►
From YouTube: [Online Meetup] Kong Spec Expose Plugin
Description
During this call, Jeremy Justus demonstrated the Kong Spec Expose Plugin built by Optum. We talked about the announcements at Kong Summit, and Thibault discussed his take on the changes in 1.0. We also had a detailed discussion of plugin debugging tools available in Kong, and herd from some of our most vocal community members.
Resources:
Spec Expose Plugin from Optum - https://docs.konghq.com/hub/optum/kong-spec-expose/
Kong integrations Hub - https://docs.konghq.com/hub/
Path prefix plugin - https://github.com/alexashley/kong-plugin-path-prefix
Join our next Online Meetup: https://konghq.com/online-meetups/
A
So
welcome
everybody.
Thank
you
for
joining
us.
This
is
the
first
ever
Kahn
community
call
we're
pretty
stoked
about
it.
We're
gonna
have
these
things
once
a
month
for
a
while
they're
gonna
move
around
in
weeks,
because
we
wanted
them
to
be
earlier
in
the
month,
but
they'll
probably
settle
around
the
second
week
on
the
goal,
for
these
calls
is
to
provide
a
place
for
community
members
to
present
their
work
with
calm,
have
a
place
to
talk
about.
A
The
community
have
a
place
to
talk
about
new
developments
with
the
technology,
and
so,
if
you
have
ideas
about
what
you
want
to
talk
about
on
this
call,
we
hope
that
will
be
driven
by
you.
So
agenda
doc
is
open.
I've
put
it
in
the
chat
and
I'll
chat
it
back
in
there
one
more
time.
If
you
could
all
do
me
a
favor
and
just
put
your
names
in
the
agenda
doc
so
that
we
knew
who
came
to
the
meeting.
That
would
be
super
helpful.
A
A
C
So
right,
well,
terma
gets
that
sorted
like
cool
we
so
Optimas
a
company
which,
among
other
things,
provides
IT
services
to
the
healthcare
industry,
we're
part
of
United
Health
Group,
which
is
fifth
on
the
fortune
500
list
or
a
pretty
large
company
and
as
part
of
an
overall
effort
to
modernize.
Our
back
we've
been
evaluating
calling
for
about
the
past
year.
C
During
this
time,
we've
created
several
small,
relatively
simple
plugins
that
fit
our
specific
needs
in
the
use
cases
and
we've
open-sourced
I
think
eight
of
them
that
we
think
the
community
should
be
able
to
benefit
from
as
well,
and
one
of
these
plugins
is
called
Kong
spec
expose
and
the
idea
behind
this
plug-in
is
this.
Plugin
allows
Stargate
well
allows
Kong
to
expose
an
API
interface
to
a
customers.
C
This
is
like
a
swagger,
an
open,
API
spec
and
we
still
a
waddle
just
some
type
of
API
interface,
and
let
me
back
up
a
minute
and
unpack
the
problem
here.
So
right
now,
if
I'm,
an
API
provider
and
I
have
an
open,
API,
spec
I
basically
have
two
avenues
for
how
I
could
possibly
expose
this
to
one
of
my
consumers.
The
first
is:
if
we
just
expose
this
as
part
of
the
proxy,
then
the
consumer
does
what
they
need
to
do
to
register
for
off
and.
D
C
So
the
point
of
Kong
spec
expose
is
to
give
a
standard
end
point
for
each
Kong
proxy
that
we
publish
that
will
allow
a
consumer
to
gain
access
to
an
open,
API
spec
with
wins,
Dola
Waddell,
or
what
have
you
and
you
can
sort
of
see
the
comparison
on
the
slide?
Where
were
the
standardization
that
we
would
do
is
we
would
expose
that
endpoint
on
spec
Z
for
any
given
proxy
and
I'll
turn
it
over
to
Jeremy
and
he's
going
to
give
you
a
little
bit
of
a
live
demo
of
what
this
looks
like
sure,.
B
Yeah,
so
first
I
want
to
show
you
a
given
service
endpoint
right
here,
you
can
see
in
my
browser
the
Gateway
endpoint,
with
API
pet
store
v1.
You
can
see
when
I'm
trying
to
hit
the
service
directly
in
browser
with
just
an
HTTP
GET.
We
an
error
code
access
token
is
missing.
That
is
because
this
proxy
is
secured
with
a
lot
to
client
credential
pattern.
So
if
I
were
to
go
to
the
slash
spec
Z
path,
added
to
the
URI,
you
can
see
a
JSON
document
is
returned.
B
This
JSON
document
is
the
spec
that
contains
information
about
what
the
request
the
API
expects,
as
well
as
the
response.
The
API
would
return
to
a
given
consumer
and
if
you
wanted
to
browse
this
programmatic
spec
here
in
a
human,
readable
format,
you
can
use
something
like
swagger
inspector
here
to
evaluate
the
spec
Z
path.
And
then
you
can
see
here
on
the
user
interface.
B
We
see
the
different
methods
that
this
API
exposes
and
the
different
resources
you
can
manipulate
from
HTTP
kids
to
HTTP
posts
to
deletes,
and
that's
essentially
how
the
spec
Z
plug-in
provides
values
to
our
API
providers,
as
well
as
our
consumers
and
I
think.
We
also
want
to
talk
a
little
bit
about
what
it's
been
like
to
integrate
with
calling
and
develop
plugins
as
well.
C
Yeah,
so
so,
when
we've
done
I
think
we've
developed
eleven
plugins,
eight
of
which
we've
made
open-source.
The
three
we
haven't
are
just
there's
a
real
possibility
that
anybody
in
the
community
could
benefit
from
this
type
of
logic.
It
really
is
specific
to
our
company,
but
we've
run
into
some
positives
and
negatives
during
this
process
and
I'll
talk
about
a
couple
of
them,
so
probably
the
biggest
positive
and
something
that
I'm
afraid
might
get
overlooked.
Kind
of
often
is
the
general
structure
of
a
comm
plugin.
C
You
never
touched
the
language
so
just
having
that
structure
and
having
that
standardization
and
being
able
to
work
with
that
has
really
helped
us
to
expedite
the
efficiency
of
our
development,
something
that
I'll
just
throw
in
here
is
a
con
that
that
we've
definitely
run
into
is
the
relative
lack
of
any
type
of
robust
debugging
framework
for
when
we're
actually
doing
these
development.
So
what
we've
typically
done
for
debugging
is
we're
just
using
nginx
Eric
stateman
to
print
debugging
information
to
standard
out
and
kind
of
going
about
it.
C
That
way,
and
there's
there's
nothing
wrong
with
that
really
I
mean
that's
more
or
less
the
first
way
that
anybody
ever
learns
to
debug
a
thing,
but
compared
with
some
of
the
tools
and
some
of
the
resources
that
are
available
for
other
languages.
It
does.
It
is
a
little
bit
lacking
and
it
does
slow
down
the
process
a
little
bit.
This
is
compounded
by
the
fact
that
we're
often
working
with,
like
very
specific
nginx
and
Kong
I,
guess
variables
that
we
couldn't
reproduce
in
a
more
standardized
lua
testbench.
C
So
we
end
up
having
to
build
and
redeploy
our
Kong
every
time
we
want
to
test
a
new
iteration
of
one
of
these
plugins,
which
isn't
so
it's
not
a
big
problem
for
us,
we're
deployed
in
a
cloud
doc,
rised
environment,
so
it
doesn't
take
up
too
much
time.
It's
just
a
little
bit
slower
than
it
might
be,
with
other
platforms
and
other
languages
and
I
think
that
there
might
be
room
for
improvement
there.
But
to
kind
of
finish
with
a
positive
working
with
the
Lua
package.
C
Manager'
Lua
rocks
has
been
also
definitely
a
great
part
of
developing
these
plugins,
just
the
ease
of
installation.
The
ease
of
uploading
Arak.
How
uncomplicated
the
whole
process
is,
how,
when
you
have
a
specific
column,
plug-in
Lua
rocks,
is
smart
enough
to
install
that
to
the
correct
directory
such
that
that
information
is
accessible
to
Kong.
These
are
all
definite
positives
that
we've
seen
during
a
development
cycle
and
I'm
happy
that
you
does
have
that
type
of
relationship
with
with
a
package
manager.
So
that's
just
at
a
very
high
level.
C
B
Yeah,
just
to
add
to
that
I
think
also
plugin
development
has
been
very
smooth
with
calling
simply
because
of
calling
the
company
itself
all
their
engineers
are
very
helpful.
You
can
go
to
the
community
forum
and
you
know
post
questions
ideas
you
have
for
a
plug-in
and
within
you
know
the
same
day
you
usually
get
a
response
with
an
idea
on
how
to
build
your
plugin
and
how
to
integrate
that
and
which
say,
nginx
phage
you
would
use
and
the
variables
you'd
use
to
access
it.
B
C
It
was
always
a
struggle
to
get
responses
when
you
did,
they
were
generic
and
canned,
but
work
at
being
able
to
just
go
on
the
Kong
forum
and
have
a
detailed
discussion
about
issues
with
the
actual
Kong
developers
and
the
level
of
support
we've
seen
from
the
community
is
so
wildly
superior
to
the
type
of
relationship
and
the
type
of
support
and
the
results
that
you
will
see
from
the
bender
integration.
That
I
think
this.
C
This
old
idea
that,
in
order
for
a
platform
to
be
really
stable
and
and
to
be
proper
for
an
enterprise
solution
that
it
needs
vendor
support,
is
totally
wrong.
Headed
these
days,
and
my
experience
with
calling
is
basically
why
I
think
that
now,
but
so
Jeremy,
if
you
didn't,
have
anything
to
add
to
that
I'd
love
to
open
it
up
for
questions.
A
D
D
A
B
Right,
yeah,
open
sourcing,
our
plugins,
our
process
is
a
little
bit
clunky.
It
involves
using
a
couple
of
bad
applications
that
are
old
for
documenting
out
the
plugin,
we're
trying
to
open
source
and
basically
getting
approval
from
a
person
to
kind
of
forms
up
our
our
legal
documents
that
we
put
into
the
repos
as
well.
C
General
I
like
to
think
that
this
company
is
shifting
more
especially
in
a
leadership
level,
towards
an
open-source
oriented
idea
of
doing
things.
We
are
a
large
organization
in
total,
there's
more
than
200,000
employees
that
work
in
this
company,
so
not
Optimus,
Pacific
Lee,
but
the
parent
company
in
general.
So
there's
there's
a
lot
of
minds
to
change,
but
I.
Think
management,
especially
jeremy
and
I's
direct
management,
has
been
supportive
in
this
effort.
It's
really
not
too
bad.
C
So
Jeff
and
Roger
are
are
in
fact
Jeremiah
is
direct
management,
so,
as
you
can
see,
they
are
on
this
call
like
this.
Is
you
know
this
is
a
change
that
I'd
like
to
see
that
I
think
is
more
or
less
industrywide,
as
open-source
becomes
more
popular
as
the
results
of
the
process
become
more
proven.
I
think
you'll
see
that
more
large
organizations
will
start
to
see
value
in
this
and
will
start
to
invest
in
it.
I
think
Optim
is
one
of
the
first
to
really
do
so
at
this
level.
E
That
makes
us
that
mix
it's
really
happy.
We
are
very,
very
happy
that
you're
moving
forward
with
that
with
open
source.
Thank
you
for
us
and
Jeremy
I.
Think
that
was
that
was
a
great
demo.
I,
don't
have
a
question
for
you,
so
you've
open
sourced
your
plugins
and
you
open
sourced
a
number
of
those
I
think
a
plan
regarding
the
maintenance
of
those
plugins.
Do
you
have
new
features?
No
any
roadmap
items
that
you'd
like
to
address
or
have
you
received.
C
So,
every
time
that
Jeremy
and
I
encounter
a
problem
or
have
a
bug
reported
to
us,
we
we
do
make
sure
to
change
the
open-source
repo
we
change
the
github
we
change.
What's
on
the
package
manager,
keeping
this
up-to-date
is
important
to
us,
not
just
for
the
support
of
the
community,
although
that's
obviously
concerned,
but
also
because
we
build
directly
from
Lua
rocks
like
that
is
how
we
deploy
our
services.
So
in
order
for
them
for
our
product
to
be
effective,
we
need
to
keep
the
open
source
versions
of
this
stuff,
up-to-date
and
and
I.
C
Think
that
is
the
way
to
do
it.
It's
definitely
not
correct
to
have
a
forked
private
company
repo,
where
the
latest
and
greatest
limbs,
and
then
everything
you
know,
that's
just
not
that
defeats
the
point
of
this
whole
process
and
I
think
the
community
does
have
a
lot
to
add.
With
respect
to
the
quality
of
these
things,
great.
A
C
Mean
the
way
it
works
basically
is
we
will
take
the
specific
URL
that
exposes
the
speck
in
the
schemer.
I
mean
that's
the
way
that
the
Kong
spec
expose
plug-in
works.
So
it's
not
using
any
kind
of
it's
not
smart
enough
to
be
able
to
examine
all
possible
end
points
on
the
API
on
the
upstream
to
figure
out,
which
is
the
spec
I'm,
not
sure.
If
I'm
understanding
the
question
right.
E
B
But
as
far
as
what
you
can
configure,
the
plug-in
on
whether
you
configure
it
on
a
route
or
a
service
would
be
up
to
you.
The
plug-in
could
be
configured
globally
for
that
matter.
If
your
gateway
is
only
fronting
one
API
how
you,
how
you
enable
the
plug-in
on
top
of
the
Kong
resources
is
up
to
you.
They
they
provide
that
flexibility.
C
And
you
know
this
is
this:
is
a
this
is
a
simple
plugin
right.
Most
of
what
we've
we
have
open
sourced
is
relatively
simple.
If
only
because
the
problems
we've
encountered
admit
of
simple
solutions,
so
I
mean
you
know,
feel
free
to
take
a
look
at
the
the
source
code
of
it,
and
if
you
have
a
question
it
supposed
to
github
issue
and
we'll
be
sure
to
respond,
but
but.
E
Yeah,
absolutely
when
I
realized
that
I
was
thinking
you
were
going
to
mention
as
part
of
the
comms
or
liberal
pin.
Dragon
was
you
know
you,
you
could
say
the
lack
of
a
test
framework
or
the
lack
of
a
you
know
in
externalize
test
framework
that
is
independent
from
Kong
that
you
know
you
can
install
independently
and
then
assume
that
comes
installed
on
your
system
and
test
the
plug-in
on
the
contest.
Instance.
I
have
some
sort
of
utility
methods.
So
how
did
how
did
you
test
the
plug-in
and
then
we
can
elaborate
into
our
plans?
C
So
I
mean
basically
what
we
did
and
I
kind
of
covered
a
little
bit
of
this,
but
what
we've
done
effectively
is
for
debugging.
We
use
entry,
NextEra
statements
to
log
debugging
info
to
standard
out
and
we
basically
just
build
and
redeploy
every
time
we
need
to
test
on
deployment
unlike
go
live.
There
isn't
necessarily
like
what
would
be
the
equivalent
of
a
J
unit
which
I
would
really
like
to
have
so
so
that's
something
that
stands
for
improvement,
but
yeah
testing
in
the
most
basic
sense.
We
have
a
development
environment.
C
E
C
E
E
So
what
so,
if
there
are
two
ways
to
approach
this
today,
if
you
are
using
kong
and
having
star
kong
locally,
you
have
access
in
your
luma
doors
to
this
peg
helpers,
module,
which
is
a
motor
that
allows
you
to
stop
and
calm
the
test
instance,
and
you
can
also
specify
your
own
like
custom
plugins.
So
you
can
inject
and
and
declare
to
the
test
instance
that
you
have
a
test
begin
to
run
and
you
give
it
the
name
of
the
plug-in
and
the
test
instance
with
load.
E
We
load
your
your
your
custom,
plugin,
then
still
in
the
setup
phase,
you
can
make
a
few
admin
API
course
yeah
helper
functions
as
well.
That
would
configure
count
a
certain
way
say
the
raw
at
a
service,
a
consumer
configure
the
plug-in
on
the
consumer,
and
then
you
enter
the
test
cases.
You
can
have
n
test
cases
and
each
of
them
is
it's
a
very
kind
of
declarative
test
framework
in
which
you
describe
what
is
the
behavior?
You
were
expecting
to
see
in
the
test.
E
You
implement
a
function
that
will
run
and
will
execute
the
test,
and
the
test
would
be
make
the
request.
The
add,
a
proxy
port
instead
of
the
API
right
and
test
that
the
status
code
and
the
buddy
match
whatever
you're
expecting
and
that
is
that
is
doable
with
the
spec
helpers
module
if
Kong
is
installed
locally
on
the
development
machine.
That.
F
E
The
VAT
grantee
majors
worse,
we
want
to
develop
against
the
background.
Image
is
a
very
easy
way
to
get
started
as
well
with
all
the
dependencies
included.
The
other
possibility
that
we've
also
been
exploring
more
recently
when
we
all
serve
start
developing
plugins
that
we
develop
externally
as
part
of
another
git
repository
is
a
more
full
end-to-end
integration
approach,
kind
of
where
the
functions
that
are
used
to
start
and
stop
Kong
assume
that
Congress
in
style,
but
it
doesn't
use
the
spec
helper
spring
and
that
way
we
are
able
to
start
a
coincidence.
E
Yeah
commands
as
if
we
were
a
regular
user
and
then
so
real
early,
our
command
lines
from
STS.
We
spawn
shell
from
the
test
script.
That
starts
come
with
the
curation
things
toward
the
environment,
variables,
settings,
etc,
and
then
we
configure
it
via
HTTP.
It
was
HTTP
or
in
the
other
way,
and
we're
also
Bridal
tests
on
cases
after
that,
so
we
kind
of
follow
both
approaches.
Of
course,
in
conch
itself
we
use
the
spec
helpers
module
in
the
long
term.
We
do
have
the
vision
of
a
test
framework
that
would
be
indefinite.
E
Of
course,
I.
Would
assume
Kong
is
installed,
but
wouldn't
use
the
spec
helpers
utility
of
commit
surf,
which
we
reserve
the
right
to
break
anytime
on
with
a
block
counter.
So
for
a
native
in
isolated,
more
independent
test
framework,
the
idea
would
be
utility
functions
to
do
all
these
things
that
do
not
break
and
support
all
test
cases,
not
the
ones
that
we
encounter
here
when
we
leave
that
combo
also
the
ones
that
community
encounters
when
community
develops
custom
plugins
and
regarding
a
yeah
regarding
on
debugging
statements.
E
B
D
E
E
But
this
one
is
included
in
count
so
I'm,
so
I'm
not
referring
to
the
inspect
module
that
you
can
install
the
Lua
rocks,
but
really
of
one
that
is
included
in
come
with
the
plug-in
development
kit.
That's
what
I
was
referring
to,
which
allows
you
to
calm
the
global,
lock
inspect
and
it's
what
you
don't
have
to
install
install
inspect
and
you
don't
have
to
require
and.
A
A
F
A
F
A
Alex
says:
I:
don't
have
a
mic
on
a
mic.
Unfortunately,
I
work
for
a
retail
in
the
Midwest,
where
we've
been
using
Kong
since
January,
really
enjoying
working
with
Kong
and
developing
some
plugins.
Oh,
that's
really
really
cool.
Have
you
by
any
chance
used
the
PDK
yet
or
have
you
been
developing
all
of
those
plugins
prior
to
the
P
DK.
E
D
Marcelo,
the
crews
and
with
McAfee
as
I
thought
on
the
notes,
I
think
I
was
I.
Wasn't
it's
I
mean
last
week
been
using
Khan
for
three
years.
First
lady,
we
developed
against
version
C
level.
7
I
do
remember
the
spec
helpers
by
the
way.
That's
awesome.
That
was
a
lot
of
help
for
for
unit
testing
yeah.
How
did
you
hear
awesome.
C
A
A
E
A
G
G
G
E
E
A
D
E
Days
where
Khan
came
from
misshape,
the
API
marketplace
was
the
proxy
that
we
use
to
developed
or
API
marketplace.
It
was
the
underlying
proxy
making
the
request
from
API
consumers
to
it
providers,
and
then
we
would
do
all
of
the
monitoring
and
all
the
the
the
load
balancing
and
everything
for
them.
So
we
exercised
called
I
made
it
open
source,
and
today,
my
role
here
at
Kong
is
principal
engineer
and
I
maintain
open
source
with
repository
and
product
with
the
help
of
the
entire
team.
E
E
A
D
D
A
F
Even
I'm
really
looking
right
from
the
beginning
days
of
Kong
and
it
looks
like
freely
it
like.
We
move
into
our
slips,
how
lik
I,
remember
it
creating
my
first
plugin
and
now
like
we
have
STK
and
all
those
things.
So
we
still
mix
makes
me
happy
like
well
going
down
same
time
like
I.
Also
wanna,
add
like
all
these
community.
Plugin
really
is
trending
like
a
Kong
as
a
big
platform,
but
same
time.
F
We
also
need
a
setup
for
a
sick
like
like
more
of
a
scanning
thing
like
we
need
to
be
more
responsible
while
using
those
plugins,
basically
right,
they
are
treated
as
a
first-class,
like
citizen,
in
the
concert.
So
probably
we
will
walk
towards
a
system
where
we
can
scan
all
those
plugins
for
vulnerabilities
and
all
those
things
make
sense.
A
E
A
The
next
thing
on
our
agenda
is
actually
a
con
summit
recap,
so
I
just
thought
that
we
would
share
I,
hear
all
over
Jeremy's
slides
in
case
they
work
and
work.
Well,
so
we
had
our
first
in-person
conference,
which
was
Kong
summit.
It
was
on
September,
18th
and
19th
in
San
Francisco.
It
was
really
really
great.
We
had
more
than
200
people
come,
which
was.
It
was
a
pretty
packed
event,
and
so
I
was
really
glad
to
hear
about
that.
E
E
So
we
want
to
make
calm
future
proof
and
when
we're
talking
about
1-0,
we
were
talking
about
the
the
stability
and
how
it
supported
major
organizations
in
the
world
over
the
past
three
years,
but
also
how
is
it
going
to
evolve?
And
we
want
to
maintain
that
stability.
We
want
to
maintain
the
the
notion
that
we
don't
want
to
make.
We
want
to
keep
improving
home,
but
we
don't
want
to,
but
if
we
want
to
consolidate
ecosystem
that
we've
acquired
so
far,.
A
And
so
we'll
give
you
a
question
we'll
give
you
a
second
to
talk
about
one
dot
on
a
SKU
questions
about
it
later,
but
the
other
things
that
happened
at
summit
Cooper
mentioned
one
of
them,
which
is
the
launch
of
the
integrations
hub.
So
if
you
have
anything
that
you
want
to
get
writer
visibility
from
the
community-
and
you
should
definitely
make
a
pull
request
to
that,
we-
the.
A
E
So
we
developed
an
or
vision
of
what
we
call
the
service
control
platform,
which
is
a
way
to
an
umbrella
that
covers
API
usage
and
micro
service,
newsagent
and
monitoring
and
deployment
and
and
the
routing
kind
of
like
as
an
umbrella
in
your
company
right.
So
we
have
the
vision
of
controlling
in
multiple
places,
so,
whether
that
is
in
the
notes
of
traffic
and
also
in
east-west,
and
for
that
we
announced
also
service
mesh.
E
Those
were
immunity
and
brain.
Yes,
so
brain
is
a
way
to
Auto
configure
count
based
on
the
swagger
generated
specifications.
An
immunity
is
what
we
call
anomaly:
detection
to
detect
an
animal.
An
animal
is
in
your
traffic,
whether
that
is
potentially
malicious
users
or
msconfig
your
client
being
able
to
detect
with
confidence
that
a
parameter
of
the
wrong
type.
A
All
the
videos
are
coming
out
stay
tuned,
but
the
other
thing
that
happened
is
we
heard
from
a
lot
of
amazing
user
stories
at
the
summit.
My
absolute
favorite
thing
was
that
Zillow
open
source,
their
terraform
module
for
deploying
and
Konya
to
AWS,
which
I
thought
was
really
really
cool
and
they
gave
a
great
talk
that
just
went
from
like
bare-bones
I
know
nothing
about
terraform
to
now.
A
A
E
Just
wanted
to
elaborate
on
the
whole,
the
simple
dissent
was
organized,
so
we
had
two
trucks.
We
had
an
industry
track.
The
industry
track
was
where
you
know
what
happened
and
where
some
of
our
enterprise
customers
came
on
stage
and
elaborated
in
their
use
cases
or
some
of
the
you
know,
founders
or
you
know,
visionaries
in
the
micro
services,
ecosystem
came
and
stages
were
and
elaborated
their
vision
of
what
micro
services
are
and
what
programs
they
solve
and
what
programs
they
do
not
solve.
E
On
actual
you
know,
production
use
cases,
maybe
more
of
like
tricks
and
tips
and
how
to
best
use
Kong
how
to
look
technical
demos
of
how
combs
of
their
problems
etc,
and
we
also
had
the
workshops
at
the
end
where
we
learn
how
to
develop
plugin,
we
learned
how
to
use
the
open
data,
connector,
plug-in,
etc,
yeah
and
so
Marcelo.
If
you
want
to,
if
you
have
any
thought
like
what
were
your
favorite
moments
during
the
assuming
except
being
on
stage,
of
course,.
D
So
well,
I
I
think
the
usual
truck
was
amazing.
I
love
the
workshops.
I
think
they
were.
You
know
quietly
I
love
having
everything
you
know
being
hands-on
and
just
you
know
dropping
in
and
just
being
quick.
You
know
to
code
and
and
I
think
that's
that's!
You
know.
We
mentioned
this
during
the
the
the
workshops,
but
I
think
it's
the
the
best
way
to
put
for
people
to
their
price.
D
Having
hands-on,
you
know
workshop
type
of
training
getting
people,
you
know
to
drive
that
that's
I,
think
that's
amazing
and
on
the
only
mic,
surgeries
in
the
industry
track
gonna
be
I.
Think
something
was
interesting
for
me.
It
seemed
people
saying
hey.
We
went
back
to
the
monolith
right
and
just
you
know,
I
think
it's
hard
for
people
to
understand
and
micro-services
it's
not
for
everybody,
it's
not
for
everybody
in
the
sense
that
it
takes
a
while
to
make
the
transition
it's
an
organization
and
change.
D
E
Think
that
was
actually
one
of
my
favorite
takeaway
is
know
my
favorite
parts
but
takeaways
from
the
same
it
was
the
the
the
semi
was
an
or
or
as
as
an
organization
is
not
just
to
talk
about
micro
services.
It
is
to
prepare
for
micro
services
or
set
expectations
right,
and
that
was
really
valuable.
I
think
I
think
I
learned
a
lot,
so
everything
is
in
the
hashtag
on
Twitter
come
see
me
I'm.
She
only
see
pictures
or
if
you
want
to
have
some
insight,
see
some
quotes
or
notes
and
everything
and.
A
A
So,
as
we
talked
about
here
is
just
a
brief,
rundown
I
think
we
hit
most
of
these,
except
for
the
little
HTTP
health
checks
and
main
routes.
This
is
a
quick
rundown
of
the
things
that
are
new
with
one
dotto.
So
if
you
have
questions
about
what
any
of
these
bullet
points
mean,
that
would
be
service
mash,
the
new
database
abstraction
object,
removal
of
deprecated,
and
it
is
HTTP
health
checks
and
named
routes.
C
Yeah,
do
you
sort
of
have
one
simple
question
about
about
the
named
routes,
so
this
is
the
first
time
I'm,
seeing
this
I
haven't,
read
patch
notes
or
changelog
or
anything
just
yet.
So
if
this
is
explained
somewhere,
I
apologize,
but
our
the
way
that
we
have
things
structured
adopt
them.
Basically,
our
names
required
now
as
part
of
this,
for
that
no.
C
E
No
they're
not
the
to
elaborate
a
little
bit,
so
everybody
has
a
bit
more
background
context
when
we
initially
redesign
our
routing
model
to
have
rots
and
services.
The
idea
was
that
services
represent
an
existing
concept
where
you
do
have
my
services
in
your
organization
and
you
need
a
place
to
inject
them:
income,
that's
the
service
entity
and
then
the
rot,
we're
more
of
a
second-class
citizen
where
there
are
just
what
maps
to
a
given
service
based
on
rules
from
the
request,
but
they
don't
really
deserve
a
name
they're
expendable,
and
there
are
many
right.
E
E
Stack
name
is
a
human
concept
right,
so
it's
very
easy
for
a
human
to
interact
with
the
I
mean
API
in
which
we
were
given
raw
if
the
human
remembers
the
name,
but
it
is
also
due
to
the
fact
that
we
added
support
for
food
requests
and
a
way
to
identically.
Configure
congee
I
mean
API,
and
so,
with
the
previous
model,
you
were
required
to
generate
a
UID,
and
you
had
to
make
sure
that
you
know
your
purity
level.
Library
was
your
parent.
E
He
was
seated
and
everything
it
was
very
tricky
to
get
it
right
in
a
row,
generate
proper
UID
and
upload
a
route
even
I
didn't
Lee,
and
so
then,
when
that
rod
was
created,
if
you
were
to
try
to
recreate
it,
the
same
UID
not
necessarily
be
generated
so
you'll
be
creating
multiple
rods
so
effectively
the
the
poor
Walker
was
working
to
her.
So
that
is
why
we
added
names
or
routes
so
you're
able
to
the
resource
identifier
would
be
wrong.
E
A
D
A
A
E
Yeah,
so
yeah
I
think
it's
a
good.
It's
a
good
yeah
you're
right.
It's
it's
important
to
mention
it
in
wonder
that
we
are
removing
everything
that
we
consider
duplicated
and
that
we
have
been
maintaining
so
far.
That
includes
the
API
entity
that
includes
for
custom
plugins.
The
way
schemas
are
defined
and
some
of
the
Global's
that
are
accessible
by
plugins
so
know
before
we
go
further
and
everybody
scared.
E
Those
changes
are
mostly
syntax
changes
and
therefore
the
best
in
the
sense
that,
for
example,
so
first
of
all
you
should,
if
you're
playin
does
not
make
any
use
of
the
database
access
layer,
you
do
not
need
it
must.
It
must
actually
do
not
need
any
change,
but
if
your
plugin
exposes
a
admin,
API
endpoints
or
your
plugin
has
custom
entities
in
the
database
that
it
stores
you
are
likely
to
need
to
implement
some
changes.
E
So
the
way
the
schemas
are
defining
the
new
model
is
much
more
declarative
and
I
would
like
to
say
possible
for
front
ends,
so
we
we
also
have
a
better
error
framework.
We
have
more
validators
that
you
can
specify
for
if
match
is
not
match
it
satis
feared
and
that
offset
must
be
filled
to
one
off,
etc,
etc.
The
the
the
framework,
the
new
schema
framework,
is
very
powerful
and
almost
entirely
declarative.
E
We
want
to
rely
as
less
as
possible
on
the
self
check
function
that
some
of
you
may
have
seen
contributing
or
writing
your
own
plugins.
The
subject
function
was
kind
of
a
catch-all
for
all
validations
right,
so
you
just
wanted
some
validation
rules.
You
know
on
one
or
M
fields
and
you
just
implement
it
in
your
function.
E
The
problem
with
that
is
that
a
front-end
such
as
such
as
conga
could
not
no
know
what
were
those
rules
because
they're
implemented
inside
of
a
function,
so
there
is
no
way
for
the
slash
ischemic
or
to
retrieve
that
schema
and
build
the
UI,
a
form
that
was
powerful
enough.
So
that
is
one
one
reason
behind
is
change.
The
other
change
is
the
new
database
access
object
itself,
so
the
object
that
makes
queries
the
database.
You
know
that
we've
validated
the
schema
make
the
query
to
the
database.
E
What
was
missing
there
on
what
we've
improved
sorry
is
a
more
modularized
database
access,
object
and
instruction
object,
and
what
that
means
is
that
we've
refined
for
a
way
like
we
factor
some
of
the
internals.
We
fact
all
the
TTL
handling
for
Postgres
for
all
suit
against
which
is
much
more
performant.
Now
we
have
greatly
improved
the
modular
motorisation
around
the
Cassandra
all
pathways
and
we're
able
to
make
much
better
improvements
in
the
future
and
for
new,
eventually
I,
don't
want
to
say,
but
eventually
support
you
that
it
is
much
easier.