►
From YouTube: hyper TSC Meeting June 1, 2021
Description
* Status Update
* Demo of DynamoDb Adapter progress
A
All
right
welcome
to
the
hyper
technical
steering
committee
meeting
june
1st,
it's
hard
to
believe
it's
already
june.
It's
crazy
just
to
catch
everyone
up
on
kind
of
the
the
road
map
we
you
know
in
may
launch
new
website
and
we're
working
on
kind
of
a
workshop
trip,
and
I
have
been
working
on
a
workshop
that
kind
of
showcases
hyper
in
a
real
world
app
and
we
completed
the
api
kind
of
first
pass.
A
I
would
say
last
week
and
now
we're
starting
to
look
at
the
front,
end
and
scope
out
some
user
stories
and
then
look
to
put
a
front
end
on
top
of
and
then
the
app
is
this
kind
of
movie
review
app
but
trip
you
want
to
add
any
kind
of
details
to
that.
B
Yeah,
it's
it's
basically,
you
know
you
have
a
movie,
letting
the
users
add
the
reviews
to
the
movie
and
then
letting
people
react
thumbs
up
thumbs
down
to
those
reviews.
B
So
it's
a
concept
is
build
the
whole
app
and
then,
as
part
of
the
workshop,
you
would
build
the
middle
tier
connecting
the
piper
to
the
ui.
That's
already
there.
So
you
fill
in
the
middle.
A
Yeah,
so
what
we're
doing
is
actually
building
the
full
app
and
then
we'll
go
back
to
construct
the
workshop
with
kind
of
a
fill
in
the
blank.
So
you
kind
of
build
the
middle
piece
and
you
you'll,
as
you
take
the
workshop.
You'll
already
have
the
front
end
and
of
course,
you'll
already
have
hyper.
A
So
all
you
have
to
do
is
kind
of
work
through
that
that
middle
tier
or
the
api
tier
or
the
application
layer,
whatever
you
want,
and
if
you
want
to
check
out
what
we've
done
to
date,
I've
put
the
repo
in
the
show
notes,
but
I
think
it
I
think
it's
looking
pretty
good
and
and
again
very
much
discovery.
We're
gonna
find
some
pain
points
and
work
on
refining
those
probably
come
down
as
change
requests
to
hyper.
A
And
then
the
other
item,
the
refining,
the
the
dev
service
for
hyper
and
getting
the
documentation
up
to
date
and
probably
going
to
take
a
another
look
at
the
documentation
or
style
of
documentation.
A
A
distinguished
engineer
from
cypress
gleb
actually
posted
a
pretty
awesome,
blog
post
on
just
how
to
think
about
documentation,
and
I
think
it's
worth
unless
anyone
disagrees.
I
think
it's
worth
using
as
kind
of
a
guide
for
our
our
documentation,
I'll
paste
it
in
the
chat,
but
he's
got
some
really
good,
clear
concepts
and
you
know
really
good
layout
on
what
to
put
on
the
website
what
to
put
in
your
kind
of
guides
and
in
workshops
and
what
to
put
in
your
blog.
A
So
if
you
guys
get
a
chance,
I
would
recommend
reading
that
and
maybe
providing
your
input
on
that
and,
let's
see,
if
you
know,
if
you
guys,
disagree
or
agree
or
or
whatever,
so
that
we
can
improve
our
documentation.
A
I
met
with
a
group
today,
that's
going
to
be
taking
over
one
of
the
projects
of
hyper
one
of
our
customers
or
clients,
they're,
switching
the
development
team,
and
so
so
that
was
kind
of
cool
got
to
do
some
onboarding
today
with
this
other
team,
and
I
think
we'll
get
a
lot
of
good
feedback
from
that.
You
know
one
of
the
things
that
they
noted
and
it's
a
team
kind
of
an
agency
like
group,
and
they
noted
yeah.
A
This
architecture
is
a
little
different
than
what
we're
used
to
and
it's
like
well
one
of
the
there's
kind
of
three
things
to
keep
in
mind.
Is
you
know
this
should
improve
testability,
make
testing
easier,
make
it
easier
to
maintain-
and
you
know,
kind
of
reduce
unintentional
technical
debt,
so
I
think
they
seem
pretty
excited
from
the
the
gist
of
it.
But
you
know
they
were
like
wow
this.
This
is
totally
different
than
what
we're
used
to
so
thought.
That
was
some
good
feedback.
C
Did
they
give
any
examples
as
to
what
they
were
comparing
it
to
like
what
they've
worked
with
in
the
past.
A
Yeah
most
of
their
stuff
is
very
database.
Centric,
very,
I
would
say,
model
view
controller
kind
of
style.
From
from
what
I
gathered-
and
you
know
very,
very
much
with
like
business
logic,
kind
of
spread
everywhere,
either
like
heavy
on
the
client
or
heavy
on
the
back
end,
but
the
the
lead
developer
did,
you
know,
understand
you
know
what
it
was
and
and
he
he
referenced,
how
good
the
documentation
was,
how
clean
the
code
was
and
said
that
it
was
really
easy
to
follow.
A
The
only
only
thing
that
he
had
trouble
following
was
the
way
we
implemented
web
hooks
in
this
app
and
kind
of
the
q
service,
because
it
is
kind
of
hard
to
follow
because
they
create
a
cue
with
a
target
and
the
the
api
layer
passes
that
back
to
hyper,
because
the
api
layer
is
serverless,
so
it
can't
maintain
any
state,
so
it
passes
it
back
to
hyper
and
then
when
a
job
is
invoked,
hyper
calls
the
api
and
the
api
calls
the
client.
A
So
it's
kind
of
a
couple
of
steps
there
and
I'm
gonna
once
he
gets
a
little
further
down.
I'm
gonna,
you
know,
walk
through
the
sequence
diagram
with
them,
but
that's
just
not
you
know.
Hooks
and
events
are
not
super
clear.
When
you
look
at
code
right,
it
doesn't
show
you
the
the
path.
So
so
you
need
a
little
bit
more
visualization
to
get
out
of
that,
I
think,
but.
D
A
But
that
was
good.
You
know
again.
They
said
that
the
documentation
was
was
good
still
like.
I
said,
a
need
to
do
better
and-
and
you
know,
really
kind
of
clean
up
some
of
the
old
stuff
and
better
define
kind
of
the
three
pathways.
You
can
run
hyper
two
pathways,
whether
you
want
to
run
self-hosted
or
like
local
development
or
run
on
the
you
know
the
the
dashboard
process
or
the
little
developer
host.
A
So
I've
got
some
work
to
do
there
just
to
tie
that
all
together,
I
still
have
like
hyper63.com
and
some
places
in
the
docs
and
need
to
change
it
all
to
hyperio
and
and
all
that
it's
like
changing
names
is
fun,
but
it's
great
because
hyper
is
a
cool
name.
So
so
that's
really
kind
of
the
current
status.
A
For
me,
you
know
june
we're
going
to
be
looking
at
maybe
trying
to
launch
kind
of
the
dev
service
and
show
the
workshop,
and
I
know
casey
has
been
working
on
the
dynamodb
adapter
okay,
so
you
have
any
updates
to
share
there
yeah.
I.
E
Can
share
whatever,
at
the
end,
all
the
all
the
the
routes
are
working.
I
haven't
tested
three
that
I
sent
before.
I
think
it's
a
bulk
listing
query
they're,
all
just
you
know,
like
stub
functions
right
now,
yeah
everything
works
pretty
well,
and
I
can
talk
a
little
bit
about
some
of
the
implementation,
but
also
some
of
the
things
that
were
easy
versus
the
things
that
were
more
challenging
with
sort
of
getting
started,
and
also
the
thanks
a
lot
to
tyler
for
setting
up
that
skeleton.
E
I
think
that
helped
a
lot
and
it
I
don't
know
if
you
guys
have
already
been
thinking
this,
but
it
might
be
nice
for
contributors
to
have
like
an
npx,
create
hyper
data
adapter,
because
I
really
think
that
just
having
that
boilerplate
there
like
prevented
me
from
having
to
figure
out
a
lot
of
stuff,
so
you
know
it
was.
It
was
fairly
easy.
Once
I
figured
out
how
to
actually
start
the
development
version
of
hyper
and
where
to
put
things
for
it
to
just
start
working,
so
that
was
that
was
very
big.
A
No,
that's
awesome
input,
one
simple
way
to
do
it
in
you
know,
code
generation
is
another
subject,
but
I
really
like
what
github
has
started
doing
with
their
templates.
So
so
that
might
be
a
lightweight
approach
is
to
just
create
a
repo
of
that
skeleton
and
make
it
a
a
template.
And
then,
when
you
create
your
new
repo,
you
can
just
inherit
from
that
template.
E
Right
and
tyler
mentioned
too,
which
is
a
good
point
that
a
lot
of
the
the
tests
are
basically
going
to
be
the
same
for
all
of
them.
So
you
know
that
could
be
part
of
that
template.
So.
A
Yeah,
yeah
and,
and
that
would
be
cool
if
the
tests
were
already
written
and
all
you
had
to
do
is
run
them.
C
C
Just
the
the
wrapping
code,
written
in
zod,
provides
a
lot
of
the
testing
for
the
inputs
and
the
outputs,
and
that
could
just
be
tests
in
and
of
itself
and
then
just
some
tests
around
each
ap
like
each
api
on
that
specific
port
and
then
that
test
suite
could
just
be
reused
for
every
single
adapter
for
that
port,
and
so
the
only
test
that
the
adapter
developer
would
have
to
write
or
just
things
like
branches
within
the
implementation.
C
A
E
So
I
hope
to
have
this
pushed
and
finished
david.
I
had
a
family
emergency
come
up
this
weekend,
so
I
have
not.
I
haven't
looked
at
this
closely
for
the
last
few
days.
I
did
test
all
of
the
routes
in
postman
this
morning
on
hyper
and
they
were
all
good,
and
I
noticed
one
thing
that
I
saw
is
a
in
the
hyper.
I
o
docs
the
create
an
index
request
body
didn't
work.
I
had
to
do
something
different.
E
So
basically,
this
is
what
what
it
had
for
creating
the
fields
for
the
index
and
it
didn't
work
for
me
when
it
was
nested
in
this
index
key.
I
had
to
just
put
it
in
its
own:
got
it
on
its
own
level,
so
that.
E
Typo
yeah,
I
think
that's
the
only
the
only
hyper
io
documentation
thing
that
I
ran
across,
so
the
hyper
I
o
stuff.
Once
I
got
the
once,
I
got
it
connected
into
hyper
was
very
easy,
worked
very
quickly,
but
figuring
out
how
to
connect
it
to
hyper
was
not
as
well
documented,
and
I
guess
for
probably
good
reason.
E
The
main
priority
is
probably
more
the
users
versus
the
contributors,
but
it
was
mainly
just
for
me
that
there
may
be
a
place
to
go
find
stuff,
but
I
was
it
was
mainly
just
sort
of
like
poking
around
a
trial
and
error
until
I
figured
out
and
the
way
that
I
ended
up
doing.
It
is
not
the
way
that
I
would
do
it
again,
which
is
I
basically
cloned
the
entire
hyper
63
and
in
editing
the
removed,
the
the
get
folder
and
I'm
editing
the
dev
image
and
just
pointing
it
somewhere.
E
I'm
assuming
that
there's
a
an
easier
way
to
do
that
to
just
grab
it
somewhere,
but.
A
Yeah,
no,
that's
that's
a
good
point,
but
there's
it's
not
a
lot
of
documentation
out
there.
I've
got
a.
A
Repo
that
I'll
share
in
the
chat-
but
this
is
a
repo
of
like
a
stand-alone
hyper,
where
you
can
just
install
the
the
core,
the
express
app
the
ports
and
the
adapters
and
then
create
a
config.
A
And
basically
you
you've
got
an
index.js
file
with
a
couple
of
lines
and
and
that's
pretty
much
it
but
definitely
need
to
document
that.
E
Let's
see
some
other
random
things,
so
the
hyper
config,
I
don't
care.
If
anyone
sees
these
secrets,
I
did
notice,
so
it's
using
the
the
the
env
like
process.ev,
but
it
dot
env
is
not
required
as
a
dependency,
and
I
saw
that
for
for
the
default,
it's
just
being
passed
in
the
command
line.
I
don't
know
that
that's
realistic
for
this.
I
don't
know
if
I
may
have
been
overthinking
this
or
if
there's
essentially.
A
E
Happy
to
I
may
have
been
asking
the
wrong
question
so,
okay,
if,
if
I,
if
I
have
like
a
a
dot
env
here,
which
I
I
do,
you're
not
using
any
external
library
to
to
grab
it
you're
just
saying
process,
process.dnv.whatever.
A
Right
so
there's
a
couple
ways:
I
use
the
the
dot
emv
module
yeah
and
you
can.
A
You
can
run
that
on
the
the
command
line
with
the
dash
r,
so
you
can
do
like
dash
r
dot,
emv,
slash
config,
and
then
it
would
load
your
emv
file
in
at
the
command
line
or
you
can
just
require
it
at
the
top
of
your
index.js,
like
it's
kind
of
the
first
thing
that
runs
and
you
just
do
like-
require
dot,
emv,
dot,
config
and
open
paren
close
print,
but
I
I
prefer
to
do
it
on
the
script.
A
E
A
E
Okay,
good
to
know
this
was
just
the
one
that
I
happened
to
land
on.
Let's
see
I
mean,
is
there
anything
in
particular
that
you'd
want
me
to
show.
E
The
only
sort
of
strange
thing
that
I
can
think
of
right
away
or
not
strange,
which
is
slightly
different,
is
for
dynamodb.
There
are
two
sort
of
main
clients
for
it.
E
A
Are
you
using
version
two
or
version
three
of
the
aws
sdk
version.
E
Three,
I
believe
that
I
didn't
specify-
and
I
saw
that
that
was
in
the
original
template-
that
tyler
had
made
okay,
so
I'll
go
ahead
and
add
that
before
doing
a
pull
request,
let's
see
a
lot
of
these
are
fairly
straightforward.
E
This
is
my
first
time
using
croc,
so
I
wouldn't
be
surprised
if
there
are
a
couple
of
awkward
usages,
but
for
the
most
part
the
documents
were
pretty
straightforward
for
it,
except
for
update
the
only
thing
for
update.
Is
you
have
to
generate
a
string
so
there's
just
a
helper
function
that
will
probably
eventually
be
in
its
own
file.
E
Let's
see
and
see
the
index,
I
don't
remember
the
specifics
here.
It's
been
a
few
days,
but
I
there
are
a
lot
of
like
a
lot
of
different
parameters
and
I'm
I'm
guessing
that
we're
just
going
to
use
a
relatively
small
subset
of
them,
but
I
mean
even
things
like
read
capacity
units
right
capacity,
units
that
you
know,
I'm
just
gonna
put
the
lowest
number
here.
I
doubt
that's
something
that
we'd
ever
want
in
the
in
the
interface,
but
I
don't
really
know
there
yeah.
No,
that's
cool.
E
See
yeah
and
then
these
these
three
have
some
have
some
like.
Basically
things
to
think
through
before
implementing.
E
If
I
remember
correctly,
the
main
one
that
sort
of
made
me
pause
was
for
list
documents,
there's
a
there's,
a
sort
feature,
and
because
dynamodb
has
that
sort
key
which
I'm
not
using
right
now
to
do
a
sort.
I'd
have
to
implement
the
sort
key
which
would
then
make
it
so
the
partition
key
alone
being
unique,
was
not
enough,
so
that
would
potentially
require
rethinking
how
everything
else
was
implemented
as
well.
Okay,
I
believe
that
was
the
the
main
pause
to
just
sort
of
move
on
to
something
else.
A
Yeah
I
mean
it
all,
looks
good,
very,
very
clean.
E
Cool
and
yeah,
I
have
to
write
some
tests
for
it.
One
thing
looking
at
the
tests
is
this:
the
first
adapter
that's
exclusively
like
a
vendor-specific
cloud
thing
like
this
runs
in
aws
that
there's
not
like
a
I
mean
you
can
run
it
locally,
but
I
don't
know
that
it's
it's
the
norm,
we're
doing
like.
A
Yeah
so
there
there's
a
library
called
mock
fetch
that
essentially
allows
you
to
capture
any.
A
C
Yeah
and
and
like
the
adapter
itself,
it
gets
past
a
or
I
guess,
the
the
adapter
factory.
It
gets
past
an
instance
of
like
dynamodb
instantiated
object
from
the
aws
sdk,
so
since
you're
using
that
you
could
just
mock
that
and
pass
it
as
the
dependency
and
so
and
you
could
just
yeah,
because
it's
just
passed
as
an
argument.
So
it's
just
good
old
dependency,
injection,
okay,.
E
Cool-
and
I
you
guys
have
some
references
for
for
some
other
ones,
so
I'll
probably
check
some
of
those
out.
C
Yeah,
I
think
I
think
most
of
them
are
using
like
fetch,
and
so
we've
used
mock
fetch
for
mocking
those
those
calls.
So
they
don't
actually
go
over
the
wire.
But
in
this
case
I'm
sure,
since
this
is
using
an
sdk
like
that,
is
specifically
why
we
pass
it
as
a
dependency
to
the
factory
so
that
it
can
be
mocked
and
as
part
and
just
make
testing
easier.
C
E
There's
anything
else
in
particular
to
show.
Unless
so
I
don't
know
if
this
is
the
case
for
some
of
the
other
ones.
Remember
correctly,
you
can
only
create
one
index
at
a
time,
a
weird
edge
case,
for
that
would
be
somebody
tried
to
create
multiple
indexes.
They
would
get
a
false,
even
though
they
passed
the
right
stuff.
D
D
C
At
that
I
mean
if
the
api,
if
the
hyper
api
supports
like
creating
multiple
indexes,
then
that
would
be
where
like
within
that
method,
we'd
have
to
have.
So
there
would
be
some
way
to,
if
possible,
like
get
around
that
limitation
for
dynamodb,
whether
that's
like
create
an
index
then
like
separately
and
then
create
the
next
index
as
opposed
to
trying
to
create
them
at
the
same
time.
C
That
would
be
sort
of
like
the
internal
logic
of
the
implementation
of
this
function.
If
I'm
understanding
what
your
limitation
here
like.
E
So
in
this
case,
I
mean
more
less
that
I
would
like
that
there
would
be
a
new
function
to
create
multiple
indexes
more
if
I
happen
to
hit
this
function
twice
like.
If
I
think
it's,
maybe
this
one
yeah
well,
maybe
not.
I
can't
remember
one
of
these.
One
of
these
is
oh.
This
is
the
thing
to
create
an
index.
If
I
click
this
twice
and
I
change
one
of
them
this.
E
This
is
actually
pretty
fast,
because
there's
basically
no
documents
and
very
simple,
but
if
I
hit
it
twice
in
succession,
I
would
probably
get
a
failure
on
the
second
one,
even
though
it
was
a
valid
api
call,
the
failure
being
because
the
index
already
exists,
because
it's
in
the
process
of
creating
one
and
it
will
just
fail
if
you
try
to
create
another
one
at
the
same
time,
I
don't
think
it's
going
to
be
a
particularly
common
error.
It's
just
something
that,
while
I
was
coming
across
it
probably
worth
mentioning.
B
That
is
an
interesting
you
could,
if
there's
a
way
to
check,
to
see
if
it's
still
processing
the
previous
call
or.
B
On
the
sequel
call
right
you
could
have,
you
could
send
back
some
sort
of
message
saying
too
many
requests.
I.
C
Right
and
there's
a
way
to
I
mean
there's
it's.
It
would
definitely
be
feasible
to
like
stack
the
promises,
like
kind
of
store
that
as
like,
I'm
just
like
spitballing
here
like
it's
like
local
state
in
the
adapter
like.
If
there's
like
an
index
being
created,
you
like
store
that
promise
and
then
train
any
additional
calls
to
create
index
onto
that
promise.
Train
and
so
they'll
just
go
sequentially.
C
But
then
that
doesn't
take
into
account
like
if
there's
multiple
instances
of
hyper
or
something
like.
D
E
Yeah,
I'm
fairly
sure
that
you'd
get
a
a
consistent
error
response
back
that
you
can
check
every
time.
So
I
don't
know.
If
that's
you
know,
if
that's
common
enough
to
be
worth
that
logic,
it
probably
is,
but
that
actually
does
remind
me
when
I
did
get
errors
that
I
don't
know
if
this
is
something
that
I
may
have
been
doing
the
response
incorrectly
or
if
this
is
just
what
was
to
be
expected,
I
don't
think
it
always
like.
E
I
would
get
okay
false,
but
I
would
not
necessarily
get
the
error
code
and
I
don't
know
if
that's
intentional,
so
that
you
don't
see
what's
going
on
how's
that
handled
like
what's
the
thought
process
there.
A
A
I
think
like
like
there's
a
couple
in
there
that
you
know
there
there's
there's
some
retry
logic
for
errors,
but
some
of
them.
You
know
some
things
shouldn't
retry
right.
If
it's
404
not
found
it
shouldn't
retry,
it
should
just
return.
404
yeah,
so
there's
some
some
quirks
in
there
that
just
need
focus
and
to
go
through
and
make
sure
everything's
consistent
and-
and
you
know,
returning
the
proper
status
code,
returning
the
proper
description
for
the
status
code,
etc.
E
And
I
think
I
think
there
are
a
handful
of
things
that
would
have
solutions
in
the
range
of
what
trip
and
tyler
both
talked
about,
which
is,
I,
I
think,
two
there's
something
with
like.
E
Was
it
something
with
like
creating
duplicates?
I
think
there's
a
lot
of
basically
things
to
you
know
you
might
have
to
have
logic
inside
the
adapter
to
to
check
for
duplicates
and
maybe
depend
on
the
the
individual
database.
I'm
assuming
that's.
I
don't
know
how
hyper
would
handle
that.
Otherwise,
actually
I
mean
that'd
be
that's
ideally
an
adapter
specific
implementation.
Would
it
not.
A
Yeah,
I
think
it's
you
know
just
defining
the
so
so
you've
got,
you
know
the
resolved
side
and
then
the
rejected
side.
The
rejected
side
from
the
adapter
can
pass
a
status,
a
message
and
an
okay
and
just
getting
that
standard
eye,
so
that
if
the
adapter
doesn't
pass
a
status
then-
and
it
is
an
error,
then
it
will
just
like
default
to
500.
A
You
know
make
the
adapter
be
explicit
about
the
status
code.
That's
one
thought
but
open
to
other
thoughts.
E
Okay,
I
guess
that
could
also
be
argued
to
be
better
as
a
as
a
business
logic
use
case
as
well.
What
do
you
want
to
do
if
someone
tries
to
create
a
duplicate,
because
maybe
two
people
using
dynamodb
adapter
would
have
a
different
idea
of
what
you'd
want
to
do?
In
that
case,
I
don't
know
yeah.
A
Yeah
I
mean
again,
you
normally
would
throw
a
409
document
conflict
for
the
last
person
and
let
that
person
you
know
that
client
deal
with
it,
but
you're
right
there
could
be
potentially
other
ways
to
go
about
it.
I
guess.
E
E
So
I
think
that
once
there's
some
like
you
know
either
like
a
simple
tutorial
or
some
like
easy
to
follow,
docs
on
like
where
to
basically
start
developing,
I
think
it'll
be
pretty
nice
because
it
was
definitely
cool
and
satisfying
to
just
you
know,
basically
just
made
the
functions
first,
based
on
the
schema
and
was
able
to
more
or
less
just
plug
them
into
hyper
after
that,
and
they
just
kind
of
started
magically
working,
which
was
nice
cool.
B
A
Have
any
updates
or
comments.
C
And
first
of
all,
just
casey
is
looking
great
so
far
and
like
the
questions
that
you're
asking
and
sort
of
the
outstanding
like
what
the
error
should
look
like
the
conversation
that
we
have
here
have
been
really
valuable.
So
thanks
for
that
and
the
work
so
far
I
mean
as
far
as
updates
for
me,
I
really
enjoyed
the
session
trip
tom
myself
had
last
week.
C
Well,
it
seemed
like
things
were
ramping
down
and
then
they
were
like
well
hold
on.
He
actually
is
leaving.
Let's
get
him
to
write
lots
of
code.
C
Yeah,
so
no
real
update
for
me,
though,
just
really
excited
to
get
to
solving
some
of
these
problems
and
acting
on
some
of
the
feedback
we've
been
getting
really
excited
for
it.
B
You
covered
it
with
the
movie
workshop
api
and
starting
design
on
the
front
end.
So
that's
what
we've
been
working
on
lately.
C
B
D
C
C
A
Yeah
yeah,
I'm
gonna,
like
I
said,
take
that
blog
post
as
a
guide,
but
you
know
again,
there's
there's
several
different
products
that
we're
like,
and
you
know
and
there'll,
probably
be
more
that
that
other
people
will
mention.
Oh,
it's
kind
of
like
this,
or
it's
kind
of
like
that,
and
I
think
the
easiest
conversation
to
have-
or
at
least
I
I
really
like
how
gleb
put
it
is
the
the
the
drills
versus
the
holes
analogy.
Have
you
guys
ever
heard
of
that?
A
Okay
good?
So
when
you
go
to
the
store
to
pick
out
a
drill,
you're,
not
thinking
this
brand
or
this
brand
you're
thinking
about
the
hole
you've
got
to
drill
right,
you're
thinking
about
the
problem
you're
trying
to
solve
so
his
recommendation
is,
you
know,
really
focus
on
the
problems
that
we're
trying
to
solve
and
talk
towards
that
and
again
making
apps
more
easy
to
test
easy
to
maintain
minimizing
unintentional
technical
depth.
Those
are
kind
of
the
holes
and
hyper
is
the
drill
for
that.
A
There
may
be
other
drills
out
there,
but
at
the
end
of
the
day,
that's
the
the
question.
You
know
that
that
I
am
kind
of
I
feel
like
this.
This
solution
solves
right
is
the
you
know,
keeping
business
logic
separate
from
implementation.
Details
in
the
day
results
in
very
highly
testable
things.
A
When
you
use
hyper
to
be
able
to
solve
problems
without
digging
through
postgres
documentation
or
digging
through
amazon
db
documentation,
it's
just
hyper
documentation
and-
and
you
don't
really
care
what
the
in
data
point
is
you're
concerned
about
solving
your
problem.
So
I
think
dapper
is
kind
of
this
mesh
event,
bus
system
and-
and
it
does
do
some
things
similar
to
us,
but
but
I
don't
think
it
shares
our
same
mission.
I
think
its
mission
is
more
of
an
integration
mission
of
you
can
write
code
in
any
language.
A
You
want
and
glue
it
all
together
through
this
microservice
architecture,
no
matter
what
language
you
use
or
what
service
you
use,
but
I'll
I'll
dig
into
more
of
dapper.
It's
definitely
a
cool
project
and
I
think
there's
a
lot
of
value
for
for
certain
use
cases
for
it,
but
but
I
do
think
we
have
kind
of
two
different
missions
if,
if
that
makes
sense,.
C
No,
I
think
it
does.
I
mean
again,
I
haven't
used
dapper,
but
and
given
that
we're
such
a
small
group,
we're
kind
of
like
people
are
going
to
be
asking
us
about
hyper,
we're
kind
of
all
in
sales
in
a
sense
and
so
yeah
just
being
able
to
like,
like
you
said,
there's
people
that
do
some
of
what
we
do
being
able
to
differentiate
those
and
know
the
distinction
like
and
it's
clear
for
for
all
of
us,
so
that
when
it
comes
up
in
a
conversation,
would
you
be
like
oh
well?
A
C
A
That's
right,
that's
right!
No,
and-
and
you
could
even
take
it
to
you
know,
the
drill
has
bits
and
hyper
has
adapters.
Hyper
may
not
have
a
zillion
adapters
yet,
but
the
the
interchangeability
is
frictionless
right
to
change
out
one
adapter
to
another
adapter
or
one
bit
to
another
bit.
A
Well,
cool
beans
thanks
everyone
for
all
the
work.
I
am
working
on
the
swag
shop
to
get
the
new
graphics
up
and-
and
I
should
have
that
by
the
end
of
this
week-
so
we
should
have
you
know
new
new
swag,
with
the
new
logos
and
everything
all
hyper.
So
I
think
that'll
be
cool
and
that's
all
I
have.
I
hope
everyone
has
a
great
week
and
we'll
you
know,
touch
base
through
slack
and
catch
up
next
time.