►
From YouTube: Plan | Weekly Sync 2023-03-13
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
C
B
D
Let's
try
them
all
okay,
so
what
we're
trying
to
do
is
that
we
are
able
to
change
any
work
item
to
any
other
work
item
type
like
it's,
if
it's
possible,
based
on
our
restrictions
and
for
now
I'm
also
I'm
working
only
on
the
backhand
side.
So
far,
so
we
don't
have
any
fronted
for
this,
but
yeah
we
are
close
to
having
like
the
notation
working
and
changing
the
types
from
one
to
each
other
and
I
want
to
represent
like
what
happens
today
when
we
change
our
custom
types.
D
So
if
you
have
an
example
of
an
objective-
and
let's
pretend
we
want
to
change
this
to
a
key
result
which
doesn't
make
sense,
so
in
this
case
we
would
just
get
an
error
that
this
is
not
possible.
But
assuming
we
want
to
move
this
objective
to
an
issue
because
we
maybe
use
the
wrong
type
in
the
beginning.
So
I
need
to
remove
the
child
item
that
I
have
here,
so
otherwise
it's
not
convertible
and
assuming
I.
Remove
that
and
change
the
whole
thing
to
an
issue.
D
D
Yeah,
it
was
the
wrong.
Well,
it
was
the
wrong
one.
Sorry
so
now
I
said
so.
What
happened
now
I
changed.
I
changed
I
had
the
wrong
idea
here,
but
what
I
changed
so
I
changed
it
from
objective
to
an
issue
type.
What
changes
like
we
removed
the
progress.
D
It's
here's
a
system
note
for
that,
because
an
issue
doesn't
have
a
progress
and
to
change
the
issue
type
to
issue
just
now.
The
rest
keeps
the
same
like
we
have
the
same
as
Chinese
labels
and
if
I
reload,
the
UI
also
updates
them,
and
we
now
have
all
the
things
that
are
available
for
issues.
D
Let's
pretend
I
like
add:
weight,
I
had
a
milestone
at
the
iterations
and
so
on,
and
let's
say
issue
was
also
the
wrong
type.
I
want
to
change
the
test
case
so
submitting
this
notation
again
will
actually
delete
all
data
that
was
but
it's
not
available
in
the
test
case
anymore.
So,
if
we
reload
again,
we
only
see
what's
available
for
test
cases
and
at
least
have
like
the
data
available
that
was
deleted
in
our
system
notes.
D
D
This
is
currently
working
progress,
It's
Like
on
a
review,
but
if
this
gets
merged,
I've
found
one
box
today
we
should
be
able
to
move
basically
any
work
item
to
any
other
one
and
have
correct
data,
and
this
gives
us
at
least
an
easy
way
to
promote
key
results
to
objectives
and
back
as
long
as
they
are
valid.
Yeah.
That's
pretty
much
the
demo
thanks
for
watching.
B
Cool
so
looks
awesome
as
somebody
who
has
not
been
in
the
weeds
with
work
item
type
conversion
like.
Why
do
we
remove
those
that
data
instead
of
just
keeping
it
but
not
showing.
D
It
because
of
we
have
the
data
that
is
like.
Let's
pretend
Jan
made
a
good
issue.
Writers
is
where
we
should
keep
that.
Let's
pretend
we
have
like
a
issue
that
has
a
weight
of
five
and
then
we
convert
it
to
a
test
case,
and
every
time
we
want
to,
for
example,
accumulate
all
the
weights
that
we
have
available
like
that
are
accumulated
over
all
issues
on
this
table.
We
also
need
to
keep
care
of.
D
Like
is
this
data
actually
accurate,
like
does
the
it
should
work
item
that
currently
has
this
data
on
it
like
we
have
data
that
is
not
visible
to
the
user
either.
So
it's
it's
strange
and
actually
other
I
think
other
system
students
as
well
like
Amanda,
did
research
and
I
think
Nick
as
well.
So
it's
a
very
destructive
action
and
we
can
even
in
beforehand
show
the
user
what
data
will
be
deleted.
A
D
D
D
I
only
know
like
from
One
Direction
but
Amanda.
Maybe
you
know
more
about
like
going
back.
E
Yeah
I
I
just
tested
in
a
couple
of
different
applications
and
the
going
back
only
had
the
data
available
that
was
now
in
the
new
type.
It
didn't
carry
forward
its
original
data
I'm
sure
there
are
some
systems
that
do
that,
but
the
ones
that
I
looked
at
did
not.
B
Right,
so
when
do
you
think
we'll
be
able
to
to
change
key
results
to
objectives,
because
we're
currently
talking
about
this
between
the
EMS,
like
at
the
minute,
we
have
to
like
delete
and
objective,
delete,
a
key
result
and
create
an
object,
and
so
on.
Yeah.
D
Both
merge,
requests
and
review
I
hope
they
get
merged
in
this
Milestone
and
then
we
still
need
the
front
end,
but
we
also
have
a
task
like
an
issue
for
a
quick
action,
so
this
will
be
easier
to
implement.
So
you
can
like
promote
I
know
we
don't
want
to
use
to
promote
anymore
but
like
at
least
convert
a
work
item,
and
so
we
can
add
a
quick
action
for
you,
then
so
maybe
two
three
weeks
awesome
thanks.
E
Real
quick,
John
I'm
sorry
I
just
want
to
highlight
that
point,
though,
like
how
important
this
is
going
to
be
for
the
next
q
Q2
okrs
in
get
lab
they're
going
to
start
writing
those
soon
right.
They
do
it
like
eight
weeks
out
or
something,
and
so
this
is
the
time
when
people
accidentally
will
create
something
as
a
key
result,
and
they
meant
it
to
be
an
objective
or
they
later
find
out
that
it
needs
to
be
an
objective
because
it
needs
to
be
children.
E
G
B
B
Yeah
sorry
Jackie,
just
on
that
point,
we
were
working
really
closely
with
avalash
when
we
first
built
okrs,
but
is
that
still
ongoing
now
like
do
we
have
a
plan
to
get
okrs
in
front
of
regular
customers
at
some
point
soon,
yeah.
E
Yeah,
we
are
still
burning
down
s
that
we're
contributing
to
just
they're
related
to
work
items
and
the
okay.
Our
feature
will
need
that,
but
the
okr
specific
work
that
is
needed
for
General
availability
is
is
going
to
be
navigated
by
abolish
but
I'm
going
to
drop
in
the
the
zoom
chat.
Here.
The
link
to
kind
of
how
we're
classifying
the
work
that
both
seg
and
the
plan
teams
are
doing
and
how
it
kind
of
rolls
up
to
availability
for
okrs
for
reference.
G
Good,
how
can
can
hear
me?
Okay,
I,
like
the
Monday
technology
problems,
so
the
agenda
item
I
added-
is
just
around
updates
on
the
work
item.
Okr.
G
If
you
aren't
aware,
there's
a
design
objective
around
like
iterating
on
the
design
direction
for
work
items.
There
are
some
gaps,
there's
some
questions
that
people
have
had
and
we're
trying
to
Wrangle
all
of
those
identify
what
the
gaps
are
see.
We
were
real
confident
in
the
direction
and
where
we
might
have
to
do
some
additional
validation
or
you
know
just
additional
prototyping
and
stuff
like
that.
So
that's
there.
G
We
got
a
late
start
because
of
the
reorgs
and
all
of
the
changes,
so
we're
kind
of
just
we're
diving
in
really
deeply
now,
there's
three
KRS,
but
on
the
top
level
objective
I
couldn't
link
to
the
comment.
But
if
you
scroll
down
into
the
comments
I'm
going
to
add
a
progress
update
just
with
a
couple
notes
about
what's
going
on
and
then
when
we
have
stuff
to
share
out
we'll
definitely
share
it
either
in
this
meeting
or
some
other
way.
E
Yeah
I
had
a
question,
but
you
can
link
to
the
comments,
even
though
there
isn't
a
control
for
it.
You
just
right
click
on
the
the
time
next
year,
name
just
for
reference.
Oh.
G
Okay,
yeah.
E
Yeah,
so
my
question
is:
whenever
we're
working
on
large
initiatives,
I
think
that's
really
good
to
like
have
benchmarks
or
Milestones
that
we
could
have
as
a
readout
or
even
ideally
use
bring
in
to
the
planning
of
the
debit
cycle.
So
I'm
wondering.
Is
there
any
way
to
break
these
down
into
something
that
you
could
either
demo?
Maybe
in
these
meetings
of
progress
or
something
that
we
can
actually
take
and
start
planning
into
the
dev
cycle,
because
waiting
a
full
quarter
is
a
long
time
for
us
to
yeah.
G
G
The
first
KR
is
around
inventorying
the
current
and
proposed
patterns
and
changes,
and
then
the
second
KR
is
kind
of
page
level
patterns
and
the
third
KR
is
like
IA,
so
navigation
and
workflow,
and
we're
doing
we're
not
necessarily
doing
them
in
that
order,
because
you
kind
of
have
to
we
were
a
little
bit
too
much
in
the
weeds
it
felt
like
so
we
went
up
and
we
started
working
on
kr3,
which
is
more
IA
and
then
once
we
have
kind
of
a
better
framework
there
we
could
we
could
dig
into
the
weeds
more.
G
So
that
being
said,
it's
not
like
immediately
obvious
to
me
what
actual
usable
outputs
like
that
y'all
could
take
and
run
with
in
terms
of
Dev.
But
if
you
have
certain
things
in
mind,
then
I
would
say
chat
with
Melissa
because
we
I
know
she's
out
this
week,
but
we
we
meet
with
Melissa
every
week
and
talk
about
the
progress
on
the
work
items
KR.
So
if
there's
something
you're
looking
for,
maybe
we
could
maybe
we
could
look
at
it,
but
I'm
not
sure.
H
Because,
like
it
feels
to
me
that
we
should
have
this
information
architecture
breakdown
before
we
kind
of
move
on
with
all
of
the
work
items
that
the
group
level
stuff,
so
it
does
feel
like
this
is
a
blocker
you're
moving
forward
with
all
of
the
implementation
or
the
question
would
be.
How
much
is
this
a
blocker
to
whatever
we're
trying
to
do
now
in
a
way.
H
G
Yeah
and
well
I
can't
speak
to
like
what
happened
before,
but
it
does
I
do
feel
like
it
does
feel
like
we're
a
little
behind,
so
we're
trying
to
move
really
fast
faster
than
we
normally
would
and
the
product
designers
were
kind
of
asking
about
that
and
I
was
like
I.
Think
it's
because
we
feel
a
little
behind
like
some
things
are
already
being
worked
on
and
already
done
and
in
motion,
and
it
doesn't
necessarily
mean
that,
like
the
design
direction
that
we
reach
would
require
anything
the
change
on
the
back
end.
G
Maybe
on
the
front
end
I,
don't
I,
don't
know,
but
I
I,
don't
know
I,
don't
know
what
to
say.
I
think
we
still
have
to
kind
of
keep
iterating
and
going,
but
it
is
important
to
do
this
work.
So
I
would
say
it's
better
late
than
never.
E
G
I
mean
the
so
we're
we're
using
the
mental
model
study
to
inform
like
workflows
and
how
things
should
be
organized
in
the
app
and
they'll
have
had
that
for
a
little
while
so
I
I,
don't
know
that
it's
going
to
be
vastly
different,
and
if
it
is
it's
for
a
really
good
reason,
and
it's
based
on
Research
that
we've
done.
I
Yeah
I
mean
we're.
You
all
are
also
taking
into
account
the
current
state
I'm
assuming
right,
so
it
doesn't
necessarily
mean
that
the
the
way
that
things
are
currently
in
are
going
to
be
completely
revamped
I'm,
assuming,
as
we
figure
out
the
implementation
plan,
will
have
to
take
into
account
okay.
This
is
in
order
to
do
this
part.
It
would
require
a
complete
change
of
this,
so
maybe
there's
not
a
whole
lot
of
value
in
actually
doing
that.
I
So
I
think
like
once
once
this
is
done,
we'll
have
to
sit
down
and
kind
of
plan
it
plan
everything
out,
but
I
don't
think
we
should
block
any
of
the
dev
work
or
any
of
the
design
work.
That's
already
been
done
for
group
work
items
or
any
of
the
work
items
work
we're
doing
this
quarter.
G
B
E
Well,
I'll
just
call
out
that
I
want
to
understand
what
folks
would
expect
when
converting
an
issue
to
an
epic
which
has
child
or
related.
What
would
you
expect
to
happen
to
those
tasks,
and
no
answer
is
right
or
wrong.
Just
want
to
know
what
you
expect
we're
going
to
be
conducting
or
I'm
going
to
be
conducting
some
validation
around
this
to
figure
out
what
we
should
be
doing
and
there's
been
several
very
good
related
discussions
mentioned
in
that
right
up
there
in
people's
comments
about
well.
E
J
The
only
thing
I
was
going
to
say
is
that
we
have
a
database
table
that
we
added
for
relationship
types
and
restrictions
which
will
eventually,
in
the
long
run
what
customers
sort
of
configure
their
own
relationships.
J
So
I
think
whatever
we
do,
we
should
start
with
that
and
then
make
it
also
not
static
and
then
great
use
your
feedback,
whatever
happens
just
so
it's
very
clear
that
was
my
tldr,
but
I
agree
with
what
Brett
was
saying
to
in
terms
of
like
faking
generic
is
sort
of
what
I
would
what
I
would
do
is
organized.
Every
organization
has
a
very
different,
almost
sort
of
tribal
way
that
they
develop
their
own
processes,
and
things
like
that
so
sort
of
where
we
can't
do
like
a
one-size-fits-all
approach.
With
this.
J
H
Yeah
I,
agree:
I
think
you
should
follow
the
whatever
priority
definition
you
have
on
that
on
that
work
item
type.
So
if
it
doesn't
Define
any
children,
you
should
get
an
error
message
or
an
awning
saying
that
you'll
convert,
but
whatever
children
are
going
to
be
be
orphaned
or
whatever,
whatever
that
is
because,
like
once,
you
define
a
hierarchy.
You
cannot
really
break
that
definition
in
a
way
because
then
you
get
done
inspected,
children
and
all
that
that
relationship
is
now
no
longer
defined,
as
as
you
wanted
it
to
be
defined.
F
And
the
other
thing
you
need
to
be
careful
of
is
is
having
two
different
models
of
handling
data
when
you
convert
from
one
type
to
another
so
and
some
you're
going
to
delete
the
data
and
then
others
you're
going
to
break
a
relationship.
So
we
just
that
just
needs
to.
If
we
do
that.
That
just
needs
to
be
very
clear.
A
H
When
we
convert
a
task,
but
we
convert
the
time
to
another
type,
we
should
be
showing
which
data
is
being
removed,
sort
of
in
a
way
or
cleaned
up,
and
similarly,
when
you
convert
it
to
a
different
type,
when
it
has
children,
we
should
be
showing
that
these
children
are
either
removed
completely
or
the
the
relationship
is
is
removed
or
whatever
we
decide
like
whatever
we
decide
to
do.
We
need
to
be
very
explicit
about
what's
happening
between
in
between
that
conversion
and
in
between
that
transition.
B
It's
funny
because,
like
we
just
had
this
conversation
about
weight,
and
even
in
that
context,
it's
the
exact
same
thing,
because
if
you
convert
an
issue
to
a
test
case,
we
clear
the
weights
right.
Why?
Because
the
way
the
context
is
different
like
the
way
it
might
be
different
for
this
piece
of
work
as
a
test
case
than
it
would
be
as
a
as
a
as
an
issue.
But
if
we
converted,
like
you,
know
an
issue
to
an
epic
and
they
both
have
tasks,
we're
going
to
retain
the
tasks
why
you
know
like
why?
C
Maybe
a
weekend
just
a
wild
idea
mixable
and
offer
the
user
a
choice
like
what
would
you
like
to
do
with
tasks
if
we
are
promoting
issue
to
Epic,
yes
twice
as
much
work,
but
if
we're
going
to
have
issues
on
Epic
and
if
we
allow
tasks
on
it,
it
makes
sense
to
be
to
ask
user.
What
do
you
want
to
do
with
the
tasks.
D
I'm
I'm
I
like
to
get
a
bit
into
technical
details
so
like
I,
make
this
like
very
specific
notation
right
now,
with
like
work
item,
convert
notation,
one
of
our
the
like
future
ideas.
That
happened,
maybe
in
my
mind,
was
that
you
can
grind
around
it.
So
maybe
we
could
like
drive
around
this
notation
you
get
in
front
and
all
the
deletions
that
would
happen,
and
then
we
could
ask
the
user
like
what
do
you
want
to
change
it
to?
D
H
I
I
think
they've.
The
user
decision
is
kind
of
nice
to
do
once,
but
then
like
on
every
conversion.
If
you
have
to
do
that
decision
over
and
over
again,
it
kind
of
gets
annoying
like
I.
Do
it
once
twice
like
ten
times?
Do
I
really
need
to
tell
you
that
every
time
I'm
converting
I
want
these
items
to
go
into
this
specific
type,
so
yeah
there
are
pros
and
cons.
Definitely
so.
B
I
think
chat
cpts
for
coming
the
answer
to
everything.
It's
fine.
B
We
can
talk
about
that
later:
okay,
I'm
gonna,
jump
on
Brett.
You
had
a
demo
for
us.
F
So
really,
this
is
a
very
simple
demo,
I'm
I'm
playing
with
well
that
doesn't
matter,
okay,
let's,
let's,
let's
just
do
this,
can
you
see
you
can
see
the
screen.
F
Okay,
so
moving
around
doesn't
do
any
good
okay,
so
basically,
what
I'm
doing
is
playing
with
replacing
our
markdown
parser
with
a
a
rust
version,
because
we're
going
to
be
we're
going
to
have
to
do
this
at
some
point,
and
this
is
going
to
provide
us
some
more
extensibility
Etc.
So
just
real
quick,
if
you
look
down
here
at
the
price,
we've
got
the
text
equals
equals
a
test
of
the
emergency
broadcast
system.
F
I've
got
some
strong,
some
Bolding
and
some
italics
in
there
and
if
I
run
that
through
bonsai
I
get
what
typically,
what
you
would
expect.
The
difference
you
see
here
that
is
visible
is
the
data
source.
This
is
actually
we
don't
get
that
currently.
F
This
actually
gives
us
position
on
inline
elements,
and
this
is
going
to
be
very
beneficial
to
the
content
editor
in
terms
of
preserving
the
markdown
as
it
moves
back
and
forth.
So
that's
a
that's
a
very
helpful
capability
of
this.
The
neat
thing
here
is
that
this
is
a
little
gem
that
I
put
together
rubygem,
that
has
the
rust
extension
in
it,
and
you
can
see
that
here
in
the
live
Main
and
the
main
piece
here
is
using
the
new
markdown
at
rust
package
or
crate,
and
so
it's
just
calling
out
to
that.
F
So
that's
it's
very
early
stages,
I'm
still
doing
a
lot
of
Investigation,
but
anyway,
that's
the
that's
the
that's
the
deal.
H
F
Well,
yes,
the
data,
the
data
source.
Actually,
the
source
position
actually
gives
us
the
ability
to
track
lines.
The
position
of
them
I'm,
not
sure
what
you
mean
in
a
different
way.
We
will
be
able
to
have.
F
We
should
have
access
to
the
AST
if
we
need
it,
but
the
neat
thing
is
we'll
be
able
to
push
down
a
lot
of
our
filters,
such
as
the
multi-line
block,
quote:
math
those
things
down
into
a
compiled
extension
that'll,
run
much
faster
and
and
allow
us
to
do
some
of
these
things
that
we
can't
really
do
at
the
Bonsai
level.
Math
is
one
definition
list
is
another
very
difficult
to
implement,
either
before
or
after
the
markdown
processing.
So
doing
it
in
the
parser
is
going
to
be
a
huge
win.
F
It's
not
really
intended
for
that.
It
might
help
a
little
bit
because
of
the
additional
Source
positioning
but
I'm
I,
don't
know
I,
don't.
F
F
This
this
would
give
us
the
ability
to
to
do
some.
Of
that
there
are.
There
are
components
of
the
processing
that
we
can't
do
on
the
front
end
very
because
they
require
database
access,
yeah,
that's
yeah,
but
but
but
being
able
to
package
this
as
a
as
a
wassum
would
give
us
much
closer
front-end
capabilities
in
terms
of
either
previewing
or
whatever
else
so
I
yeah
I
see
a
lot
of
I
see
a
lot
of
possibilities
with
this
cool.
H
But
for
the
for
the
for
the
database
access
if
we
would
replace
those
components,
those
things
with
with
some
like
the
lab
UI
components
where,
where
those
would
make
calls
to
the
database
or
I've
I've
I'll,
share
a
link
actually
with
an
idea
from
Gabriel
I
think
he
was
where,
like
we
can
store
stuff
on
the
local
storage
like
as
cached
and
then
once
that
cash
is
expired,
we
can
make
the
the
calls
to
the
database.
H
So
if
we
have
that
component
as
a
gitlab
UI
component,
then
then
we
can
probably
do
all
the
rendering
on
the
front
end
anyway.
F
But
possibly
one
thing
you
got
to
remember
is
that
we
do
a
lot
of
redacting
So
based
on
user
permissions.
What
you
have
access
at
the
time
which
can
change
underneath
you
various
things
like
that,
so
I'm,
not
against
that.
But
there's
there's
definitely
some
challenges
there.
H
Yeah
yeah,
so
the
thing
is
that
gitlab
component
would
only
contain
a
reference
as
an
ID,
and
then
it
will
send
a
request
to
the
to
the
back
end,
basically
and
return
the
the
contempt
as
as
a
result
as
a
response
to
that
anyway.
Yeah
makes
sense.
H
I'm
not
sure
we
do
have
a
bit
of
time
right,
so
I
can
share
my
screen
now,
so
this
I
think
I've
downloaded
part
of
this
some
time
ago
about
work
items
being
at
the
group
level.
So
you
can
see
this
is
a
group
and
you
can
see
it's
not
very
visible.
C
H
Not
sure
if,
if
I
need
to
make
it
larger
or
not,
but
you
can
see
that
the
sum
of
these
issues
are
or
tasks
are
created
exactly
at
the
group
level,
and
then
there
were
a
couple
at
the
project
level.
So
what
I've
been
working
on
is
work
item
creation
mutation.
H
So
if
I
have
this
new,
it's
like
I,
don't
have
the
Y,
it's
just
the
back
end
part,
but
basically
I
do
have
a
backend
where
I
can
create
a
key
Result
One
fill
like
on
the
UI
will
show
a
an
error,
but
in
fact
it
is
created
at
the
group
level,
and
this
is
a
POC.
So
this
is
not
all
in
the
Master
yet
and
it's
not
really
production
ready,
but
you
can
also
browse
our
work
item
at
the
group
level
now,
and
you
can
also
browse
the
this
one
at
the
project
level.
H
So,
in
order
to
have
that
what
I
have
to
do
is
basically
make
the
namespace
type
into
a
interface
so
like
before
that
you
can
see
where
we,
when
we
want
to
to
pull
the
users
from
a
for
a
work
item.
We
always
look
into
the
project
because
work
items
before
we're
tied
to
the
project
and
what
we
can
do
instead
is
look
them
up
by
the
namespace
full
path,
because
that
will
resolve
to
either
a
project.
And
now
we
have
project
name
spaces.
H
H
H
No,
it
should
not
at
least
not
not
that
I'm,
seeing
what
the
the
breaking
changes
actually
making
sure
all
the
permissions
would
work
in
in
the
sense
that
we
do
a
lot.
We
used
to
rely
a
lot
on
the
project
authorizations
table
where,
where
you
cache
the
rolls
from
groups
into
that
project-
and
we
need
to
skip
that
now,
especially
when
you
pull
the
list
at
the
group
level,
you
want
all
of
the
work
items
from
all
of
the
child
groups
and
child
projects.
H
So
you
cannot
join
only
the
project
authorization
you
at
least
need
to
do
a
left,
joy
and
then
also
compute,
the
memberships
within
the
groups,
because
we
don't
have
a
groups
authorization
table
where
you
cache
the
memberships
within
the
subgroups
and
so
on.
So
that's
where
it's
kind
of
tricky
and
we
need
to
check
with
the
permissions
and
how
how
you
have
access
to
issues
beneath
those
groups
and
so
on.
So
you
only
can
access
subgroups
that
you
actually
have
access
to,
but
otherwise
should
be
fairly.
Okay,
I
think.
B
H
Yeah,
we
definitely
should
feature
flag
because
again,
there
is
potential
in
in
terms
of
not
only
leaking
data.
That's
one
part
of
it,
but
also
a
performance
hit
again
because
we
don't
have
that
cash
version
of
the
memberships
and
and
so
on.
So.
B
H
I
think
we
still
have
a
lot
of
feature
Flags
in
terms
of
like
traversal
ideas,
but
that
may
be
helpful
in
Computing
all
of
these
memberships,
but
again
I,
don't
think.
That's
enabled
like
all
over
the
place
and
all
the
feature
flies
on
I.
Don't
think
all
the
features
lives
are
unable
to
to
be
using
that
I'm
testing
it
out
how
it
works,
but
that
that
can
be
probably
helpful
in
in
doing
some
of
the
performance
optimizations.
D
I
only
have
a
request
to
remove
lots
of
content
from
this
file
because
it's
like
lagging
so
we
make
an
archive
of
the
plan.
Google
log.
B
So
yeah
like
this
is
the
Sledgehammer
that
I
think
we're
applying
to
every
problem
and
I
like
so,
but
yeah
I
was
curious
to
find
out
like
how
effective
it
would
be
to
actually
start
training
this
on
some
of
our
data.
So
if
you're
interested
like
you
can
check
out
that
thread,
I
put
it
in
plan
in
the
plan
Channel,
and
but
it
was
at
the
end
of
a
of
a
Friday
a
minute
time.
So
I
don't
expect
that
everyone
will
have
seen
it
a
couple
of
things
about
it.
B
That
I
think
are
interesting.
First
is
the
cost
they
famously
like
cut
the
cost
by
10x
a
couple
of
weeks
ago,
but
it's
still
quite
expensive,
given
that
it
was
one
issue,
it
was
quite
a
long
issue,
but
nevertheless
it
cost
about
one
and
a
half
to
two
dollars.
B
Just
to
ask
you
know
maybe
like
15
questions
like
something
like
that
and
I
picked
out
some
of
them
more
interesting
ones
and
put
them
in
the
thread
that
said,
like
I'm,
no
expert
I
also
use
their
most
like
accomplished
model
which
is
DaVinci,
which
is
quite
expensive
to
use.
So
I'll.
Try
using
other
Babbage
at
some
point
and
also
I
didn't
do
any
fine
tuning.
So
fine
tunings,
when
you
like
tune
at
once
on
your
data,
and
then
you
ask
it
questions
which
are
the
cheap
part.
B
I
did
the
in
context
method,
which
is
it
basically
trains
it
every
time
you
ask
something
so
yeah
like
anyway,
so
it
was
quite
quite
expensive
and
the
answers
were
like
quite
funny
at
times,
because
they
were
sort
of
hilariously
wrong,
but
it
was
very
confident
about
them
and
then
also
it
remembers
the
context.
So
you
know
anytime,
you
try
to
ask
it
something
afterwards
like
it
would
include
that
same
erroneous
assertion.
B
B
They
were
correct
to
a
point
like,
even
though
they
looked
wrong,
but
we
just
simply
moved
on
since
then
made
different
decisions,
but
it's
also
quite
funny
that
it
tried
to
convince
me
that
I'm
using
a
single
table
for
work
items.
Although
that's
the
decision,
we
went
with
that
it
was
because
it
was
more
performant
and
easier
to
scale,
which
is
like
pretty
much
the
opposite
of
what
we
decided,
even
though
we
decided
to
go
with
it
anyway.
B
But
what's
funny
is
that
it
remembered
that
it
had
told
me
that
and
then,
when
I
asked
questions
afterwards,
it
would
always
refer
to
that.
I
could
ask
it
if
the
sky
was
blue
and
it
would
have
said
the
sky
was
blue
and
also
that
we
should
reuse
the
issues
table
for
everything,
because
it's
easier
to
scale
so
like
some
interesting
stuff,
so
anyway,
like
so
yeah.
This
is
the
end
of
my
big
talk.
B
But
the
point
is
this
one
issue
and
it's
a
long
issue:
it's
got
like
100
comments
or
something
but
I'm
gonna
try
next
to
maybe
train
it
on
the
whole
Corpus
of
get
labor,
gitlabs
issues,
discussions
and
epics,
just
the
public
ones
and
then
see
if
it
gets
any
more
interesting
I'll,
also
try
some
of
the
cheaper
models
as
well
that
are
faster
to
train
and
we'll
go
from
there.
B
This
is
definitely
like
a
Spare
Time
Project,
though
so
I
don't
know
in
the
week
of
Saint
Patrick's
day
whether
I'm
gonna
get
any
spare
time,
but
we'll
have
to
see.
H
B
That's
a
great
idea
and
I
didn't
actually
include
so
here's
the
interesting
thing
about
it
didn't
clean
the
data
at
all
right,
so
what
I
did
was
literally
copy
all
the
discussion
and
everything
but
I
didn't
I
copied
in
system
notes,
but
I
didn't
make
any
attempt
to
expand
the
system
notes.
So,
instead
of
saying
like
this
issue,
was
added
to
an
epic
called
complete
work
items
by
this
time.
You
know
what
I
mean
like
it
just
said.
This
issue
was
added
to
Epic
four
five,
one
whatever.
B
So,
if
you
would
expand
all
that
out
and
really
clean
the
data,
it
might
be
interesting
because
it
wouldn't
know
it
because
I
didn't
give
it
that,
but
it
would
know
it
from
the
system
notes.
So
it
would
have
seen
that
the
way
it
was
added,
so
it
would
be
interesting
actually
to
to
train
it
on
and
to
try
and
include
things
like
weight
and
everything
on
the
whole
public,
gitlab,
Corpus
or
whatever,
and
and
then
try
to
like
give
it
tasks.
Like
estimate,
this
issue.
H
Yeah
yeah,
like
I,
was
thinking,
estimate
and
then
learn
like
three
iterations
where,
like
what?
What
can
we
achieve
in
next
quarter
or
something
like
that?
I'd
be
curious?
If,
if
that
would
be
something
that
we
can
do,
but
because
that's
where
I
think
the
cost
of
It
kind
of
pay
for
themselves
in
a
way?
If
you
can
do
that
kind
of
where
you
can
get
the
the
benefit
of
paying
for
such
such
a
thing.
But
yeah.
B
Yeah
one
one
of
the
things
I
thought
was
interesting
was
so
it
recommended
a
team
member
former
team
member
now,
but
a
team
member
who
I
asked
it
couldn't
recommend
somebody
who
I
could
go
to
to
ask
further
questions
and
it
recommended
this
former
team
member
who
was
heavily
involved
in
the
decision
making
around
database
issues
what's
funny
about
it
is
that
it
recognized
that
that
person
it
it
that
person
was
a
maintainer
but
Naya's.
A
former
team
member
has
contributor
access.
B
It
understood
the
word
contributor
and
then
told
me
that
he
had
made
contributions
to
the
project
and
it's
very
subtle,
but
it's
also
very
interesting
that
it
understood
that
that
role
do
you
know
what
I
mean
and
that
it
kind
of
understood
like
when
I
asked
for
some
recommendation
about
who
I
could
talk
to
it
told
me
all
about
this
person
and
the
contributions
that
they've
made
and
so
on
so
yeah.
It's.
It's
definitely
interesting.
I.
B
Yeah,
so
there
are
a
few
different
things
you
can
do
with
it.
I
didn't
I
didn't
put
that
in
a
thread
because
I
did
it
late
on
Friday
I'd
have
just
shared
curiosity,
but
there's
a
summary
mode,
and
it
was
quite
good,
but
I
think
it
needs
a
lot
more
tuning
to
to
give
decent.
You
know
reliable
summaries.
B
One
thing
that
I
think
might
be
interesting
actually
is
just
to
give
it
the
discussion
element
and
do
an
auto-generated
summary
of
any
discussion
in
any
issue,
because
some
of
them
get
to
a
certain
point
where,
if
you're,
trying
to
catch
up
on
an
issue
which
I
try
to
do
a
lot,
especially
when
I
know
when
we're
planning
the
next
Milestone.
If
I
need
to
know
like
very
quickly
if
I
have
like
10
issues,
I'm
considering
and
I
need
to
know
what
the
current
state
is.
B
It
would
be
very
interesting
to
have
like
just
give
me
a
summary
of
these
60
comments.
I
don't
have
time
to
read
them
all
and
that
it
would
give
you
some
rough
idea
of
the
major
decisions
that
were
made
and
things
like
that.
It
seems
like
it's
pretty
much
designed
for
that
kind
of
work,
asking
questions
and
giving
summaries.
B
If
you
have
any
questions,
you'd
like
like
me
to
ask
until
my
original
ten
dollars
of
credit,
runs
out,
you're
welcome
to
pass
them
along
and
I'll.
Give
you
the
answers
in
the
thread.