►
From YouTube: 2021-10-06 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).
B
A
E
All
right
so,
let's
get
started.
I
think
the
first
thing.
Let
me
just
briefly
share
here
the
things
that
we
have
there,
but
if
people
want
to
bring
something
else
up,
feel
free.
E
So
robert
start
the
issue
for
us
to
finally
move
that
we've
been
talking
a
few
days
a
few
weeks
about
making
the
plc
the
main
and
really
start
the
work
there
I'm
out
for
it.
I
just
checked
that
there
was
zach
and
chris
also
giving
the
thumbs
up.
So
I
think
you
are
set
to
go.
I
the
only
thing
that
I
have
kind
of
in
my
mind.
E
I
want
to
finish
the
the
pool
from
upstream
on
may
before
we
switch
main
just
because
I
have
the
pr
opening
it
passed
at
everything.
Zach
pointed
a
few
days
back
some
issues.
We
did
some
investigation
and
also
fixed
some
rename
issues
that
were
breaking
linux.
Robert
helped
with
that
investigation.
E
So
I
think
we
are
good
there.
I
think
the
last
thing
is
for
me
to
remove
the
headers
and
then
zach
do
the
merge
right
like
last
time.
So
I
will
try
to
remove
those
headers
later
today,
then
I'll
bring
zach
via
slack.
So
we
can
get
that
closed
once
that's
closed,
let's
not
open
anything,
your
main
and
let's
do
the
switch
click,
since
we
don't
have
many
pr's
at
the
same
time,
it's
easier
if
you
do
the
switch,
while
no
pr
is
open
against
me.
E
So
I
think
you
are
good
there.
Robert
do
you
want
to
add
anything
else
or
any
preparation
step.
No.
Thank
you.
Let's
move
on,
I
like
that.
E
Okay,
I
think
the
the
most
interesting
thing,
then,
is
the
startup
hook.
I
think
we
have
the
change.
The
changes
is,
as
far
as
I
understand
is
ready.
There
was
some
question
about
from
rasmus
regarding
the
location
of
the
the
dll
itself.
I
think
it's
a
it's
a
good
thing
for
us
to
discuss
erasmus.
Can
you
bring
up
the
the
issue
that
you're
thinking
about.
F
Yeah
so
we
suggested
to
use
the
hotel
manage
profiler,
but
instead
of
oh,
no
sorry,
auto.net
tracer
home
instead
of
automatic
profiler,
and
so
this
environment
variable
can
be
cross
used
between
dotnet
framework
and
dotnet
core,
which
means,
if
you
point
this
one
to
the
home,
then
you
need
to
do
additional
app
local
changes
to
make
the
dot
net
core
application
pick
up.
The
startup
hook,
dll.
D
Yeah,
you
are
correct,
so
I
I
currently
remove
that
once
so
we
are
no
more
depending
upon
that
the
tracer
home
environment
variable
at
all.
Now
so
it's
completely
removed.
So
that's
why
I
specified
in
slack
so
the
startup
hook,
library
lives
in
the
like
tracer
home
now,
so
it
knows
how
to
pick
the
library's
host,
which
is
in
the
same
folder
now.
E
So
but
but
do
you
have
to
specify
on
the
location
kind
of
the
the
core
version
kind
of
because
you
have
to
put
net
core
app
3.1
in
the
path?
Assuming
that
the
default
structure
that
we
put
on
the
home,
you
have
to
put
net
core
3.1.
D
E
This
is
kind
of
it's
related,
but
kind
of
something
that
happened
in
the
best
behalf.
Zach
has
a
memory
of
that.
I
I
don't
recall
so
in
the
past
for
everything.net
core,
it
was
using
net
standard
2.0.
E
Then
now
we
have
net
standard
to
zero.
In
that
part
app21,
do
you
know
why
zack.
C
Yeah
mostly
so,
when
we
first
started
out,
we
had
net
standard
2o
because
for
nuget
publishing
guidelines
it
was
recommended,
if
you
had
a
net
461
target,
or
I
guess
yeah
yeah,
like
a
target
framework
for
net
461,
that
you
should
also
have
a
net
standard
2o.
So
we
have
both
of
those
and
then
to
take
advantage
of
some
built-in
libraries
on
netcrab31.
E
A
Thing
I'm
curious
about
why
the
choice
of
net
core
app
versus
net
standard
2.1,
because
I
think
that
they
were
one
in
the
same,
at
least
as
far
as
the
improved
apis
were
part
of
net
standard
2.1
and
not
separated
out
into
net
core
app
3.1.
C
That's
a
good
question,
I
don't
remember
the
decisions
at
the
moment.
I
think
there
still
might
be
some
libraries
that
are
not
in
a
center
to
one.
I'm
thinking,
maybe
around
I
don't
know
are
else,
is
the
alc
api?
Is
that
supposed
to
net
standard,
or
is
that
just
net
core.
C
Yeah
I
yeah,
I
can't
recall,
I
think
we
felt
more
confident
that,
if
we're
running
on
the
platform
that
instead
of
having
the
redirection
through
net
standard,
we
feel
more
confident
about
using
the
platform
assemblies,
since
there
would
be
direct
like
system
private
corelib
reference.
C
So
that's
the
only
reason
I
could
think
of
that.
We
preferred
net
core
app
31
over
net
standard
2
one
until
I
think
about
the
moment.
D
E
So
so,
basically
both
for
the
profiler
and
they
start
up
hook,
we
can
use
you
can
keep
use
they're,
not
net
core
app
3.1.
It
should
work
both
for
net
five
and
net
six.
E
We
just
need
to
add
separate
ones
if,
like
the
reason
zach
pointed
if
we
want
to
take
advantage
of
something
that's
provided
on
the
runtime,
then
we
need
to
specialize,
but
for
now
trick
network
31
that
folder.
What
actually
means
is
the
minimum
requirement.
That's
what
it
means
right
now:
cool
cool
yeah,
I
think,
makes
sense.
I
think
I
think
perhaps
is
is
a
bit
confusing
when
you
look
because
then
they
as
soon
as
you
see
net
core
app
21.
The
first
thing
that
crossed
their
mind
is
kind
of
hey.
E
I
need
one
for
each
or
version,
but
that's
not
the
case.
That's
just
the
minimum,
so
I
I
think
it's
fine
and
the
behavior
is
great.
It
is
just
that
kind
of,
I
think,
is
a
bit
deceiving
because
of
the
name.
E
My
question
that
I
have
actually
for
raj
about
the
startup
hooks
and
in
follow-up
in
that
is
regarding
the
the
next
step
right.
I
understand
that
the
next
step
is
kind
of
showing
an
application
with
the
old
version
and
how
they
startup
hook
is
gonna
use
a
prox
to
kind
of
unify
those.
My
main
kind
of,
perhaps
you
showed
this,
but
I'm
sorry
so
many
things
going
on
and
I
kind
of
perhaps
forget
how
that
works
with
the
sdk.
In
that
case,.
D
If
you
have
the
procs,
so
still,
we
need
to
explore
that
option
like
if
sdk
comes
into
picture.
There
are
so
many
questions
in
foot
in
front
of
us.
What
happens
if
sdk
is
present?
If,
again,
on
top
of
that,
customer
has
a
like
a
processor
written,
ordered
custom
like
telemetry
processor
and
everything,
so
we
still
don't
have
an
answer
for
that.
D
E
Okay,
if
you
wanna,
feel
feel
free
to
bring
up
discussions
on
his
lack
and
things
I
I
think
people
will
be
willing
to
kind
of
contribute
and
giving
if
they
are
read
right,
something
or
consider
something.
You
know
yes,.
E
Yes,
yes,
yes,
both
both
ways:
work,
really
nice.
I
I
was
just
gonna
mention
that
one
of
the
things
that
guided
us
to
the
sdk-
I
think
I
I
perhaps
mentioned
this
before-
was
that
with
the
procs
that
greg
had
prototype
and
then
we
we
saw
before
it
was
that
we
still
couldn't
reuse
the
sdk
code
directly.
So
there
are
paths
I
can
see
some
paths,
some
advantage.
Some
disadvantages,
but
at
at
first
was
one
of
the
main
motivators
for
us
to
really
try
to
move
to
the
sdk.
D
So
well
I
just
a
simple
scenario,
so
thought
process
to
enable
so,
for
example,
auto,
like
instrumentation
from
instrumentation
library
we
are
bringing
we
are
going
to
bring
in
sdk
from
our
and
open
telemetry.
So
if
customer
is
also
bringing
something.
So
if
we
do
that
diagnostic
source
snapping,
so
one
of
them
is
going
to
win
and
we
need
to
disable
the
like
the
telemetry
collection
from
the
other
end.
D
So
the
first,
the
first
part
is
that
later
we
can
think
about
how
it's
going
to
collaborate
and
work
with
each
other
like
what
we
are
doing
for
diagnostic
source
proxy.
We
can
think
about
like
if
customer
has
an
sdk,
customization
and
everything
so
which
one
do
you
think
we
should
disable
it.
Is
it
the
one
from
our
end
or
the
sdk
which
customer
brings
it.
E
Yeah,
I
I
I
I
think
we
should
ideally
try
to
ways
that
we
can
combine
them,
but
we
know
that
there
are
some
challenges
there.
That
are
not
easy.
I
I
think
the
main
thing
is
kind
of
the
ideal
word
we
combine,
but
I
think
that
in
a
sense
the
open
telemetry
for
for
me
because
of
the
conventions
and
all
the
things
should
be
the
priority.
E
But
it's
hard
to
tell
if
that's
the
case
that
satisfy
all
the
possible
case
that
people
are
using.
You
know,
but
if
you
ask
my
perspective,
my
perspective
will
be.
If
I
have,
though
both
I,
you
rather
have
open,
telemetry
coming
out.
You
know.
E
D
So
these
are
something
the
questions
I
have
every
now
and
then
I'll
be
posting
it
like
I'll.
We
will
do
the
base
work
and
we
will
keep
posting
these
questions
in
the
slack
instead
of
like
just
asking
a
question:
it's
better.
If
we
have
it
documented
so
we'll
have
the
everyone's
opinion
there
and
we
could
arrive
to
a
better
conclusion.
There.
E
Yeah
yeah
and
just
to
be
be
clear.
We
also
kind
of
have
that
questions
when
we
consider
the
those
approaches
you
know,
so
we
didn't
kind
of
really
got
into
implementing
and
getting
to
solutions
to
those
in
principle.
We
are
just
not
using
the
sdk,
but
since
we
are
getting
out
of
that,
we
are
going
to
now
have
answers
for
those
you
know
kind
of.
E
E
Yeah
no,
but
it's
great,
I'm
gonna!
Try!
Sometime
before
friday,
I'm
gonna
try
the
the
startup
hoop
specially
so.
D
D
So
here
is
my
visual
studio,
like
with
the
current
changes
like
whatever
we
have
it,
so
the
3.1
app.
So
what
the
like,
what
I'm
doing,
yeah.
E
I
think
you
have
dual
moon
monitor
and
the
window
came
up.
D
Okay,
so,
which
one
are
you
able
to
see
my
visual
studio
screen
now
like.
B
D
Okay
cool,
so
this
is
the
startup
hook
earlier
I
was
using
the
tracer
provide
like
so
many
environment
variables.
I
removed
all
of
them,
so
it
is
very
simple.
What
it
is
going
to
do
is
like
it's
going
to
look
at
us
like
wherever
it
is
located.
It's
going
to
look
at
the
same
in
the
same
location,
it's
going
to
look
for
the
clr
profile
manager,
library
and
it's
going
to
load
this.
So
to
achieve
this,
what
I've
done
is
like
in
the
like.
D
Well,
I
think
you
guys
would
have
done
the
review
already,
but
still
just
want
to
call
it
out
here
I
did
a
change
to
the
one
of
the
publish
still
need
to
get
used
to
the
names.
D
So
this
is
what
I
did.
I
did
a
change
to
the
publish
manage
profiler
here,
so
it's
going
to
move
the
library
from
the
dot
net
hook
up
project
also
to
the
same
output
location.
So
we
had
this
one.
We
were
writing
all
of
the
library
to
tracer
home
and
net
core
app
3.1,
so
the
startup
hook
is
going
to
live
in
this
location
over
here
and
all
I
need
to
do
yes
in
the
sample
app
whatever
we
have
it.
Let
me
set
as
a
startup
project.
D
All
I'm
doing
here
is
like
set
the
data,
dotnet
startup,
hookup
location,
to
that
specific
one
right,
tracer
home
net
core
3.1
and
that
library
setup
hook
for
to
work.
It
should
have
the
dll
name,
so
we
need
to
go
and
point
to
that
one
so
remaining.
I
think
everyone
will
be
familiar
with
these
ones,
so
not
going
through
that,
so
just
launching,
so
it
should
work
like
a
magic.
I
I
don't
have
anything
else
about
from
it's
a
default
app.
I
don't
have
any
like
customization
to
this,
like.
D
D
This
app
needs
a
reference
to
this
type,
so
this
is
something
we
need
to
work
to
remove
it.
So
if
it's,
if
we
move
to
the
sixth
version,
we
don't
need
this,
but
three
point
one
or
five
needs
this
at
this
point,
so
our
future
work
will
remove
this
one
also
so
without
any
of
these
just
an
app
with
dotnet
startup
hook
or
manage.
I
I
feel,
even
though
our
profiler
managed
also
should
work.
E
But
because
we,
when
we
build,
we
put
the
version
that
we
need
on
that
3.1
folder.
If
you
will
need
this,
the
application
fails.
D
F
B
F
B
B
B
D
D
B
D
It
will
not
work
because
framework
will
default.
For
example,
this
application
does
not
have
anything
related
to
diagnostic
source,
so
I'm
just
going
to
remove
and
show
it
to
you
the
behavior
like
what
happens.
D
The
application
is
going
to
bring
a
diagnostic
source
which
is
471
along
with
it,
so
I
might
be
incorrect
also,
it
won't
run
it
so
whenever
the
application
time
before
the
build
time,
it
creates
a
like
a
trusted
platform
assembly
list.
At
that
point,
it
says:
hey
I'm
going
to
accept
a
diagnostic
source
which
is
at
framework
level,
nothing
more
than
that.
If
it
does
not
have
this.
D
E
But
if,
if
we
have
the
the
handle
like
we
have
in
the
loader,
that
is
the
assembly
resolver
that
is
called
after
it
failed
and
then
we
we
point
to
the
directory.
Wouldn't
that
solve
that
case.
D
No,
it
won't
because
this
is
something
I've
been
working
with
dotnet
or
the
reason
it
won't.
Work
is
even
before
that,
like
during
the
build
time,
even
before
the
startup
hook
or
the
profiler
is
loaded.
The
dotnet
builds
up
something
called
as
trusted
platform
assembly
list,
so
that
point
in
time
it
decides
hey.
This
is
the
highest
version
of
some
library
that
I'm
going
to
load
so
yeah,
even
whenever
there
is
a
failure.
If
you
try
to
load
a
6
version,
it
will
directly
crash
very
simple
thing.
D
Is
you
have
a
console
app
with
built-in
3.1
framework
and
you,
if
you
load
a
6.0
library
from
somewhere
dynamically,
it
will
crash
that
application
no
need
of
instrumentation
of
anything.
So
it
is
as
simple
as
that,
because
this
is
a
framework
assembly
and
they
have
some
different
check
in
the
framework
for
these
libraries.
E
I
see
so
so
what
you're
saying
is
that
they
run
time
constraint
for
certain
assemblies,
the
top
version
that
it
loads.
D
Yes,
so
that
was
the
reason
for
me
to
get
tarik
tariq
was
the
like
the
guy
who
designed
this
system.diagnostic
source,
so
we
have
the
first
time
when
I
got
him.
We
have
this
especially
this
this
issue,
where
we
cannot
upgrade
even
with
the
assembly
resolve.
If
we
try
to
load
the
latest
version,
it
will
crash
the
application,
that's
what
is
happening
in
the
other
case,
we
are
trying
to
load
that
dynamically
and
the
6.0
and
it
is
crashing
the
app.
So
it
is
it's
not
a
problem
with
the
startup
hook.
E
Yeah,
so
what
we
did
on
the
plc
were
basically
you
and
you,
you
you're
aware
of
that,
but
we
are
basically
kind
of
requiring
the
application
to
have
the
top
reference
to
the
correct
ones.
So
then
the
runtime
allows
that
to
work
correctly.
I
see.
D
Correct,
but
this
can
be
removed
with
the
other
work
which
I
was
saying
about
the
diagnostic
proxy
this
this.
Whatever
this
requirement
we
have,
we
could
remove
it
in
some
time,
like
maybe
I'm
just
forcing
it's
a
month
work
from
now
after
a
month.
We
should
not
see
these
this
here
if
everything
goes
as
planned.
E
Yeah
I
okay,
I
keep
thinking
yeah
because
then
we
we
get
back
to
that
initial
question.
It's
kind
of
and
the
area
that
we
have
to
explore,
then
how
we
be
sure
that
we
can
leverage
the
sdk
in
this
case.
You
know.
D
E
D
Okay,
that's
a
different
thing:
whenever
someone
brings
an
sdk,
that
means
he
has
an
indirect
reference
to
the
latest
diagnostic
source,
also
with
him,
so
we
have
a
possibility
that
until
the
dot
like
7
gets
released,
so
we
have
that
possibility
to
auto
upgrade
so
I
hope
in
dot
net
7.
We
have
something
at
the
run
time
so
which,
where
we
could
avoid
this
issue
so
literally
whatever
you
were.
Thinking
is
possible.
If
someone
brings
a
sdk,
so
you're
yeah,
it
is
doable.
D
It
is
clr
profiler,
okay.
I.
F
D
That
already
with
this
app
having
both
of
them,
so
it
just
takes
the
the
profiler
one,
not
the
startup
hook
in
startup
gets
loads
for
the
second
time,
but
the
loader
is
written
in
a
way,
not
loaded.
The
instrumentation
is
written
in
a
way.
If
a
second
initialization
happens,
we
just
ignore.
That's
the
start
code
that
we
have
it.
A
D
D
Yeah,
so
what
I'm
going
to
do
is
my
next
work
like
as
follows,
say
there
are
two
to
do's
after
the
startup
hook.
So
one
is
the
integration
test
to
add
this,
and
the
next
work
is
to
do
right
now,
I'm
loading
the
instrumentation
directly.
That's
not
the
right
thing
to
do,
because
I
looked
at
the
c
plus
native
libraries
how
it
is
doing
it.
It
is
rather
using
your
loader
to
load
it,
so
we
can
tweak
the
startup
to
use
directly
load
the
loader
rather
than
the
instrumentation.
D
E
Yeah
yeah,
I
think
I
I
think
that's
the
best
path
we
share
more
hold.
We
have
more
than
that
in
that
path.
Yes,
yeah,
no,
very,
very,
very,
very
interesting.
I
think
I
I
think
it's
really
a
an
area
that
we
can
do
a
good
case
for
what
we
call
the
develop
scenario,
which
gets
me
just
to
think
about
something
down
the
line.
We
will
really
need
a
nuget
package
to
ship
that
right,
so
we
you
need
to
ship
this
as
a
new
get
package
which
was
not
necessary
with
the
profiler.
E
It
was
desirable
with
the
profile
but
not
necessary,
but
in
this
case
I
think
it
becomes
necessary
because
I
think
for
azure
functions
and
aws
lamp
does
that
have
layers
this
kind
of
thing
you
can
have
then
a
layer
with
this
a
trap
hook
and
then
they
just
add
a
layer
and
set
the
environment
variable
they
get
all
the
things
that
are
enabled
by
the
layer.
E
E
Yeah,
I
I
I
have
used
the
clr
profiling
web
services
app
web
apps,
but
those
have
custom
ways
to
deploy
stuff.
You
know
so
that's
possible
to
work.
Azure
functions.
I
don't
recall.
Oh
azure
functions
the
only
thing
that
I
remember,
but
that
was
with
custom
linux
vms.
E
Then
we
could
deploy
the
profiler.
You
know,
I
don't
know
if
they
stand
the
images,
if
there
is
any
way
of
doing
that
for
azure
functions,
I
don't
think
so.
A
A
Know
how
desirable
it
is
to
rely
on
bytecode
instrumentation
in
something
like
azure
functions
and
lambda.
From
from
what
I've
seen
the
overhead
is
often
too
much.
E
Very
exciting
stuff,
both
on
the
branch
and-
and
this
change,
I
think,
is
a
is
good
time
for
a
lot
of
things
happening
in
the
siege.
I
I
love
to
see.
I
love
the
momentum
that
I'm
seeing,
I
think
the
rest
of
the
the
people
here
that
have
been
participate
for
some
time
too
yeah.
I
I'm
looking
forward
for
the
next
stuff.
You
know
it's,
it's
look
exciting.
E
F
So
basically,
I
just
opened
the
dotnet
5
super
vr.
I
see
that
somebody
already
managed
to
look
this
one,
but
I
just
wanted
to
open
this
as
a
discussion
pr.
So
if
it's
not
needed,
then
you
can
just
throw
away
but
yeah.
You
can
look
into
and
say
what
you
think
about
this
one
if
there
is
any
benefit
at
all
or
and
the
second.
E
Yeah,
I
didn't
look
yet
I
don't
know
the
context.
I
don't
know
if
you
prefer
to
discuss
on
the
api
itself
or
if
you
want
to
kind
of
give
an
overview
here.
Also,
it's
fine.
F
F
F
E
See
what
you
get
if,
if
you
don't
get
the
answers
by
the
next
week,
then
we
we
kind
of
bring
more
details
and
try
to
see
if
we
can
get
some
help.
Perhaps
perhaps
I'm
I
don't
want
to
put
this
on
raj
but
raj
is
the
is
on
the
azure
side.
But
I
know
it's
a
huge
team.
You
know,
so
it's
not
like
he'll
know
the
person
or
something
like
that.
But
let's
let's
wait
until
next
week.
If
we
don't
get,
then
we
we
see
if
haraj
can
help
us
with
that.
D
And
also
I
just
did
look
at
the
pier,
so
the
thing
is
that
it
will
like
made
from
a
maintability
like
aspect
if
we
start
doing
this,
adding
like
versions
like
newer
versions
to
our
like
project-
it's
very
it
gets
at
some
point
in
time.
It
will
get
very
difficult
to
like
maintain
it,
because
we
need
to
retire
some
version
at
point
in
time
and
need
to
keep
supporting
newer
version.
D
So
it's
always
recommended
to
keep
the
lowest
possible
version
supported
version,
only
the
target
version
as
that
one
and
create
packages
for
that,
because
that's
that
can
get
loaded
with
any
other
higher
version
of
dot
net.
So
it's
it's
a
recommendation
from
the
dot
net
p.
Probably
david
is
not
that
he
could
have
spoken
more
about
this,
so
I
will
also
ask
him
to
take
a
look
into
it
and
see
if
he
has
any
inputs
for
us.
There.
E
Yeah
one
thing
that
I
think
our
integration
tests
tend
to
kind
of
cover
all
the
versions
that
we
support
is
because
it
involves
the
cli
and
the
profiler.
So
when
it's
doing
the
actual
code
injection,
I
think
that
is
more
chance
of
things
going.
D
It's
not
yeah
not
about
the
integration
test.
I
agree
like
integration
tests
should
be
done
for
all
the
versions
supported
questions,
so
the
change
which
has
come
is
for
the
actual,
the
main
project,
not
for
the
tests.
So
that's
the
same
thing
I
propose.
The
test
should
have
all
version
support,
but
not
the
product.
The
actual
code,
that
is
only
the
minimal
version,
is
needed.
B
E
All
right
everyone
does
anyone
want
to
bring
anything
else.
E
All
right,
exciting
times
for
our
sick,
alright,
see
you
guys
next
week
see
you
bye.