►
From YouTube: 2020-05-14 Python SIG
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
A
A
A
C
I'm
I'm
Conners
host
for
the
internship,
oh.
A
C
C
A
D
D
E
F
E
D
I
did
that
last
time,
oh
good
and
then
yeah
I
was
planning
to
return
to
school
this
term,
but
I'm
here.
Instead,
it
is
probably
better
than
remote
school,
so
yeah.
So
you
better
like
off
stream.
Now
right,
it's
CS,
there's
streams
that
really
exist.
Oh
really,
oh,
it's
just
chaos!
Yeah!
You
can
do
it
ever.
You
want
basically
yeah
I,
guess.
E
D
D
F
B
F
B
B
B
A
A
E
E
A
A
B
E
B
Yeah
that
that
didn't
go
as
well
as
I
wanted
it
to
go,
be
honest,
I.
It
took
way
longer
and
it
broke
about
halfway
through,
which
was
pretty
pretty
unfortunate,
because
there
was
no
way
of
rolling
back
either.
So
we
there
was
a
bunch
of
pipe
or
packages
that
were
out
for
0.70
for
a
little
while
that
would
have
been
very
sad
if
someone
had
tried
to
install
them.
I
think.
B
I
mean
I,
it
was
probably
only
not
to
do
a
release
when
neither
you
or
Chris
were
was
around
so
I'll
I'll.
Think
the
blame
on
that
one,
but
otherwise
I
mean
it
was.
It
was
fine.
I
think
the
other
part
I
was
painful,
was
just
the
amount
of
time
it
took
for
any
new
builds
to
get
through
like
it.
It
takes
what
like
between
10
and
13
minutes,
bro.
G
B
A
You
know
this
is
some
potentially
low-hanging
fruit
if
we're
looking
for
like
intern
projects.
So
right
now
we're
looking
at
these
like
10
15
minute
build
times,
but
when
we
run
the
tests
locally,
they're
really
fast.
So
we
have
like
a
few
tests
that
are
slow
because
they're,
you
know
like
locking
waiting
for
a
server
to
come
up
or
down
or
installing
dependencies,
and
it's
possible
that
we
can
speed
this
way
up.
Just
by
like
running
more
jobs
or
caching,
some
dependencies
or
excluding
some
tests.
B
Yeah
there's
so
I
started
poking
around
the
build
last
night.
Just
for
this
for
fun,
because
optimising
builds
is
what
I
do
for
as
a
hobby,
now
yeah
there's
so
there's
a
handful
of
things
like
we
could.
We
could
easily
split
out
the
lint
darker
tests
and
darks
generation
like
out
into
their
own
tasks,
and
they
all
have
to
be
part
of
the
Python
3/8
runtime
and
that's
a
is
potentially
like.
You
know
three
to
four
minutes
on
on
that
bill.
B
B
B
A
B
Think
there
was
the
only
thing
that
I
wanted
to
catch
was
there
was
a
couple
of
steps
that
I
ended
up
having
to
do
manually
like
updating
the
stable
tag,
wasn't
documented
anywhere
so
I'm
gonna
document
that
and
I
want
to
document.
I
also
want
to
update
the
automation
script,
to
update
the
setup
and
configuration
for
dependencies
as
well,
so
so
doing
those
two
things
and
creating
an
actual
release.
Markdown
doc,
like
they
have
in
the
JavaScript
Sega
thing,
will
be
really
helpful.
B
B
B
D
Yeah
sure
so
this
is
just
letting
auto
instrumentation
work
with
the
Postgres
library,
and
so
it's
a
little
also.
It
actually
uses
all
of
the
DB
API
stuff,
as
best
as
possible.
I
had
to
make
some
changes
to
get
it
to
work.
So
I
I
would
like
a
in-depth
review
here,
because
I
think
there's
some
pythonic
stuff
that
could
be
made
better,
but
it's
not
that
complicated.
D
A
D
E
E
But
Diego
and
Mauricio
were
doing
Alex
where
we're
doing
like
some
pretty
good
work
for
that.
A
lot
of
the
magic
of
I
didn't
I,
think
Django
a
flask
of
my
sequel
way
of
Pi,
my
sequel
now
yeah.
B
B
B
B
A
B
E
B
E
B
E
B
E
Another
PR
by
it
was
a
waste
or
something
about
the
you
know
automatically
finding
the
libraries
that
are
installed
think
he
meets
more
reviewers.
For
that.
E
B
B
E
B
B
C
B
Love
to
get
more
more
thoughts
on
this.
This
is
just
around
having
mechanism
for
users
to
configure
collecting
or
generating
metrics
from
their
systems.
So
this
is
like
CPU
memory,
metrics
Network
metrics.
So
if
you
thoughts
are
comment
on
this
I'd
love
to
do
them,
yeah.
A
I
know
they're
doing
about
your
work
in
the
like
the
collector
for
this
too,
because
you
know
in
general,
like
if
you're,
using
the
the
Python
client
it's
running
in
some
Python
application,
and
so
you
can't
really,
if
you
just
wanted
to
collect
system
metrics,
you
can't
deploy
the
Python
client
on
its
own
right
right,
but
the
the
collector
kind
of
fills
this
niche,
because
even
if
you're
not
using
it
to
do
anything
like
you
know,
like
tail
sampling,
you
just
have
it
running
in
and
scooping
out
system.
Metrics.
E
A
I
mean
we
should
probably
maybe
this
is
too
early
to
do
this
now,
but
we
should
probably
plan
to
have
the
same
semantics
conventions
as
the
collector
because
they
have
to
decide.
You
know
how
to
represent
all
these
system.
Metrics
right,
you
need
to
do
a
name
for
things
like
you
know,
bytes
in
memory
and
yeah.
E
Yeah
yeah,
it's
good
that
we
don't
have
to
be
like
developing
it
in-house
and
it's
like
yeah.
Now
we
have
some
sort
of
standard.
I
can
follow
so
yeah
I.
A
B
Yeah,
basically,
this
this
working
amount
of
the
same
kind
of
thing
that
Layden
you
guys
are
doing
on
the
exporter.
We
we
have
the
same
like
goal
of
having
metrics
being
sent
from
our
exporters
as
well,
so
it
seemed
like
it
would
make
more
sense
if
there
was
a
like
a
community-wide
mechanism
for
collecting
those
metrics
and
then
like
whatever
you
need
to
do
to
translate
those
you,
you
can
do
it
on
ingest
or
whatever,
but
yeah.
B
Yeah
one
one
note
that
was
kind
of
fun
to
read
this
morning
was
from
someone
on
the
like
the
salary
maintainer
who's
chiming
in
on
the
instrumentation
for
salary,
which
I
think
is
pretty
exciting.
Apparently
they're
planning
on
adding
open
telemetry
directly
into
the
into
the
celery
framework.
But
that's
no.
B
B
F
B
D
B
D
F
That's
in
the
PR,
that's
supposed
to
be
merged
into
the
spec
that
the
active
span
there
can
always
only
be
one
and
there's
no
active
spam
for
tracer,
and
so
by
doing
that,
it
vastly
simplifies
those
sort
of
get
current
span
operation
where
we
can
just
lift
it
to
a
top
level
API.
So
that's
what
this
PR
does,
and
so
the
other
data
point
I
want
to
throw
out
is
by
doing
this,
we
could
also
simultaneously
deprecated
well
with
we
can
deprecated,
certainly
without
any
impact
of
the
user.
F
F
F
List
to
actually
the
active
span
today
is
actually
a
single
is
a
singleton,
the
wrong
word,
but
the
active
span
is
there's
only
one
active
span
ever
in
the
case
of
a
context,
so
I'll
find
another
PR
I
did
that
so
yeah
give
me
a
sec,
I'll
pulled
up,
but
I'll
pause.
If
anyone
has
any
questions
or
comments.
E
So
just
users
would
be
interacting
with
the
span
just
through
the
context.
Instead
right.
F
Exactly
okay
and
the
reason
we
do
that
is
because
you
don't
want
to
like
it's
nice
to
encapsulate
and
not
rely
on
people
to
somehow
pull
in
the
global
key.
That
I
mean
you
could
also
clip
pull
in
the
global
can
do
the
get
as
well,
but
I
think
it
currents
fan
encapsulates
things
a
little
bit
better.
Yeah
cool
makes.
E
F
E
E
A
A
To
do
and
Josh
is
like
neck
deep
in
the
oh
four
spec
rewrite,
but
any
work
that
you
do
now
on
views
will
probably
inform
the
views
spec
so
I,
don't
think
you're
you're
blocked
by
the
spec
work,
but
I
mean
this
bad
work
is
gonna
have
to
happen
whether
or
not
the
users
complete.
That
makes
sense.
Yeah.
A
E
You
think,
in
terms
of
like
priority.
E
A
E
E
A
F
Hey
sorry,
rerouting
to
sort
of
my
current
span.
Pr
again,
does
anyone
have
thoughts
on
getting
rid
of
tracer
tracer
tracer
I'll
get
a
current
span
which
does
break
a
bunch
of
stuff,
but
again
I
feel
you
know
I
guess
I
feel
like
eliminating
any
I
guess
like
legacy
API.
Is
this
really
important
before
the
one
point?
Oh
yeah.
G
F
D
F
That
is
because
traced,
I'll,
update
name
is
a
discouraged
API
because
it
impacts,
sampling
or
head-head
based
sampling,
where
you
may
want
to
basically
look
at
the
span
name
and
choose
whether
you
want
to
disable
that
span
and
so
doing,
update
name
after
the
fact
requires
additional
post-processing
and
potentially
recovering
a
span
that
basically
doesn't
exist
anymore
or
that
was
chosen
to
be
sampled
away.
So
there's
complications
in
that
and
I
have
not
been
able
to
push
that
discussion.
F
So
but
so
I
guess
my
my
point
is
my
use
case.
It's
a
little
bit
invalid
right
now
or
my
need
to
modify
an
existing
spam.
But
that
said,
I
think
it's.
You
know
it's
there's
good
use
cases
for
modifying
or
amending
in
the
current
span
that
you're
running
in
without
creating
one
yourself
good
use
case.
Yeah.
F
So
I
tell
people
it's
use
case
without
the
s
yeah
yeah.
Let
me
see
yeah
I'll
link
the
discussion
around
trace
to
update
name
in
this
thing.
Okay,
so
I
guess
just
to
kind
of
round
out
this
conversation.
If
there
is
support,
do
I.
Does
anyone
want
to
say,
nay,
right
now
to
potentially
breaking
and
pretty
much
everyone
who's
using
get
current
span
right
now,
which
is
hard
to
verify
and
remove
that
API.
F
B
All
right,
yeah,
I
guess
unless
anyone
has
anything
else
to
talk
about,
we
couldn't
get
this
short.
Okay,
great.