►
Description
In this session we pair on and discuss some BE topics.
00:00 - Discussing a new DB seed script
14:00 - Testing out the seed and talking about the next problem
17:55 - Refactoring EE permission check out of CE
40:30 - Discovering that the root problem actually doesn't actually exist
42:42 - Testing out dead code we ran into
54:20 - Final question about DB seeds
B
D
B
Right,
let
me
copy
and
paste
a
couple
of
messages,
so
I
am
oh
they've
kind
of
smushed
together
a
little
bit
there,
but
I
I've
recently
created
an
issue
to
to
sort
of
suggest
that
we
start
encouraging
like
more
seed
data
creation,
because
I'm
sure
you
find
this
all
the
time.
Paul
right,
review
and
merge
requests
the
recreation,
slash,
validation,
steps
are
like
right,
go
into
the
rails,
console
and
and
run
these
commands
and
create
these
models.
And
then
you
know
go
here
and
do
this
and
all
this
kind
of
crazy
stuff.
B
E
B
Working
on
the
achievement
stuff
at
the
moment-
and
you
know,
I-
should
make
sure
that
we've
got
achievements
in
the
database
user
achievements
in
the
database
and
revoked
achievements,
and
you
know
a
bit
of
a
mixture
of
data
so
that
anyone
that
picks
it
up
today
in
the
future
a
week
a
year
a
month,
you
know,
can
hit
the
ground
running
rather
than
have
having
spend
a
bunch
of
time
sort
of
manually
seeding.
The
database
CI
is
one
of
the
hardest
ones.
B
C
Well,
that's
already
come
up
actually
when
we
were
just
chatting
because,
yes,
Marco
can
tell
you
more
about
the
seed
data
he's
putting
together,
but
it's
like
when
you
try
to
make
it
random
by
saying,
like
first
eight
projects
do
this:
it's
not
always
clear
from
just
reading
the
code
which
eight
projects
received
the
data
right,
so
you
have
to
kind
of
go
into
your
console.
Look
at
your
own
database,
State
and
figure
out
where
you
can
get
look
at
this
data.
So
we
that
kind
of
came
up
already,
but
Marco.
E
F
B
B
A
B
Well,
we
we
kind
of
covered
that
in
some
matches
that
so
we
were
saying
these
scripts
pretty
good
at
the
moment,
because
it
will
look
if
there
are
any
merge,
requests
and
any
issuables
sorry
issues
or
merge
requests
and
for
and
it
will
create
time
logs.
But
if
there
aren't
any
it
will
just
continue.
B
A
E
C
Yeah,
okay,
but
going
back
to
your
original
question
about
either
item
potency
of
this
script,
I
believe,
if
you
were
to
run
this
seed
script
as
Marco
has
it
today
it
would
increase
the
number
of
time
logs
every
time
you
run
it
right.
So
it
is
not
item.
Potent
I
would
argue
that
maybe
it
should
like
do.
We
want
to
get
into
a
situation
where
maybe
an
issue
has
like
thousands
of
time
logs
right,
because
you've
run
your
seeds
over
and
over
again
I.
B
B
If
the
report
explodes
when
I,
when
I
do
X,
I've
I
think
you've
got
almost
got
free,
free
ideas,
you've
got
the
script
explodes
if
you
try
and
run
it
and
it's
already
been
run
before
you've
got
the
scripts.
If
you
run
it
a
second
time,
won't
kind
of
it
will
almost
not
run
and
say:
oh
I've
already
been
run
so
I
don't
want
to.
Maybe
even
something
super
crazy
that
strips
out
what
it
put
in
originally
and
then
starts
clean
and
re-puts
it
back
in
to
kind
of
really
recreate
it.
But.
A
Sorry
what
if,
instead
of
this
finding
issues
to
attach
time
blocks
to
this
created
the
issues
it
was
going
to
attach
timelines
to
and
that
way
we
could
create
an
issue
of
like
call
it
warm
ipsum
with
time
logs
and
then,
if
that
issue
already
exists,
we
know,
can
we
run
this?
We
don't
have
to
do
it
again
or
something
like
that,
and
that
way
we
may
have
the
and
dipotents.
A
B
So
I
mean
what
a
lot
of
the
scripts
do:
well
platonic,
that's
getting
back
to
the
impotent
one!
Now
I,
don't
know
what
the
platonic
there
are
a
number
of
them
right
that
have
a
flag
that
you
can
set
to
say
like
how
many
or
or
something
along
those
lines.
So
maybe
I
I
feel
like
we're
going
a
bit
overboard
realistically,
like
what
would
be
nice.
A
B
I
can't
I
haven't
been
able
to
decide
yet
whether
I
think
Randomness
I
went
with
Randomness
previously,
but
I
I
can
see
a
benefit
for
predictability
over
Randomness
as
well,
so
I
think
but
then
I
guess
the
the
whole
singing
all
dancing
would
be
a
script
that
looks
at
the
environment
variables
and
you
can
tell
it
either
how
many
projects
you
want
to
seed
for
or
how
many
time
logs
you
want
to
seed,
or
in
your
case
it
looks
like
you
were
only
even
gonna
seed
them.
If
you
actually
set
a
flag
as
well.
A
A
A
B
B
I
haven't
read:
I
saw
the
email,
but
I
I
mean
I.
Think
you've
had
my
conversations
that
I've
found
that
if
you've
got
self-hosted
instance,
you
pretty
much
need
a
single
root
group
or
a
lot
of
functionality.
Just
doesn't
work
anyway,
so
I'm
not
too
adverse
to
that
as
a
first
iteration
anyway,
but
would
be
nice
if
it
could
be
instance-wide.
B
A
B
C
B
E
D
A
B
B
E
A
Because
we
have
customers
that
are
groups
within
there
and
like
are
we
going
to
be
I?
Don't
know,
it
just
seems
maybe
strange
that
there's
a
view
for
us.
If
this
is
an
admin
only
view,
maybe
that
makes
sense,
but
hopefully
we
don't
at
the
very
least.
Hopefully
we
wouldn't
expose
any
private
groups
time
logs
or
something
like
that,
like
that.
B
No,
no
no
100
would
adhere
to
policies
that
already
in
place,
so
the
the
time
log
policy
without
diving
into
it,
definitely
takes
into
consideration.
Can
you
see
the
issue
or
merge
request
and
whether
you
can
see
the
issue
or
merge
request
depends
on
whether
you,
whether
it's
confidential,
whether
the
Project's
private,
whether
you're
a
member
of
the
project,
etc,
etc,
but.
B
C
C
C
C
B
C
Yeah,
so
that's
the
big,
that's!
The
fish
I
have
to
fry
that
I'm,
avoiding
by
doing
this
refactor.
Okay.
So
it's
a
really
simple
concept,
which
is
that
or
simple
it's.
It
should
be
straightforward,
which
is
that
we
have
these
methods
called
node
visible
to
user
in
lib
Bonsai
reference
parser
directory
I.
C
Don't
know
the
first
thing
about
what
lib
Bonsai,
but
that's
fine,
because
I'm
just
trying
to
move
this
code
to
EE
and
looking
at
this,
it's
a
combination
of
like
overriding
something
in
EE
and
then
it's
also
calling
super
to
a
base
class,
that's
overwritten!
So
this
is
the
code
I
just
commented
out.
This
is
what
is
on
master
I.
Have
these
two
methods
in
different
classes
that
look
very
similar
and
then,
if
you
go
up
to
this,
doesn't
exist
a
master?
C
If
you
go
up
to
like
the
base
parser
for
these
non-ee
classes,
this
is
what
nodes
visible
to
user
does.
So
it
actually
refines
projects
which
doesn't
seem
ideal
right,
because
here
we're
finding
projects
and
then
we're
pre-loading.
These
custom
roles
yeah
and
then
we
super
app
and
we
I
don't
know
what
this
lazy
thing
does,
but
I
think
we're
finding
projects
again
and
then
doing
some
other
stuff.
D
C
A
C
C
D
A
C
C
C
C
C
C
B
A
B
B
F
F
C
Yes,
yes
and
I,
but
I
would
like
to
get
around
that
and
I
mean
the
other
thing
complicating.
This
is
that
there
are
other
classes
that
inherit
from
base
parser
right,
so
I
am
I'm
not
wanting
to
like
change.
I
can't
remove
projects
from
here,
because
at
first
I
was
like
oh
just
remove
this
line
right
and
then
I'm
only
evaluating
once,
but
there
are
other
classes
that
inherit
from
base
parser
that
don't
override
this
method.
So
I
want
to
keep
the
behavior
of
this
method.
The
same
could.
E
A
A
B
A
A
A
A
C
Yes,
so,
instead
of
passing
a
collection
just
pass
a
Boolean
that
tells
us
if
we
want
to
load
the
collection,
but
then
how
would
we
get
access
to
projects
in
the
super
in
the
base
class
if
we
weren't
passing
it
through,
we
wouldn't
have
access.
It's
not
like
an
instance
variable.
It's
like
not
so
since.
A
A
C
A
It'll
call
the
the
ee
method
and
then
Super
is
going
to
call
the
CE
map,
but
it's
strange
because
what
I
think
the
strangest
thing
is
that
these
behave
like
mixins.
They
don't
really
behave
like
extensions,
yet
somehow
we
can
still
call
this
super
oven,
which
is
I,
think,
is
some
of
the
ee
magic
that
we're
doing.
C
Yeah
and
then
that
will
include
this
and
then
I'll
call
Super
and
I'll
pass
in
the
projects
to
Super
so
that
we
don't
look
up.
Do
the
lookup
again,
okay,
that's
the
piece
that
was
missing
for
me
before
I
hadn't
considered
that
I
was
like
gosh
I,
like
I,
don't
know,
I
was
actually
to
find
a
method
that
I
like
one
of
them
called
preload,
and
one
of
them
doesn't
like
an
empty
method
in
CE
or
something
I've.
C
A
C
A
C
C
A
So
what
this
can
mean,
so
here's
here's
something
that
might
be
going
on
there
I
feel
like
Steve
Abrams,
is
going
to
be
able
to
correct
me
so
much
on
this,
but
it's
possible
that
the
ee
base
parser
is
like
an
ee
specific
one
that
we
don't
necessarily
want
to
override
into
CE.
So
we
have
ee
specific
code,
which
we
won't
do
the
whole
pre-pin
mod
with
that's
the
reason
why
I
would
think
that
doesn't
it
does
not
exist.
D
A
Where
No
One's
Gonna
initialize
bass,
it's
kind
of
like
an
abstract
class,
I,
guess
and
so
I'm
assuming
we
have
other
parsers
and
so
I'm,
assuming
we
have
a
collection
of
CE,
specific
parsers
and
then
it
sounds
like
we
must
have
a
collection
of
ee
specific,
only
parsers.
The
way
like
Bonsai
works
is
I.
Guess
we
just
run
through
a
list
of
all
these
parsers
to.
F
Wait
what
I'm,
confused
okay,
yeah
I'm,
so
what's
I
mean
I
I!
Think
the
way
I
think
about
the
ee
stuff
is
pretty
much
like
when,
when,
when
we
put
the
code
together,
the
ee
file
and
the
regular
CE
file
are
combined,
so
you
can
imagine
them
being
like
one
file
essentially
so
when
you're
running
ee,
you
can
just
imagine
that
you
just
have
one
file
anything
that
was
overwritten
in
ee
kind
of
just
like
takes
over
and
then
inheritance
just
works
as
you
would
expect.
F
C
C
F
C
C
Let's
see,
get
diff,
oh,
my
goodness,
forward
forward,
no
digital,
the
user,
that's
the
one!
So
nodes
visible
to
user
is
I've
deleted
it
from
here,
because
I'm
going,
but
in
theory
this
basically
nothing
should
be
doing
preload
custom
roles.
If
it's
not
in
ee,
because
custom
roles
is
an
ee
only
feature
so
I'm
just
trying
to
remove
any
references
to
customers.e.
D
C
D
A
C
C
We
only
need
to
do
this
custom
rules
preloading
if
the
authorization
check
is
for
read
codes.
This
is
read
group,
so
there's
no
need
to
preload
custom
roles,
because
read
group
isn't
part
of
custom
roles
right
now,
but
read
code
is
part
of
custom
roles.
So
if
we're
gonna
say
can
read
code
somewhere
I'm
going
to
have
to
do
that
preload,
so
I
don't
want
it
to
be
for
every
single
flat
of
these
one
of
these
reference,
parser
classes
just
for
the
ones
that
do
that
authorization
check.
D
C
A
C
F
C
F
C
C
C
I
discovered
that
we
were
preloading
so,
and
this
is
something
that
only
I
could
know,
but
when
we
first
sell
custom
rolls
the
very
first
iteration
was
actually
adding
a
different
permission.
It
was
adding
download
code
and
then
we
got
feedback
that
actually
like.
No,
no,
the
customers
just
want
read.
Code
and
I
was
like.
What's
the
difference
between
reading
code
and
downloading
code,
like
anything
you
can
read,
you
can
download
right,
but
it
was
determined
that
they
would
prefer
to
like.
C
Not
have
the
download
button
not
have
the
Clone
option
available,
like
they
can
load
the
repo
and
look
at
the
files.
So
this
new
permission
was
added.
They
split
out
read
code
from
download
code
so
that
there
there's
this
like
ability
just
to
look
at
a
repository,
and
if
you
really
wanted
to
you
could
copy
the
HTML
on
every
page.
But
you
know,
but
we're
not
making
it
easy
for
you
to
download
the
code.
C
And
so
we
switched
from
debt
from
download
code
to
read
code
for
our
customer
MVC
and
so
I
wrote
this
preloading
logic
when
it
was
download
code
and
I
was
getting
all
these
n
plus
one
errors
for
download
code
and
what
I
just
discovered
or
just
realized.
Is
that
like?
Actually
it's
a
different
permission
now
so
I?
Don't
I
can
just
remove
the
preloading.
A
Was
this
was
right
so,
while
we're
here,
though,
we
did
just
come
across
potentially
a
bit
of
technical
debt
because
it's
unclear,
if
that,
if
we
came
across
dead
code
ghosted
code,
can
we
do
you
happen
to
have
the
GDK
running
yeah?
Well,
can
we
see
if
it's
if
it's
definitely
dead
code,
can
we
raise
an
exception
and
try
to
and
try
to
trigger
it.
C
A
We
remove
it,
then
we
would
remove
the
custom
logic
and
I
want
to
see
if
that
custom
logic
is
even
being
hit.
So
let's
raise
it
and
I
think
what
hits
this
is.
If
you
leave
a
comment,
is
when
Bonsai
is
a
markdown
rendering
pick
so
I
think
leaving
a
a
comment
with
a
commit
reference,
or
something
like
that
would
would
do
it.
F
Yeah
I
used
to
do
GDK
update
not
nearly
as
much
as
I
should
have
and
I
would
just
do
a
pull
rebase
and
then
run
migrations
and
bundle
install
on
on
just
the
gitlab
pro
that
turned
into
every
time.
I
did
a
gtk
update.
It
was
it
didn't,
go
very
well,
so
I
ended
up
doing
like
I
would
do
GDK
update
like
once
a
week
and
then
now
I'm,
not
I'm
on
infrastructure.
Now
so
I
don't
use
GDK
as
much
so
I
just
do
it
anytime,
I
use
it
now.
D
E
F
F
Has
gotten
a
lot
more
stable,
like
over
the
last
year
year
and
a
half,
because
I
remember
like
I,
used
to
be
afraid
to
run
it
if
I
had
like
a
lot
of
work
to
do
like
if
I
was
like
I
need
to
get
some
work
done
today,
I
can't
do
a
gtk
update.
I
can't
afford
to
lose
an
hour
debugging
something,
and
but
now
that
doesn't
seem
to
happen
anymore.
It.
E
D
A
D
A
Yeah
or
or
if
you
do
just
the
just
the
the
incrementing
ID
and
that
was
talking
about
with
the
commit
I
was
just
talking
about
with
the
you-
can
also
any
kind
of
reference
because
we're
talking
we're
targeted
the
base
class,
so
this
seems
like
it's
all
working.
Can
we
try,
then,
let's
just
double
check
that
raising
raises
like
we
expected
to.
Can
we
raise
on
the
CE
class,
then
this
user
can
reference
thing.
A
D
C
A
C
A
F
D
A
A
A
My
experience
with
the
like
the
profoundness
of
its
answers
has
kind
of
tapered.
So
I
don't
know
if
it's
because
I'm
now
more
used
to
it
or
I
have
a
theory
that
with
how
much
it
scaled,
they
had
to
shallow
the
depth
that
they're
running
in
it
and
as
an
addition
to
making
and
handle
all
the
requests.
I
think
part
of
that
clearly
I
think
it's
running
at
a
different
depth
than
it
used
to
all
I
have
to
say
is.
A
That
would
be
interesting
to
me
if
you
subscribed-
and
there
was
maybe
some
sort
of
there
is
some
sort
of
Improvement
on
the
depth
of
the
answer.
Not
just
the
speed
subscription
only
promises
like
an
increase
of
speed,
which
I
find
to
be
interesting,
yeah
cool.
Well,
so
we
can
remove
that
file.
That's
that's
all
we
got
to
do.
D
D
E
C
F
D
A
That
clearly
applies
everything
cool
all
right.
Well,
now
we're
pretty
close
to
time.
Does
anyone
else
have
anything
else
that
they
would
like
to
I
know?
That's
about
it.
I
have
a
question
about
the
seed
stuff,
and
then
we
can
maybe
close
it
because
I
noticed
so
the
desire
was
yeah
overall,
be
great
if
we
can
have
more
seed
data,
especially
for
some
of
these
features
that
are
hard
to
get
to
that
state.
A
One
I
guess
would
be
really
great.
I
need
to
find
the
actual
documentation
for
it,
but
I
noticed
that
we
have
these
incrementing
by
integer
and
I
can
see
it
being
really
possible.
If
we
really
want
to
encourage
more
developers
to
adding
seeds,
we
end
up
with
two
Mrs
with
the
new.
You
know:
incremented
integer,
you
know
38
seed,
food
38
seed
bar
they
in
parallel
both
pass
review
and
get
merged.
F
F
Up
do
you
think
we
should
like
I,
feel
like
there's
a
lot
of
non-visibility
around
seating,
because
it's
something
that
I
definitely
wasn't
aware
of
at
first
and
then
spent
a
lot
of
time
trying
to
like
Advocate
my
team
to
spend
time
on
and
it
was
kind
of
like
well.
We've
already
got
the
feature,
and
everyone
on
the
team
has
data
seated,
so
it
wasn't
prioritized
and
I
think
that's
common
at
gitlab.
It's
that
you
know
certain
things
like
that.
Don't
don't
get
noticed
until
they're
really
really
needed.