
to
entry
could
help
us
write,
Auto
instrumentation
a
lot
faster.
I,
certainly
see
that
that
benefit
of
the
the
usefulness
or
something.
A
Mm-Hmm,
okay,
so
basically
that's
it
from
my
site.
So
I
will
try
to
you,
know
to
build
the
minimal
available
Pro
product
and
to
maybe
on
next
Mythic
I
can
I
will
try
to
present
that
how
it
will
work
end
to
end.
C
One
of
them
is
whether
we
can
have
a
configuration
setting
for
the
verbosity
of
our
order
of
instrumentation,
so
I
mentioned
before
PDO
fetch.
This
gets
caught,
lots
and
lots
of
times
in
WordPress
database.
Query
execution,
it's
called
you
know.
100
times
like
there
are
a
couple
of
things
that
are
really
you'd
only
probably
want
that
level
of
tracing.
C
If
you,
if
you
were
debugger-
and
you
probably
don't
want
it
on,
you
know
ordinarily,
so
I
really
want
to
start
thinking
about
a
configuration
where
we
can
control
the
the
level
of
logging
that
we
generate
out
of
order
of
instrumentation
I
haven't
thought
much
Beyond
wanting
to
do
that.
But
but
that's
something
you
know
maybe
I'll
create
an
issue
for
it
and
yeah
prototype
something
up
to
to
see
if
I
can
get
into
the
book
and
the
other
thing
that
I
would
be
like
to
think
about.
C
Although
it's
quite
a
hard
problem
is
asynchronous,
sending
call
sometimes
tree
so
when
we
send
traces
and
metrics.
Now
we
just
are
saying
our
batch
exporter
when
we
end
a
span,
we
check
you
know,
has
it
been
five
seconds
since
we
last
said
something?
If
yes
then
send
it?
But
if
we're
not
actually
doing
anything
say,
we've
got
a
long
running
application
server
and
the
requests
have
come
in.
Then
then
we
don't,
we
don't
send.
C
Yes,
that's
something.
I
would
like
to
look
at
at
least
enough
to
know
whether
we
would
need
any
API
changes
to
implement
it.
I
I
know
that
Tobias
has
looked
at
this
in
the
past.
It
sort
of
has
something
that's
something
that
works
is
private
well
in
his
own
area,.
C
Yeah
but
it's
it's
yeah,
it's
a
copy
of
the
batch
Explorer
with
extra
stuff
in
it
and
extra
dependencies
and
it
only
works
in
phpa
point
one
or
eight
point
two,
because
it
means
it's
fibers.
So.
C
I'm
not
sure
what
to
do
about
it.
Yet
yeah,
listen
to
think
about
it
more
and
maybe
see.
If
I
can
foreign.
A
Okay
regarding
this
first
topic
that
you
mentioned,
so
it's
a
configuration
that
you
would
like
to
control,
for
instance,
number
of
spawns
that
will
be
generated
right.
A
Yes,
because
there
is
a
as
a
topic
that
is
I,
think
connected
to
it,
because
recently
maybe
I
shared
the
screen
for
a
moment
when
I.
C
A
Again
so
Severin
Neumann
opened
a
kind
of
issue
for
for
specification,
which
is
about
log
level
for
spans,
and
he
also
I
mentioned
here
our
discussion
that
we
had
some
time
ago.
A
C
C
It
is
a
generic
problem,
yes
and
I.
Think
no
one's
really
no
one's
really
tackled
before,
except
for
maybe
rust
what
was
pointed
to,
but
but
they
do
in
a
very
rust-like
way
like
in
a
non
non-generic
reusable
way.
A
A
C
Yeah,
that
would
that
would
be
useful,
because
I
mean
I
would
imagine
we
could
drop
that
into
our
SDK
fairly,
simply
and-
and
you
know,
interrogate
that
as
part
of
bootstrapping
and
auto
instrumentation
and
that's
as
simple
as
I
thought
it
would
be
at
the
moment.
You
know,
there's
there's
a
thing
that
we
know
it's
just
very,
very
noisy
such
as
PDO
fetch.
C
Yes,
yes,
so
you
know,
if
we're
not
at
debug
level,
then
just
go.
Don't
run
that
particular
Observer.
A
C
C
Three
yeah,
nothing
else
for
me,.