►
From YouTube: 2022-01-05 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
B
A
E
A
All
right,
so
I
think
we
have
everybody
that
expect
to
be
here.
Yes,
I,
for
those
of
you
don't
know,
my
name
is
bob
strichants,
georgia
intuit
and
I'm
one
of
the
maintainers
and
teemo
aka
title
is
one
of
the
other
maintainers,
so
we're
here
to
answer
any
questions
or
talk
about
any
discussion
points
that
you
may
have
with
open,
telemetry
php
project,
I've
linked
in
the
chat,
our
weekly
sig
meeting
agenda.
Note
document.
A
So
if
you
have,
if
you
have
any
questions
or
comments.
A
Please
please
either
ask
them
or
add
them
there
we'll
make
sure
to
talk
through
them.
This
is
so
without
further
ado.
I
guess
we
can
get
started
on
our
open
disc.
My
the
open
two
open
discussion
points
that
I
made
and
then
we
can
continue
talking
based
on
what
comes
up.
Let's
start
with
the
k3s
and
equinix
metal
timo
and
timo
and
company
requested
some
equinix
metal
from
the
cncf.
They
were
nice
enough
to
grant
us
that
hardware.
A
So
now
I
think
we're
going
to
talk
through
adding
adding
kubernetes
on
that
hardware,
so
that
can
be
useful
timo.
Do
you
want
to
talk
through
what
your
thought
process
is
there?
Maybe
we
can
continue
to
talk
through
it.
B
B
So
we
asked
for
three
servers.
Basically
we
could
get
more
so,
but
we
will
use
the
three
ones
we
have
asked
for
for
now,
I've
been
in
the
process
setting
up
it
a
little
bit,
so
we
got
the
service
yesterday,
so
I
haven't
gone
that
far,
but
I
consulted
with
brad
and
bob.
So
what
we,
what
kind
of
what
flavor,
or
what
distribution
of
kubernetes
to
install
there?
B
So
we
want
to
keep
it
easy,
so
we
decided
to
go
for
k3s,
that's
some
fairly
simple
and
lightweight
distribution,
which
you
can
also
run
locally
on,
which
is
easy
to
set
up
and
they
even
got
a
documentation
how
to
set
up
set
it
up
on
one
on
their
servers.
So
for
now
I
haven't
finished
all
the
setup,
because
there
are
some
specifics
which
I
had
to
read
through
how
to
do
networking
and
stuff
with
this
equinix
metal.
It's
not
super
complicated,
but
some
sort
of
stuff.
B
I
had
to
read
so
so
yesterday
I
did
like
a
try
run
and
I
set
up
a
k3s
there.
It
basically
worked,
but
I
still
have
to
set
up
a
lot
balancer.
They
have
one
solution.
They
are
one
way
to
do
it
in
the
documentation,
but
I
don't
know
what
did
you
ever
try
and
what's
called
metal
lp.
A
B
Well,
their
recommendation
was
something
different.
It
was
a
little
bit
easier
than
when
you
say
it's
complicated.
I
will
try
theirs
first,
so
basically,
just
we
have
to
end
up
in
some
some
access
points.
You
know
if
it's
so
if
it's
a
real
alkalines
or
not,
but
okay,
I
will
try
this
first
and
then
maybe
quick
and
set
up
something
else
later.
If
it
doesn't
work.
B
Yes,
basically,
I
follow
the
guide
which
there
we
have
to
tweak
it
a
little
bit,
because
some
things
are
a
little
bit
different
than
s
as
often
with
guides.
You
know,
they're
not
really
updated,
so
there's,
but
I
will
document
everything
which
is
different
there
not
to
go
into
any
detail.
That's
not
super
important,
but
yeah.
Some.
There
have
been
some
watches,
but
you
have
to
basically
adjust
but
yeah.
Then
we
should
be
good
up
and
running
and
for
then
the
deployment.
I
think
we
should
go
for
argo
cd.
A
I
haven't,
but
I'm
I
I
am.
I
have
a
kindergarten
familiarity
with
it,
a
kidnapping
I
I
have
a.
I
have
a
very
rudimentary
understanding.
B
Okay,
well,
basically,
it's
quite
easy
to
set
up.
Have
you
maybe
used
flux
before
as
a
github's
deployment?
B
Okay,
augustine
is
easier
in
my
opinion
than
flux,
because
it
has
a
really
good
graphical
user
interface,
where
you
can
see
everything
and
you
can
see
the
drift
from
so
when
it's
up
and
running
it's
fairly
easy
to
use.
You
just
have
to
put
the
kubernetes
manifest
or
what
you
have
so
it
works
with
customize
and
works
with
helm
and
plane
manifest.
So
I
usually
use
helm
because
that's
really
easy
to
set
up
you
just
run
helm
in
it.
Then
you
have
a
shot
and
then
you
can
throw
stuff
in
there.
B
Then
you
put
this
those
charts
in
a
github
repository
and
configure
it
in
architecture
which
is
very
easy,
and
but
you
can
also
they
have
custom
resource
definitions.
That's
in
case
anybody
doesn't
know,
that's
basically
custom
resource
definitions
for
kubernetes,
which
some
protests
create.
So
you
cannot
it's
basically
like
a
configuration
which
you
can
use
so
that
kubernetes
then
understands
what
to
spin
up
or
something
like
that,
so
they
have
for
their
own
stuff.
They
have
an
application
and
crd
there,
which
you
can
use.
B
So
you
can
basically
put
everything
in
git
and
it
will
you
don't
have
to
touch
kubernetes
anymore
for
any
well,
maybe
for
some
maintenance
or
stuff
like
this.
So
but
usually
you
don't
have
to
touch
it
for
anything
anymore.
You
just
have
to
change
the
the
manifest
or
whatever
have
you
and
get
and
then
agassi
will
will
see
the
changes.
So
you
can
configure
webhook
then
it
will
see
them
immediately.
B
Otherwise
it
will
just
pool
like
every
minute
or
something,
and
then
you
can
configure
if
it
should
apply
for
certain
projects
immediately
or
you
just
basically
push
a
button
to
apply
the
changes
on
kubernetes.
So
that's
really
easy
to
use,
even
when,
once
it's
it's
not
even
complicated
to
set
up
that
some
gotchas,
like
always
with
kubernetes.
So
it's
better
to
have
some
basic
knowledge
to
set
the
stuff
up.
B
But
it's
fairly
easy
to
use
when
when
it's
set
up,
and
so
even
people
can
just
create
a
like
a
deployment
or
something
don't
not
having
to
know
too
much
around
how
this
really
works
or
how
it
gets
up
into
bran.
A
B
My
pleasure
so,
but
I'm
still
working
on
this,
but
I've
spent
time
this
week.
So
I
think
I
will
get
all
the
stuff
this
week,
get
it
up
and
running,
and
then
we
can
see
what
we
actually
yeah
do
there.
So
of
course
one
one
thing
is
the
performance
test,
but
we
have.
I
think
we
should
also
run
some
end-to-end
tests
so
to
actually
more
test.
B
If
all
the
stuff
we
have
we're
working
on
especially
the
exporters
and
are
actually
working
as
we
expected
and
then
maybe
some
some
demo
websites
and
and
so
on.
There's
a
lot
of
stuff-
and
this
will
be
also
be
helpful
because
I'm
working
on
splitting
all
the
components
of
the
sdk
and
the
api
and
the
control
repository
into
separate
composer
packages
and
with
the
setup
we
have
on
open,
telemet
telemetry,
it's
a
little
bit
cumbersome
because
we
don't
have
direct
access
to
basically
most
of
the
things
you
usually
have
as
administered
administrators.
A
Cool
thanks
for
that's.
B
A
Update
so
the
other
thing
that
I
put
on
our
open
discussion
list
is
what
we've
I
mentioned,
that
we
should
talk
about
our
next
priorities,
so
I'm
interested
to
see:
do
you
know
what
you
and
others
feel
like
our
next
priorities
for
this
project
should
be,
I
think,
we're
doing
a
lot
of
really
we're
making
a
lot
of
really
great
progress,
which
has
been
really
exciting,
but
I
wanted
to
talk
to
what
others
feel
should
be
our
next
like
goals
and
milestones
for
this
project.
B
Yeah,
well,
I
think
in
general
we
should
go
for
a
better
release
or
some
some
something
more
stable.
There's
still
one
thing,
I'm
still
working
besides
on
a
little
bit
overhaul
of
the
exporter,
so
it's
a
little
bit
more
unified
how
they
work
so
yeah
so
like
they
have
a
different
interface
and
all
the
the
parts
which
are
different
are
extracted
out
and
abstracted
out.
So
we
I'm
not
set
on
to
how
to
deal
with
this.
B
So,
but
basically,
we
have
like
a
generic
grpc
client,
which
then
gets
just
configured
to
use,
auto
or
jaeger
or
whatever,
and
the
same
for
http
and
stuff
like
this.
So
so
this
I
there
has
been
a
lot
of
work
and
commits
and
all
these
exporters
and
that's
usually
a
sign
that
there's
some
problems
with
stuff.
So
it's
not
really
abstracted
enough
to
be
easily
maintainable.
B
B
B
I
was
muted,
yes,
for
example,
this
one-
and
there
are
some
things
I
think
in
there
which,
which
we
maybe
can
discuss
later,
because
some
of
the
issues,
especially
some
of
the
old
ones,
which
have
been
collected
through
this
user
research.
B
Yeah,
I
think
they
yeah,
I
I
either
I
don't
understand
them
or
they
are
just
not
feasible
to
be
implemented,
but
we
can
go
over
those
later.
So
if
I'm
open.
A
All
right,
so
maybe
maybe
we'll
do
that
in
our
next
sig
meeting,
we'll
go
over
the
user
research
tickets.
B
Yeah,
can
you
maybe
because
actually
that's
interesting,
explain
how
they
have
been
created,
so
I've
just
seen
that
they
have
been
like
a
form
which
could
people
use
to
test.
D
A
C
B
A
Like
a
blank
version
of
this,
and
somebody
was
able
to
fill
it
out,.
A
But
this
was
done
quite
a
bit
ago,
so
it
could
it
could
things
could
definitely
have
changed.
B
Yeah,
so
there
are
definitely
some
things
which
were
outdated
and
I'm
not
sure
because
I
don't
know
how
well
the
spec
I
was
looking
back
then
so
so
some
some
things
are
just
do
not
work
anymore.
You
know
some
tickets
or
we
should
do
differently,
so
we
should
maybe
go
over
them
at
some
time.
A
Agreed,
I
think
that
we'll
hear
mark
then
we'll
earmark
the
next
next
week's
meeting
to
do
that.
B
B
Do
we
want
to
proceed
with
a
better
with
a
better
release,
or
I
mean
I
was.
I
think
I
was
the
only
one
which
were
saying
something
regarding
next
steps,
so
maybe
somebody
else
has
something
to
say.
A
I
agree
with
you.
I
think
that
we
should
probably
work
towards
crushing
required
for
beta
tagged
issues,
there's
eight
of
them
and
being-
and
I
think
that
that's
probably
a
really
great
way
for
us
to
start
working
towards
a
beta
release
which
will
eventually
get
us
to
a
ga
release.
B
A
So
anybody,
I
think
that
anyway,
you
and
I
are
probably
the
only
ones
on
the
call
and
they
study
cube,
but
that
actually
have
enough
experience
on
the
project
to
know
how
that
works.
So
I
think
that
we
can.
We
can
say
that
that's
probably
the
right
thing
for
us
to
work,
work
towards.
B
A
Yeah,
the
the
semantics
aren't
really
that
important
to
me.
I
think
we
just
have
to
work
towards
towards
a
stable
release.
However,
we
can,
if
we
want
to
call
it
data
or
rc
or
yeah,.
B
Yes,
so
a
lot
of
the
issues
I
actually
worked
on,
we
have
maybe
yeah
it's
a
little
bit
of
a
pity
that
brett
is
not
here,
because
we
maybe
have
to
think
over
the
configuration
one
because
yeah
well,
that's
something
we
actually
need,
but
on
the
other
thing
other
and
on
the
other
hand,
the
reason
why
we
try
to
want
to
post
or
why
we
want
to
postpone
it.
B
A
little
bit
was
that
there's
actually
an
official
configuration
specification
upcoming
and
we
basically
don't
want
to
do
the
work
twice,
but
I
think
we
don't
know
really
when
when
this
specification
will
be
finished,
so
maybe
we
have
to
come
up
with.
You
know
some
some
way
to
do
configuration
before
yeah
and
maybe
there's
I
actually
looked
into
how
other.
At
the
moment
we
don't
have
really
a
central
entry
point
for
the
sdk
so
and
most
of
the
other
language
sdks.
B
You
have
something
on
the
api
level
which
you
can
create
some
object,
or
some
factory
or
whatever
it's
kind
of
different,
but
and
where
which
you
can
actually
run.
Instrumentations
against.
You
know
understand
what
I
mean.
A
B
So
because
I
think
this
is
the
real,
the
real
point:
why
there's
an
api
in
the
thread
or
something
called
an
api?
That's
a
very
ambiguous
term,
but
and
actually
it
was
kind
of
confusing
for
me
in
the
first
at
first
when
I
read
it,
but
I
think
the
idea
is
that
the
api
holds
everything
bare
bones
so
and
you
can.
B
B
Are
positioned,
I
was
looking
for
a
german
word.
Sorry,
I'm
positioned
in
this
in
this
matter,
so
yeah.
I
think
this
would
be
important
to
have
this
as
well,
especially
if
we
have
the
split
packages
but
in
the
future,
but
I
think
then,
maybe
it's
easier
to
to
actually
investigate
how
this
all
works,
where
if
we
have
a
separate
api
package
and
then
having
the
whole
repo
and
I'm
just
figuring
stuff
out
yeah,
but
maybe
I
can
say
something
to
this
regarding
this
as
well,
because
I
forgot
a
little
bit.
B
I've
been
working
on
this
as
well,
and
I'm
very
good
in
how
they
get
split
is
done
and
how
they
end
up
this
only
a
little
bit
stuff.
I
have
to
work
on
to
actually
automate
all
the
stuff
and
I'm
a
little
bit
of
play
kit
when
it
comes
to
this,
because
we
have
to
create
different.
B
If
you
guys
don't
know
we
have,
or
we
have
created
another
organization
where
we
can
basically
place
the
packages
so
the
problem
which
we
have
or
yeah.
B
Basically,
it's
it's
a
problem
that
composer
packages
cannot
be
cannot
live
in
a
subdirectory
of
the
repository,
so
what
people
do
usually
is
they
split
the
code
which
is
just
the
code
for
one
of
the
packages
and
synchronizes
them
in
different,
separate
repositories
somewhere
else,
usually
in
the
same
organization,
but
in
our
case
we
have
to
use
a
different
one,
because
we
cannot
create
20
or
50
or,
however,
we
may
have
one
in
the
future
repositories
and
open
telemetry.
B
So
we
basically
saw
this.
What
it
does
is
takes
the
code
base
and
then
we
we
can
configure
which
code
goes
into
which
packages
and
then
this
process
handles
all
this
stuff
around
it,
and
I
was
looking
into
a
lot
of
different
implementations
or
ways.
You
can
do
this,
so
I
ended
up
finding
and
basically
the
servers
will
help
us
with
this,
because
some
limitations
we
have
on
github.
B
You
can
work
around
with
this,
but
basically
it's
all
done.
It's
very
good
should
be
by
the
end
of
the
week.
It
should
be
all
there
and
then
we
will
have
our
separate
packages
and
we
can
see
what
we
can
do
with
them
or
if
they
actually
work
as
separate
pictures
as
the
antenna.
B
Otherwise
I
will
be
happy
to
present
how
all
of
the
stuff
works
in
a
in
a
meeting
where
we
actually
have
all
the
stuff
running,
because
then
it's
a
little
bit
easier
and
not
too
vague.
A
A
B
I
had
some
notes
from
last
week
where
I
just
talked
with
myself:
it
was
a
little
bit
boring.
Yes,
I
have
semantic
conventions.
I
don't
know
what
I
meant
with
it.
I
know
what
semantic
conventions
is
what
created
instances.
B
I
cannot
actually
remember
what
I
meant
when
I
wrote
down
php
storm
I
have
so
maybe
can
tell
about
this.
We
have
I've
also
asked
jetbrains
for
some
open
source
licenses
for
a
php
zone
and
everything
else.
B
There
are
some
requirements
regarding
this,
but
if
you
generally
need
some
license,
we
can
talk
about
this,
but
it
can
only
be
handed
out
to
regular
contributors
to
the
project,
so
you
must
at
least
have
the
record
of
over
three
months
of
contributing
to
the
project,
but
then
we
can
talk
or
even
if
we
we
have
one
we
can
give-
could
give
out
at
the
moment
one
more
license,
but
we
could
possibly
ask
for
others
in
the
future
as
well.
So
just
for
an
information.
B
If
you
have
interest
in
this
but
yeah
you
should
be
a
regular
contributor,
because
that's
basically
the
requirement
jetbrains
puts
on
this,
so
we
cannot
just
send
them
out.
For
free
to
everyone,
but
in
case
you're
interested
then
just.
B
B
Yes,
so
doppler
as
a
secret
handling
software
as
a
service
which
yeah
basically
can
help
us
with
centralizing
secret
management,
a
little
bit
easier
because
we
don't
have
access
on
the
both
repositories.
We
have
under
the
open,
calendary
umbrella
on
this
organization,
and
so
it's
a
little
bit
cumbersome
to
deal
with
secrets.
There.
So
pastor
asks
for
access
and
then,
if
you
forget
something
so
we
will
just
use,
especially
for
the
service
that
we
will
just
use
on
doppler,
it's
kind
of
a
nice
solution.
B
I
can
put
a
link
to
this
stuff.
Maybe
in
the
future
we
can
add
more
users,
but
that's
the
free
option.
Basically
just
has
four
and
there's
it's
a
little
bit
tricky
because
everyone
is
an
administrator,
but
but
there
is
a
what's
actually
called
an
english.
B
Let
me
out
a
link
where
basically
there's
a
link,
and
if
you
sign
up
over
this
link,
you
will
get
hundreds
of
dollars
in
which
you
can
use
an.
B
Affiliate
yeah,
that
was
the
word
I
was
looking
for
and
we
couldn't
get
a
hundred
dollar
as
well.
For
so
maybe
we
can
open
this
up
in
the
future.
We
can
invite
more
people
and
go
for
a
paid
option
and
pay
like
this,
but
yeah
at
the
moment.
Just
so
you
know
we
will
use
this
tool
for
handling
secrets
and
which
is
an
easy
way
for
us
to
handle
them.
And
if
you're
interested
I
can
put
later,
I
can
put
a
link
basically
put
into
the
link
in
the
chat.
B
So
then
you
can
sign
up
and
get
hundred
dollars
and
we
can
well
more
physically,
which
you
can
use
on
this
platform.
And
if
you
have
it's
actually
really
nice.
I've
used
another
software
as
a
service
solution
before
which
was
nice
as
well,
but
they
got
bought
by
somebody
else
and
don't
offer
any
free
service
anymore.
So
players
it
is,
and
their
alternative
would
be
setting
up
something
if
you
know
like
hashicorp
vault,
which
you
probably
would
go
for
some
on
the
enterprise
level.
B
But
it's
a
little
bit
tricky
to
administer,
to
work
with
and
set
up
and
maintain,
and
so
that's
the
easiest
solution.
B
A
A
B
E
D
I
I
just
had
one
kind
of
thing
so
right
now,
I'm
working
on
revamping
softies
otr
for
integrating
those,
the
jager
legends
and,
as
I'm
kind
of
like
going
through
watching
and
they're
kind
of.
D
Yeah
yeah,
I
think
my
computer
has
just
been
like
the
event.
Is
the
fans.
A
B
D
Yep,
okay
and
yeah.
So
basically
I'm
trying
to
revamp
that
pr
and
with
the
goal
of
like
one
stats
and
and
being
able
to
kind
of
rework
some
of
that
for
the
rest
of
the
year
support
and
but
the
one
thing
I'm
kind
of
in
the
back
of
my
mind
is
that
I
don't
really
yet
know
how
to
like
functionally
test
any
of
safti's
original
changes,
and
I
know
he
mentioned
a
while
back
that
he'd
been
actually
like
using
his
fork
in
production,
and
so
I
certainly
like
hasta
and
but
yeah.
E
A
B
You
you
could
just
at
first
test
this
manually
or
you
know,
just
take
a
look
if
everything
ends
up
there,
but
I
can
we
can
send
since
we're
in
the
process.
We
can
set
up
those
enter
and
test.
So
there
are
a
lot
of
different
ways.
I
don't
want
to
make
it
too
complicated,
but.
B
A
good
testing
framework
is
codeception.
If
you
guys
know
about
this
for
php,
because
it's
so
like
a.
Let
me
look
for
the
link.
Then
I
can
post
it
it's
kind
of
a
it
sits
on
top
of
php
unit,
which
is
kind
of
the
defective
standard.
Well,
there
are
other
frame
testing
frames
with
php,
but
usually
people
use.
B
Php
unit,
and
now
I
found
the
link-
sorry
that
was
just
rambling.
I
learned
this
years
ago
when
I
was
working
in
a
call
center,
so
people
don't
just
to
talk
about
anything
if
you
just
trying
to
find
something.
So
this
is
conception
and,
as
you
can
do,
behavior
testing
with
it,
you
can
do
acceptance,
testing
and
also
functionality
and
testing,
so
there's,
basically
and
or
api
testing
and
so
for
giga
and
the
exporters.
B
I
would
basically
then
just
yeah
go
for
api
testing,
because
jaeger
has
an
api
which
he
couldn't
just
crawl
or
ask.
So
you
don't
have
to
you
know,
use
the
user
interface.
You
can
just
call
the
api
and
then
check
if
everything
ends
up
there
as
expected.
So
that's
how
I
would
do
it
if
anybody
has
any
another
idea.
So.
D
B
But
I
would
you
probably
don't
have
to
test
everything,
because
with
the
s
with
the
nature,
with
the
entire
test,
they
can
be
very
brittle
and
changing
all
the
stuff
all
the
time
around.
So
basically
just
a
smoke
test.
You
know
the
span
ends
up
there,
then
yeah.
So
I
don't
even
know
well,
you
cannot,
of
course,
or
let's
put
it
in
that
way,
you
can
actually
with
conception.
You
can
tag
certain
tests
or
some
smoke
tests.
B
B
D
B
Okay-
let's
maybe
maybe
we
can
aim
for
this
for
next
week,
then,
if
we
have
the
server
and
stuff
all
set
up,
it's
basically
a
little
bit
easier
to
talk
about
this,
maybe
having
different
options
or
but
yeah.
That's
basically
what
I
you
can
basically
use
any
testing
framework
for
stuff
like
this,
but
I
think
it's
easier
to
to
use
something
which
has
already
some
entertainment
testing
backed
into
it.
You
can
also
do
with
a
php
unit
or
even
the
but
yeah.
I
think.
B
D
B
And
did
you
see
that
I
moved
or
we
basically,
I
moved
the
generator
stuff
like
we
talked
about
in
not
the
last
meeting
where,
where
just
both
of
us
were
there.
B
So
yeah
there's
no
a
subdirectory
for
for
auto
otlp.
I
always
forgot
how
the
protocol
is
called.
B
I
moved
all
this
stuff,
basically
one
directory
on
the
lower
directory.
So
then,
when
we
have
the
same
or
grpc
implementation
for
you
as
it
can,
you
can
put
the
stuff.
Oh
actually
I
I
I
remembered
something
I
wanted
to
talk
about.
Thank
you
for
you
yeah.
But
okay
is
your
question
and
answered
so
then
I
can
talk
about
the
other
stuff.
D
B
So
you
maybe
have
seen-
or
maybe,
if
not,
that
I
changed
something
and
how
the
grpc
or
the
hd
photograph
code
is
created
because
before
it
was
well
basically
any
change
from
the
main
line
would
have
ended
up
in
protocol
in
code.
B
So
I
changed
it
to
use
the
last
tag,
and
so
my
idea
was
if
we
basically
move
the
creation
or
all
the
generated
codes
to
a
different
repository
which,
on
this
open,
telemetry
php,
because
then
we
could-
or
let's
say
we
could
do
the
same
when
we
leave
it
here,
but
it
would
a
little
bit
easier.
We
could
automate
the
creation
of
these
packages
and
use
a
different
version
number
for
them.
You
know
basically
just
the
version
number.
I
think
it
was
now
0.7
which
I'm
not
like
this.
B
B
Well,
do
you
understand
my
my
reason.
A
B
Yeah
the
the
reason
I
would
leave
it
basically
there,
but
I
think
it's
a
good
idea
to
give
the
packages
the
same
version
number
the
then
the
upstream
is
so,
and
that
would
not
really
correlate
with
the
version
number
we
have
in
our
packages
and
that
would
make
the
package
splitting
a
little
bit
more
complicated.
B
I
can
still
do
it,
but,
but
since
we
wouldn't
it's
basically
self-running,
then
we
don't
have
to
have
it
in
the
in
our
repository
and
on
the
other
hand,
and
people
can
can
just
require
that
for
something
else
if
they
want
to
do
their
own
implementation.
B
Okay,
so
that's
not
super
pressing,
but
well.
I
will
put
this
on
my
bucket
list
that
we
move
the
stuff
there
and
then
we
do
not
have
to
care
about
this
too
much
anymore.
Okay,
cool.
A
B
Yeah
and
the
same
we
could
do
then
with
thrift
stuff.
You
know.
D
B
That's
it's
automatically.
We
just,
I
don't
know,
run
once
a
day.
We
run
the
check
if
there's
something
new
and
then
we
create
the
automatically
create
the
packages,
and
you
know
maybe
create
a
select
message.
So
everybody
knows
if
in
case
there's
something
to
test
or
to,
and
then
we
just
use
in
the
code
just
use
a
fixed
version
and
then
check
if
everything
still
works,
but
you
can
just
now,
of
course,
just
create
it
how
you
did
or
how
it's
done
before
you
use
the
same
way.
B
This
I
can
remember
his
name,
but
the
other
guy.
This
generator
right.