►
From YouTube: 2021-02-25 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
It's
fine
yeah,
so
you
have
a
chance
to
meet
so
I'm
robert
from
splanc
I
have
joined
in
in
february,
basically
so
just
starting
yeah.
What
are
you
working
on
on
the
instrumentation
of
basically
on
the
stuff
that
we
are
here?
That's
my
main
responsibility
right
now
to
make
those
stuff
pretty.
I
don't
know
right
now.
The
task
which
I
will
be
working
on
is
to
support
this
open,
telemetry,
semantics.
B
B
So
I
don't
I'm
from
you
know
I
work
at
microsoft
on
the
profiling
apis.
I
don't
know
if
you
know
that,
so
that's
what.
C
So
I
see
dave
and
robert
already
met
yeah.
D
C
C
First,
is
it's
very
nice
to
see
the
asp.net
core
migrating
to
active
source
and
taking
the
hotel
stuff
in
consideration,
so
following
the
the
practice
and
recommendations
for
hotel
regarding
a
lot
of
the
conventions,
you
know
so
linking
this
here,
and
this
is
something
that
I
mentioned
some
time
ago,
when
dave
had
had
a
chance
of
looking
at
this
in
the
collector
repository,
and
it's
very
good
for
us
that
are
doing.net,
because
we
can
use
the
collector
to
collect
metrics
and
actually
opens
the
potential
for
a
lot
of
things
that
we
can
collect
via
the
diagnostics
dave
people
from
the
collector
ask
me
for
the
following:
if
you
would
like,
if
you'd
like,
if
you
would
like
to
be
kind
of
not
in
what
they
call
component
owner,
you
don't
have
any
responsibility
besides
reviewing
code,
if
people
propose
changes,
would
you
be
okay
with
that,
because
then
it's
basically
letting
your
name
there?
B
So
is
it
that
means
I
just
basically
get
a
github
notification
anytime,
a
pr
is
opened.
C
C
So
this
is,
let
me
just
to
to
to
capture
generic.
Let
me
start
a
little
bit
with
the
collector.
The
collector
has
the
concept
of
generic
components
like
receivers
and
exporters,
and
this
is
a
receiver
that
can
capture
the
net
metrics
emitted
via
event,
pipe.
C
Yeah,
the
the
the
need
for
that
kind
of
comes
from
people
that
are
migrating
their.net
applications
to
linux
containers,
and
then
we
don't
have
the
windows
performance
counters.
The
collector
already
has
the
capability
of
collecting
the
windows
performance
counters,
but
nothing
on
linux,
so
this
is
kind
of
a
barrier
for
some
people
that
want
to
switch
to
linux
containers
and
with
this
feature
at
least
the
counters
we
start
to
have.
C
But
dave
knows
much
better
than
me:
there
is
the
potential
for
much
more
stuff.
You
know
we
can
get
the
activities
we
potentially
could
get
call
stacks
and
a
lot
of
stuff.
B
E
I
was
curious,
so
there
was
talk
about
dotnet
team,
considering
doing
something
like
activity
for
metrics,
where
there
is
a
full-fledged
metrics
api,
that
you
know
that
where
people
can
kind
of
produce
metrics
and
then
they
go
into
nothing
if
nobody
is
listening,
but
then
they
go
into
somewhere
and
they
wonder
whether
this
this
is
considered
as
such
api
like
whether
this
is
even
related
with
what
what
what
you're
doing
here
and
that
stuff,
because
maybe
that
api
would
be
like
in
this
shape
or
different
shape.
E
I
don't
know
what
do
you
mean
by
metrics?
Like
so
say,
I
have
custom
metrics
or
like
standard
metrics
like
could
use
the
same
mechanism
as
they
want
to
say.
I
have
custom
metrics
and
application
emits
metrics,
just
like
it
emits
activity
is
a
sort
of
a
log
kind
of
thing
right,
so
it
emits
activity
and
then
activities
are
very
cheap.
E
If
nobody
wants
to
record
activities
and
then,
if
somebody
registered
registers
who
wants
to
record
activities,
then
you
pay
a
little
performance
price
and
then
you
can
record
them
and
persist
them
right,
and
there
was
I
I
heard
a
few
times
that
dotnet
is
considering
a
similar
mechanism
api
for
metrics,
and
I
wonder
whether
there
was
any
like
what
what
happens
here
has
any
relation
to
those
considerations.
B
So
I'm
not
actually
familiar
with
any
of
the
apis,
I
mean
it
could
be
happening
but
yeah,
but
I
haven't
heard
of
it
and
but
the
the
thing
I
will
mention
is
once
you
have
this
event
pipe
reader.
Is
it's
really
dead,
simple
to
to
use,
to
define
an
event
pipe
and
manage
sorry,
an
event
source
and
manage
code?
And
then
you
can
listen
to
it
over
the
event
pipe,
and
so
you
can.
You
can
create
your
own
custom.
B
E
Yeah
yeah
totally.
No,
I
I
just
wondered
whether
it's
related,
because
if,
if,
if
it
was
then
whoever
is
focusing
on
those
apis,
probably
have
some
sort
of
opinion
about
you
know
what
are
the
metric
types
like
how
to
think
about
matrix
types
correctly
and
how
to
think
about
metric
aggregation
correctly
and
those
topics.
C
The
the
the
hotel
sdk
had
been
doing
some
experimentation
around
metrics,
but
and
this
in
the
same
way
that
happened
with
activity
source.
It's
going
to
happen
for
matrix,
it's
going
to
kind
of
move
to
that.net
runtime,
so
people
can
have
a
generic
way
of
dealing
with
the
metrics.
That's.
C
E
Please
I'm
interested
in
this
just
kind
of
thank
you.
C
All
right
moving
to
the
other
topics
about
the
prior
release.
C
I
think
this
week,
robert
and
david-
that's
not
here
with
us
today,
probably
sleeping
there
in
poland
or
estonia,
I'm
not
sure,
but
both
both
guys
did
a
lot
of
work
regarding
kind
of
build
building
on
mac,
improving
the
the
the
ci
stuff-
and
I
think
the
the
most
important
work
regarding
to
kind
of
for
us
to
have
the
pre-release
is
a
pr
that
is
still
whip
about
generating
the
files
with
the
the
proper
names
that
we
can
separate
and
avoid
conflicts.
C
I
think
still
need
the
snk
and
idea
updates
this
kind
of
thing,
but
I
think
that
is
the
first
kind
of
milestone
for
us
to
have
something
that
we
can
build
and
use.
That's,
not
really
the
data
dog.
It's
really
have
the
name
and
things
off
open
the
limit.
E
You
mean
the
the
tags
and
everything.
C
That
is
just
for
basically
files
class
id
guides,
nothing,
nothing
like
name
space
types.
This
is,
I
want
touch.
A
E
Yeah
about
that,
I
I
so
my
read
it
on
it
and
zac.
Please
correct
me:
if
I'm
wrong,
also
from
from
what
the
the
other
folks
on
the
team
said
kevin
and
everybody,
I
think
it
looks
good.
There
was
a
bunch
of
comments
on
on
on
the
issue
where
I
describe
the
approach.
Please
consider
those
comments
I
think
they
they're
spot
on.
I
haven't
actually
dug
into
depth
in
there.
E
So
please,
please
like
consider
them
as
you
implement,
but
I
think
overall
we're
good
to
go
to
start
there.
How
did
it
touch
you.
A
D
E
E
So
I
think
yeah,
if
you,
if
you
do
that,
plus
one
integration
to
show
how
you
do
it
and
let's
make
sure
that
the
data
don't
force
review
this,
so
that
you
don't
have
any
throwaway
work
so
that
they're,
okay
with
this
before
we
proceed.
C
Yeah,
so
moving
to
the
next
thing
I
had
somebody
that
is
not
participating
on
the
sig.
Take
a
look
at
this
stuff
there
and
we
are
not
making.
This
kind
of
this
is
work
that
we
just
mentioned
kind
of
feasible
in
the
readme
and
the
docs,
because
we
still
have
the
old
world
map.
So
I'm
going
to
be
trying
this
week
to
update
those
who
reflect
this
kind
of
work
that
we
are
we
are,
we
are
doing.
C
I'm
gonna
try
to
update
the
board,
I'm
no
pm
at
all,
but
I'm
gonna
try
to
update
to
reflect
also
that
work
like
some
time
ago.
Greg
did
a
dashboard
for
the
project.
I
think
I
will
take
a
look
and
try
to
build
on
top
of
that
and.
E
If
you
like,
drop
me
a
half
an
hour
into
the
calendar,
and
you
can
do
it
together,.
C
Cool
cool
I'll.
Do
that,
and
I
don't
know
chris
eric
sometimes
ago,
was
looking
in
perhaps
helping
us
with
pm
work.
If
he
still
can
do,
perhaps
now
is
a
moment
that
he
can
help
us.
D
Yeah,
so
I
can
reach
out
to
him.
I
I
believe
he's
going
to
be
taking
a
leave
of
absence
shortly
for
a
couple
of
months,
and
so
I
forget,
when
that
starts
so
I'll
have
to
I'll
have
to
look
into
that.
E
C
Yeah-
and
in
light
of
that,
I
was
thinking
that,
since
you
are
going
to
have
this
work
kind
of
in
line
for
for
this
prayer
release,
I
was
thinking
perhaps
in
reaching
the
maintainers
meeting
because
to
see
if
there
is
other
or
other
vendors
or
people
that
want
to
kind
of
get
involved
and
participate
on
the
project.
So
we
have,
I
think
we
have
kind
of
a
momentum
to
kind
of
to
that
kind
of
invite.
C
You
know
so
I
was
thinking,
perhaps
not
maybe
next
week
or
the
week
after
go
on
the
maintenance
meeting.
I
will
go
there
and
kind
of
give
these
status
and
ask
if
more
people
want
to
join
the
project
and
contribute.
E
I'm
just
curious.
I
never
went
to
that
meeting.
So
can
you
let
me
know
when
you
go,
I
I
it
might
be
interesting
to
join
just.
C
All
right
all
right,
yeah,
it's
just
a
meaning
of
maintainers,
of
different
projects,
and
that
is
the
kind
of
place
that
people
from
different
projects,
change,
exchange
the
ideas
and
common
problems
and
common
solutions.
So
sometimes
people
ask
for
this
kind
of,
let's
say,
manpower,
help
for
specific
things.
C
Then
I
have
a
if
you
have,
if
you
don't
have
it
greg's,
totally
obvious
fine
about
diagnosed
source.
If
you
have
an
update
or
want
some.
E
E
E
So
I
may
or
may
not
be
done
by
friday,
but
if
I
I
think
I
will
be
and
if
not,
then
maybe
I
just
because
I
started
a
day
late,
working
on
other
work
related
things
I
might
just
like.
If
it
spills
over
into
monday,
then
it
goes
over
to
monday.
E
So
for
the
for
the
for
the
specific,
the
the
part
that
I'm
working
right
now,
it's
because
I'm
in
this
hackathon
mode
and
I'm
halfway
through
so
this
is
like
we
can
chat,
but
this
was
like
basically
only
this
week
and
only
like
if
you
have
a
lot
of
time,
but
once
I'm
done
so,
then
for
sure
there
is
a
lot
of
times
that
you
can
show
cool
cool.
E
E
As
a
side
note,
this
was
kind
of
my
own
hacking
thing.
I
spent
a
lot
of
time
working
on
the
logging
thing
that
that
we
have
the
tracer
users
like
what
it
uses
and
like
the
that
look
team
is
doing
some
in
incremental
improvement
into
the
way
the
tracer
logs
stuff,
but
for
other
components
like
the
different
libraries
and
that
we
have
in
an
all
over
datadocker.
E
I
have
created
this
little
thing
that
allows
you
to
have
this
logging
in
direction
where
you
don't
take
dependencies
on
some
sort
of
logging
library,
but
you
have
essentially
callbacks
for
all
sorts
of
vlogging
and
then
sync
for
it
which
gets
rid
of
the
of
the
rendered
in
log.
What
is
it
siri
log
I
think
in
the
traces
or
and
then
all
the
new
stuff,
I'm
doing
I'm
using
that
I'd.
C
C
E
The
the
the
idea
is
that,
as
we
kind
of
modularize,
we
end
up
in
a
situation
where
we
have
a
bunch
of
a
bunch
of
dlns
and
they
all
produce
logs,
but
then
they
they
come
together
into
a
in
a
product
and
then
different
vendors
might
wanna
actually
persist
the
logs
differently
so
rather
than
everybody
logging
to
some
sort
of
file,
and
then
the
vendor
going.
Oh,
my
god
now
I
have
to
I
don't
like
this
file
logging.
I
want
a
different
file
logging.
D
Greg
I'm
curious.
Have
you
had
different
use
cases
where
you
wanted
to
log
things
differently,
based
on
what
was
being
monitored.
D
Yeah
so,
for
example,
let's
say
you
were
running
in
a
containerized
environment,
did
you
want
to
output
logs
differently?
In
that
case,
is
that
kind
of
what's
driving
this
change
or
or
has
there
been
some
other
user
need.
E
So
the
answer
is
yes,
but
that's
not
what
driving
this
change
so
for
now?
Actually,
I'm
not
going
to
apply
this
change
to
existing
tracer
components
for
now
I'll
just
apply
it
to
the
new
ones
and
then
later
we
can
decide
whether
or
not
we
like
it
enough
to
apply
it
everywhere
or
not
in
that
which
case
we
don't.
So
I
don't
want
to
create
churn
with
this,
but
to
answer
your
question
directly.
E
Different
scenarios
can
require
different
things,
so,
for
example,
in
in
the
standard
that
not
case
belong
to
a
file,
but
in
a
container
you
sometimes
want
to
prefer
to
log
just
std
out
or
something
because
like
there
is
no
file
that
is,
as
such,
you
know,
container
dies
five
goals
so,
but
as
long
as
the
product
is
monolithic,
that
decision
can
be
taken
using
today's
system
right
today,
the
tracer
logs
to
file
it
can
just
decide
to
lock
somewhere
else.
E
So
the
what
drives
this
technology
is
that
the
product
becomes
less
monolithic.
So
if
we
now
have
imagine
a
bunch
of
live
libraries,
some
of
them
are
vendor
specific.
Some
of
them
are
not,
and
even
within
data
dog
if
we
have
different
products
right.
So,
for
example,
if
you
have
like
a
core
product
and
then
you
have
some
customers
this
function,
some
customers
have
that
functionality
and
some
customers
have
both
right.
So
that
means
now.
E
You
have
essentially
different
dlls
and
all
of
them
want
to
do
logging,
but
exactly
that's
the
thing
they
I
I
wanna
have
them
not
be.
I
don't
want
those
dlls,
not
a
reason
about
exactly
the
problem
you
just
described.
Where
am
I
actually
running
in
a
container
or
am
I
running
er,
where
file
logging
is
better
or
something
else
right?
All
of
those
deals
just
focus
on
what
they
do
and
I
want
them
to
produce
logs
and
then
the
guy.
Who
is
the
main
entry
point?
E
He
should
be
the
only
one
who
is
composing
all
of
these
libraries
and
he
should
be
the
one
who
decides
whether
the
log
output
is
a
file
or.
A
A
sorry
to
interrupt
you,
this
quick
question:
what
did
what's
the
difference
between
abstraction,
login
or
libloc,
which
was
created
a
few
years
ago.
E
I
obstruction
so
basically,
this
is
in
this
case.
I
all
I
wanted
to
do
is
not
take
dependencies
on
any
libraries,
so.
F
Of
no
yeah,
we
use
that
already
in
this
project.
I
think
the
thing
is
we
still
to
do
our
own
logging.
We
do
vendor
in
ceralog
and
that's
just
the
de
facto
one,
but
if
we
have
to
break
up
the
assembly
into
further
components,
I
think
what
greg
is
trying
to
say
is
like
we.
We
don't
want
to
introduce
another
third-party
dependency
like
depending
on
like
yeah,
ceralog
or
anything
or
whatnot.
F
So
I
guess
we'd
be
doing
our
own
small
abstraction
just
to
abstract
away
the
implementation.
E
So
right
now
it's
about
it's
a
file
that
is
about
100
150
lines.
That
is
included
in
that
every
library
who
wishes
to
take
advantage
of
this
includes
a
file
of
about
150
lines
and
that's
it
that
is
like
our
own
file.
E
And
then
there
is
some
more
files
for
libraries
who
would
like
to
for
the
one
library
who
does
the
output,
but
we
don't
have
to
use
it
for
now,
I'm
using
it
into
the
prototypes
that
I'm
doing
and
if
we
get
to
production
and
there
is
better
options,
we
can
always
replace
things,
but
I
spend
a
few
fun
days.
C
I
I
I
think
it
will
be
interesting
when
one
once
you
have
the
the
the
prototype
kind
of
take
a
look
in
one
of
the
meetings,
so
people
can
can
see
that.
E
E
I'm
using
it
in
the
diagnostic
stuff,
so
yeah.
C
The
next
topic
that
is
there
I
I
ask-
I
actually
asked
robert
because
both
robert
and
david
came
with
new
eyes
on
the
project
and
one
of
the
things
that
they
have
to
do
is
ci
and
building
and
running
tests,
and
they
they
had
a
lot
of
questions
and
ideas.
And
I
I
really
liked
this
idea
that
robert
presented
in
vipr
for-
and
I
asked
him
to
kind
of-
do-
a
presentation
for.
C
I
find
this
approach
a
very
interesting
to
target
the
the
integration
tests
so
robert.
Whenever
you
want
to
start
sharing
presenting,
please
go
ahead.
Okay,.
A
You
see
it
right,
we
can
see
it
yeah,
okay,
so,
first
of
all,
if
you
have
any
questions,
if
something
is
not
clear,
feel
free
to
interrupt,
if
you
do
not
understand,
probably
other
doesn't
others
do
not
understand
as
well.
That's
my
that's
how
I
think
I
feel,
and
also
I
might
talk.
I
know
too
fast
because
of
being
nervous.
I'm
also
english
is
not
my
native
or
mother
language
so
feel
free
to
interrupt
and
also,
if
you
prefer,
you
can
ask
at
the
end.
A
So
let
me
go
so
maybe
the
problem
why
I
started
making
this
kind
of
poc
whoa.
A
A
A
A
So
so
maybe
I
I
try
to
find
what
are,
in
my
opinion,
the
two
problems
for
both
of
the
questions
or
maybe
because
there
are
two
problems
that
for
me
work
hard.
So,
first
of
all,
there
are
some
kind
of
preliminary
steps
which
are
required
to
start
even
executing
the
integration
test,
so
you
can
run
them.
So
you
need
to
know
that
you
need
to
build
this
whole
tracer
with
the
profiler
to
put
it
in
the
correct
places.
A
So
the
integration
test
know
how
to
fetch
that
cetera
and
you
need
to
also
build
and
publish
test
applications
which
are
connected
with
the
integration
tests
and
also
you
need
to
make
sure
that
this
backend
services,
which
are
testing
those
those
integrations
like,
for
example,
service,
stack,
elastic,
etc.
I
need
to
be
up
and
running.
A
A
So
I
find
it's
a
kind
of
big
memory
pressure,
mostly
because
of
the
of
this
docker
compose
stuff,
because
you
need
to
have
all
of
these
services
up
if
you
want
to
run
the
whole
suit
and
in
my
opinion
this
may
cause
test
instability,
so
I
have
a
laptop
which
I'm
presenting
right
now.
It
has
16
gigs
of
ram,
and
I
have
checked.
Azure
pipelines
has
only
7
gigs
of
physical
ram
and
what's
this
more
interesting,
the
linux
machines
have
c
group
with
limit
to
six
gigs.
So
it's
even
worse.
A
So
this
is
a
screenshot
sorry
for
polish
polish
here
names,
but
I
think
you
get
what's
here
so
basically,
when
I'm
running
the
integration
test,
almost
all
my
ram
goes
consumed
and
yeah,
and
what's
more
even
on
my
own
machine,
the
integration
tests
sometimes
are
failing,
and
my
colleague
who
has
david,
who
has
32
gigs
of
ram.
He
said
that
he's
always
passing
for
him.
A
A
F
This
is
really
good
information
because
I
actually
never
encountered
that
issue
just
because
we
all
have
the
same
standard
issue
laptops
with
like
32
gigs
of
ram
over
here.
I
think
there
are
two
so
yeah
like
I've,
never
actually
run
into
the
situation
that
you've
encountered.
So
it's
really
good
to
hear
this
data
point.
A
B
A
A
So
I
have
may
I
have
like
time
books
my
my
time
to
two
days
not
to
spend
too
much
time
and
I
make
kind
of
poc
which
is
kind
of
working,
but
at
least
it's
something
which
you
can
look
at,
but
basically
I
was
focusing
on
spinning
up
the
container
that
will
be
needed
by
the
test
inside
it
as
itself.
So,
basically
the
test
when,
during
its
arrange
or
setup
step,
it
will
spin
up
the
container
and
during
tier
down
it
will
close.
It.
A
A
So
let
me
go
here
so
here
I
just
so
here,
basically
not
to
in
order
not
to
make
too
much
stuff.
I
just
tried
to
make
in
a
way
that
I
can
improve
the
test
one
by
one,
not
just
replacing
all
of
the
tests,
but
I
will
just
make
it
in
a
way
that
one
test
will
work
in
a
new
way,
but
the
other
one
will
still
work
in
the
old
way
and
that
this
docker
compose
run
integration
test
should
still
work.
A
So
I
have
basically
found
some
some
library
which
has
something
called
docker
fixture,
where
you
can
run
a
docker
compose
file
inside
xunit
and
during
setup
it
will
spin
up
this
docker
compose
and
during
third
tier
down
for
the
whole
test.
It
will
just
remove
everything.
So
this
is
the
whole
docker
compose
file
here
which
is
needed
for
this
test.
This
is
a
mongodb
test.
It
needs
just
the
mongodb
and
we
want
this
port
to
be
exposed.
C
C
I
think
everyone
here
is
familiar
with
this
fixture
model
to
do
the
that
therapeutic.
A
A
Basically,
this
is
running
inside
wsl
and
wsl
has
some
strange
behavior
that
it
doesn't
retain
the
it
like
whenever
it
comes
to
the
memory
it
keeps
it
forever.
A
So
this
is
how
much
this
one
docker
addition
this
one
additional
docker
container
this
and
the
test
run,
has
consumed.
It
was
only
seven
percent
of
my
16
gigs,
okay
and
right
now
I
will
run
the
second
part.
So
right
now,
if
I'm
just
make
sure
so
here,
you
can
see
that
no
docker
image
is
being
run.
So
now
I
run
docker
compose,
run
integration
tests
and
now
all
so,
you
can
see
that
here,
there's
no
mongodb.
A
A
A
A
A
A
A
C
So
the
the
the
dev
experience
is
it's
much
better.
I
think
you
are
planning
to
kind
of
touch
on
these
things,
but
the
dependency
about
the
docker
fixtures.
Where
does
it
come
from
how
how
reliable
is
released
or
things
like
this?
You
know
this.
This
is
my
first
question.
A
Because
we
have
time
I
was
thinking
about
technique
in
the
end,
but
now
we
have
a
good
time
to
because
working
so
basically,
I
simply
have
found
out
some
some
library
which
does
it.
But
it
is.
I
looked
at
the
code
and
it
looks
quite
okay,
but
when
I
wanted
to
import
it,
it
occurred
that
it
had
it
had
notes.
It
had
no
snake
key,
so
I
could
not
use
it
so
I
copy
paste
it
and
I
find
other
issues.
A
So
basically,
I
would
not
recommend
probably
using
it
probably
there's
a
there's,
a
net
sdk
for
docker,
so
this
is
probably
where
I
will
go
for
because
then,
if
you
run
it
inside
the
container
this
test,
you
will
not
even
need
to
install
the
docker
and
docker
compose
cli,
because
the
sdk
just
needs
the
socket
and
it
works.
A
A
A
I
have
looked
at
one
of
these
tests
and
basically
it
was
looking
dumb,
simple
and
that's
also
one
of
the
reasons.
I
think
that
the
the
the
ram
utilization
is
the
problem,
especially
that
one
of
the
tests
which
is
failing
is
like
this
http
handler
or
something
like
that.
So
it
was
not
even
spinning
up
or
using
any
additional
any
additional.
You
know
like
backup
container,
it
was
making
requests
to
example,
example.com,
or
something
like
that.
So.
F
Those
those
smoke
tests-
we,
those
are
just
supposed
to
be
like
really
quick
regression
tests
like
do
a
small
app,
throw
the
the
instrumentation
on
it
and
see
if
it
blows
up,
and
so
it
should
exit
quickly
so
yeah.
That's
it
probably
just
exceeded
a
timeout,
because
we
expect
it
to
finish
quickly
and
I
can
see
that
happening
with
lesser
rams.
A
Okay,
so
maybe
one
more
thing
which
I
will
show
here
so
just
I
will
maybe
quickly
show
how
it's
possible
that
or
maybe
I'll
show
the
presentation,
because
I
tried
my
best
to
explain
better
here.
So
this
is
how
basically
it
it
works.
An
illustration.
So
there's
a
unit
test
here
which
manages
the
container,
so
it
starts
it
and
it
runs
this
dll.
This
sample
mongodb,
for
example,
which
uses
this
container
right.
A
It's
side-by-side
point
integration
test,
and
and
because
of
that,
we
need
to
set
the
the
host
not
to
localhost,
but
to
hostdocker
internal,
which
is
something
which
is
working
on
windows
and
mac
os,
but
for
linux
machines
it's
not
yet
supported
out
of
the
box.
I
know
that
the
docker
team
docker
ng
team
needs,
wants
to
like
have
it
out
of
the
box.
Right
now
I
have
read:
there's
a
workaround,
which
is
that
you
can
simply
add
an
extra
host
which
points
to
host
gateway.
C
So
you
you
are
in
the
dev
container,
because
you
are
using
vs
code
remote
into
the
dev
container,
but
if,
for
instance,
I
could
use
these
on
windows,
I
could
just
call
the
test
and
the
text.
The
test
creates
directly
the
mongodb
and.
C
A
A
D
C
E
C
A
Yes,
you
could,
I
know
how
to
do
such
stuff,
but
it's
not
very
easy.
So.
F
I
haven't
even
attempted
some
of
that
because
running
the
x
units
just
starts
another
process,
so
you'd
have
to
you
know,
just
attach
a
new
debugger
to
that
process,
knowing
the
pid
and
stuff,
which
is
already
an
extra
step,
since
we
do
that
model
of
et's
unit
starts
new
process
to
run
the.net
app.
F
Yeah
yeah
locally,
usually
what
I
do
my
inner
loop
is
to
just
run
one
stamp
application
at
a
time.
A
A
Okay,
so
this
is
how
it
works
under
the
hood,
at
least
on
my
machine
when
I
was
presenting
I'm
using
a
docker
engine
because
I'm
using
wsl-
and
I'm
I
mean
it's
my
private
laptop,
I
do
not
want
to
mess
up
on
my
machine,
I'm
using
a
development
container,
so
this
stuff
was
even
docker
in
docker,
but
everything
was
run
in
the
engine
of
the
windows
and
I
present
in
it
no
digit
to
say
that
you
know
I'm
super
cool
extra.
A
A
I
know
so
that
was
my
mind
when
I
was
working
on
it.
So
maybe
some
conclusions
from
my
side.
What
are
the
good
parts?
Oh
man,
that's
ugly.
There
at
le,
I
have
removed
just
one.
A
I
removed
only
one
manual
step,
which
is
only
just
running
this
contain
this
container.
I
know
this
is
not
a
lot
at
all,
but
at
least
it's
something,
and
I
just
wanted
to
make
sure
that
you
know
if
it's
even
a
group
step
forward.
Basically,
I
checked
that
starting
a
container
just
takes
a
few
seconds
inside
like
three
seconds
on
my
machine,
and
I
think
that
it
was
taking
three
seconds
when
all
other
services
were
running.
So
probably,
if
I
have
no
other
service
running,
probably
it
will
be
even
faster.
A
So
I
think
that
when
you're
locally
running
it,
it
should
not
be
an
issue
to
wait
one
second
more.
In
my
opinion,
the
code
is
more
cohesive
because
you
don't
have
everything
in
one
docker
compose
and
you
need
you
can
have
stuff
side
by
side
and
this
approach
could
be
incrementally
adopted.
These
are
the
good
parts,
but
I
see
bad
parts
as
well.
A
First
of
all,
I
have
no
totally
no
idea
why
it
doesn't
work
on
azure
pipelines.
It
cannot.
The
test
cannot
cont
connect
to
to
the
mongodb,
even
though
I
see
the
ip
address
which
interiorly
should
be
properly
mounted
to
the
exposed
ports.
I
have
no
idea.
I
should
probably
test
it
on,
but
I
didn't
I
simply
type
box
my
I
have.
I
had
the
time
books
two
days
and
I
failed
to
to
just
simply
dig
any
deeper.
Maybe
it's!
Maybe
if
you
know
how
to
debug
stuff
in
azure
pipelines,
you
can
take
a
look.
A
The
the
bad
thing
is
that
probably
it
will
not
work
on
github
passion
as
well,
because
github
actions
that
azure
pipelines
like
share
the
same
machines
they
basically
they
have
the
same.
They
use
the
same
cloud
as
far
as
I
know.
A
The
second
bad
part
is,
I
think,
that
the
com,
I
think
that
this
complexity
of
running
docker
containers
from
docker
container
may
be
hard
to
troubleshoot
and
maintain
something
stop
working,
and
if
there
will
be
some
problem
there,
probably
some
developer,
which
is
not
familiar
with
it,
will
be
like
he
will
simply
lose
a
lot
of
time.
It's
different
when
you
troubleshoot
the
problem,
then
you
create
something.
I
know
that
troubleshooting
stuff
takes
more
time.
A
So
this
pr
is
a
dirty
poc,
so
I
as
as
you
apolo
ask
if
what
I
think
about
this
library,
I
think
it
would
be
better
to
use
docker.net
instead
of
docker
compose
inside
the
test.
There
are
also
other
libraries
which
are
even
like
dedicated
for
running
and
spinning
up
docker
containers
for
tests,
there's
one
like
popularity
called
test
container
but
in
my
opinion,
for
a
project
like
open
telemetry,
which.
B
A
A
I
think
that
automating
the
build
pipeline
using
something
like
nuke,
bullseye
or
sake
or
fake
or
cake
or
whatever,
would
bring
more
benefits
regarding
the
problem
of
preliminary
steps
and
basically
with
getting
into
the
projects
for
for
newcomers
and
even
for
development
on
daily
basis.
I
think
that
it
is,
it
will
be
a
lot
more
valuable
than
what
I
have
done
here
and,
as
far
I
know,
andrew
lock
from
data
dock
is
working
on
it,
and
I
really
hope
that
you're
keeping
finger
crossed
from
him
and
second,
is
that
having
tests
more
independent
is
great.
A
But
personally
you
know
this
whole
rams
thing
that
it
works
on
60,
gig
machines,
etc.
I
have
no
guarantees
that,
if
I'll
refactor
all
this
test,
it
will
be
any
better.
It's
a
guess.
I
think
it's
highly
probable
that
it
will
help,
but
it's
just
a
guess
that
you
need
to
empire
empire.
Empirically
check.
Sorry
for
my
english
and
one
thing
that
I
spotted.
I
have
no
idea
why?
A
What,
when
you
have
this
test,
when
why
the
we
are
running
the
sample
files
from
from
this
publish
directory,
not
just
using
a
dotnet,
run
to
run
from
cs
pro?
This
is
something
I
personally
do
not
understand.
It
will
at
least
remove
one
of
these
preliminary
steps.
You
know
what
I
mean
or
not,
really,
you
can
simply
use
dotron
on
cs
pro.
As
far
as
I
know,.
F
Got
it
yeah?
I
think
the
main
thing
that
we
do
in
our
our
integration
tests
is:
we
try
to
test
a
matrix
of
libraries
so
like
mongodb
testing
against
different
versions,
and
so
that
makes
it
difficult
because
if
you
just
do
net
run,
it's
just
going
to
run
with
only
one
version
at
a
time.
F
F
C
C
I
I
was
thinking
here
about
the
harder
to
troubleshoot
in
the
case
of
in
the
case
like
I'm,
on
a
windows.
Machine
using
visual
studio
is,
is
the
same
right.
It's
different
if
I'm
using
vs
code
with
a
dev
container.
A
C
Yeah
but
but
yes,
but
what's
gonna
happen,
I'm
gonna
the
test
is
gonna,
keep
the
the
docker
for,
and
I
can
run
any
of
the
platforms
that
I
want
right.
I
can
do
alpine
yep.
C
So
what
I'm
saying
is
that,
in
this
scenario,
is
really
norm
not
harder
to
debug
than
is
starting
a
docker
compose
and.
A
A
I
see
if,
for
example,
there
will
be
some
changes
in
the
docker
engine
that
this
thing
will
not
work
or
maybe
they
will
make
it
impossible
to
work
for
an
under
the
reason.
In
my
in
my
experience
they
make
this
stuff
working
easier
and
easier,
but
it's
a
very
it's
a
very
sophisticated
way.
I
will
say
that
most
of
I
don't
know
is
kind
of
popular.
A
C
I
see
yeah,
I
I
I
I
like
the
the
idea
of
the
approach
having
the
the
the
setup
and
the
teardown
going
kind
of
per
test.
Instead
of
having
to
spin
up
everything
at
once,
it
seems
really
great
that's
what
I
like,
and
I
liked
also
the
way
that
you
can
migrate.
C
My
concern
is
that
know
that
you
put
that
in
bold.
You
know
because
that
perhaps
is
a
big.
I
I
don't
have
any
idea,
I'm
just
saying.
Perhaps
it's
a
big
rabbit
hole.
That
kind
of
is
very
hard.
You
know,
because
if
this
was
already
working
on
ci
kind
of
in
in
your
plc,
I
would
be
very
kind
of
hey.
Let's
try
to
do
a
a
team
effort
here
to
not
to
to
do
all
of
them,
but
kind
of
set
up
the
standard.
C
E
In
this
context,
I
have,
I
have
like
a
command
a
question
and
a
suggestion.
So
first
of
all,
this
is
really
cool.
Thank
you.
This
is
like.
I
really
appreciate
that
you
look
into
this
and
like
tried
it
as
well.
This
is
actually
like
it
you're
picking
up
at
the
right
kind
of
problem,
so
this
is
really
cool
really
appreciated.
Thank
you,
question
is,
did
you
I
think
you
cheated
with
tony
right?
E
Was
it
just
mac
related
or
was
it
general?
Could
you
read
it?
I
think
you
chatted
with
tony
about
about
this.
Was
it
just
mac
related
or
was
it
general
about
all
this,
or
did
you
not
have
a
chance
to
chat
with
him
yet.
E
A
There's
another
player
which
is
already
merged,
which
simply
has
a
configuration
if
you
want
to
run
stuff
in
the
docker
containers
like
like.
B
E
It
will
be
easier,
good,
good,
okay,
cool,
so
yeah.
Then
then
I
guess
it's
in
the
context
of
suggestion.
I
would
really
because
I
personally
very
far
removed
from
the
deep
technical
details
of
this
and
so
jack
knows
much
better
about
it,
but
there
is
like
a
bunch
of
folks
on
the
team
who
are
looking
into
this
right
now,
actually
in
detail.
F
F
E
Yeah,
so
I
I
don't
have
context
on
the
technical
details,
but
I
would
highly
recommend
to
have
a
conversation
with
the
folks
who
do
have
those
technical
details
and
to
see
whether
it's
like.
If
it's
unrelated,
no
worries,
then
it
doesn't
contradict
each
other
and
both
of
it
can
be
done
but
like
if
it's
related,
I
think
just
like
sharing
some
context
would
be
super
useful
for
everybody.
A
D
D
F
So
if
you
want
to
send
an
email
to
me,
I
can
also
connect
with
colin
who's
doing
some
other
work
and
we
can
try
and
set
up
a
meeting.
D
So
robert,
what
you
showed
here
is
similar
to
what
we
were
doing
at
new
relic,
not
so
much
with
the
docker
compose,
but
rather
we're
trying
to
get
the
test
app
itself
to
spin
up
the
dependencies
that
it
that
it
needs,
and
then
the
idea
there
is
that
we
can
debug.
What's
going
on
without
running
the
test
framework
itself,
we
can
manually
exercise
that
app
that
application
and
try
to
see
what's
going
on
just
to
remove
some
of
the
other
moving
parts.
B
D
Test
yeah,
so
what
we've
run
into
in
the
past,
so
yeah
also
running
into
flaky
tests
for
different
reasons,
I
mean
the
tests
all
spin
up,
different
processes
with
different
apps
and
and
other
dependencies
and
so
on,
and
so
what
we
would
run
into
is
some
of
our
tests.
They
only
execute
something
once
and
so
it
becomes
harder
to
debug.
D
If
you
didn't
set
your
breakpoint
in
the
right
place,
and
things
like
that,
so
we
decided
to
be
able
to
be
able
to
run
those
test
applications
outside
of
the
context
of
the
testing
framework
itself,
because
there
was
a
coupling
between
the
two
things.
D
So
our
test
apps
were
aware
of
things
in
the
test
framework
to
be
able
to
coordinate
different
things
happening,
and
so
it
made
it
really
hard
to
run
these
test,
apps
separately
from
x
unit,
for
example,
and
so
we
decided
to
try
to
make
a
switch
to
where
we
would
try
to
push
some
of
that
logic
into
the
test
app
itself,
so
that
you
can
just
start
up
the
test
app
and
be
able
to
debug
things.
So
in
that
case,.
A
D
Yeah,
but
so
so
that
that
could
be
a
potential
option.
I
think
right
now
for
simplicity,
we're
just
saying
start
up
this
container
before
you
run
this
app,
but
that
seems
like
a
natural
progression.
A
I
was
also
thinking
about
another
thing,
but
it
was
just
a
I
was
just
thinking.
A
D
So
you're
talking
about
instrumenting
the
test
process
itself,
exactly
right.
D
Runner,
but
I
think
a
lot
of
the
test
applications
itself,
you
you
actually
want
want
to
monitor
your
iis
process.
B
D
Your
your
process,
that's
actually
running
reddit,
so
in
some
cases
yeah
you
could
put
some
of
that
into
your
the
test
runner
itself,
but
in
other
cases
you
really
will
have
a
separate.
A
F
F
I
guess
the
other
thing,
so
I
don't.
I
don't
know
how
open
the
open
telemetry
is,
but
you
could
set
up
a
build
machine
and
send
jobs.
There
have
that
as
an
agent
that
all
that's
cute
stuff.
We
usually
just
rely
on
the
manage
pools,
though.
A
F
Thank
you
for
showing
this.
This
is.
I
think
this
is
really
a
good
idea
and
and
all
the
improvements,
it's
very
helpful,
to
have
your
new
perspective,
to
highlight
the
different
things
that
we,
just
you
know,
haven't
considered
or
just
kind
of
shrugged
away.
F
Right
sounds
good
good
meeting,
see
you
soon,
yep
see.