►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
C
C
B
C
Is
very
true,
it's
very
true,
but
there's
a
certain
something.
That's
not
here
when
I'm,
not
here
I
can't
get
to
like
we
don't
get
to
hang
it
on
each
other,
like
we
usually
did,
but
that's
okay,
we'll
figure
it
out,
we'll
figure
it
out
so
I
believe.
That's
our
little
birdie
told
me
that
you
still
love
community
I.
Do
I
do
more
than
usual.
This
way,
I
would.
B
C
B
C
Can
I
just
interject,
yeah
cuz
YouTube,
just
taught
me
something
new
live
chat
replay
your
chat
messages
are
now
saved
and
may
be
seen
by
viewers
who
watch
the
recording
of
the
stream
learn
more
so,
ladies
and
gentlemen,
who
are
typing
into
the
window,
please
be
mindful
now
that
your
chats
will
be
here.
Forevermore,
not.
C
Lost
at
the
end
of
this
Jam
lefty,
usually,
but
this
is
kind
of
cool
I'm
gonna-
be
interested
to
see
what
this
looks
like
this.
B
Exciting
we've
finished
yeah
anyway,
yeah,
okay,
I'm,
gonna
click,
the
got
it
acknowledgement
there,
okay
cool,
so
there's
that
I
have
shared
that
information.
Okay,
so
cool!
We
have
two
similar
ish
takes
on
upgrading
existing
apps
to
SP
net
core
to
one
preview,
one.
So
that's
exciting!
It's
always
you
know
there.
There
is
documentation
and
both
of
these
posts
call
that
out.
But
then
they
talk
about
here's.
What
I
ran
into
and
here's
some
things
that
you
know
the
that
worked
for
me.
What.
B
Is
pretty
straightforward,
so
I
mean
you
know
the.
A
C
B
C
B
So
I
would
say
this
is
the
kind
of
straightforward,
easy
button.
This
one
from
Steve
Gordon
here
he's
talking
about
the
humanitarian
tool
box,
the
already
project,
and
so
his
are
usually
a
little
more
in-depth
because
it's
a
more
complex
project
right.
So
here
he's
talked
about.
You
know
again
the
standard
stuff,
here's
how
you
can
update
it,
some
things
he
did
talk
about.
He
talks
about
the
the
unit
unit
test
project
update.
B
C
C
Way
through
they
really
love
it.
When
people
try
the
previews
and
then
log
feedback
when
they
find
issues,
I
think
go
live
with
it
anyway,
but
like
honestly,
like
I,
keep
on
about
this
every
week,
but
we
really
really
really
need
folks
to
try
the
previews,
especially
these
feature
release
previews
yeah.
This
was
quite
a
loan
milestones.
C
We
have
added
a
whole
bunch
of
stuff,
and
you
know
we
don't
really
know
how
it's
gonna
be
received
unless
folks
use
it
and
we
don't,
we
would
much
prefer
that
to
happen,
obviously
in
the
previews,
because
we
can't
really
change
things
in
meaningful
ways
after
the
RC.
So
it's
really
great
when
we
see
people
do
this
so
I
mean
that's.
C
So
if
you're
watching
this
show-
and
you
have
not
downloaded
the
two
one
preview
and
tried
it
out-
not
just
like
done
file
new
but
like
looked
at
some
of
the
new
features
or
upgraded
an
existing
app
I
I
would
do
you
to
please
try
and
find
some
time
in
your
schedule
to
do
so
and
if
there's
anything
that
is
causing
that
to
be
difficult
for
you
give
us
that
feedback.
If
you
do
nothing
else
other
than
log
an
issue
on
they
spent
at
home,
that
says:
here's.
B
Okay,
so
well,
I'm
well,
I
have
Steve
Gordon
here,
here's
another
one
from
him,
this
one
he
posted
last
week
during
MVP
summit,
so
kudos
for
that,
and
here
he's
looking
at
creating
a
host
builder
for
micro
services.
So
his
the
overall
here
is
set
web
host
builder
is
web
focused
and
the
idea
here
of
creating
a
host
builder
just
for
hosting
micro
service
itself,
so
not
with
all
the
kind
of
web
hosting.
It's
not
like
there's
much
overhead,
but
you
know
so.
This
is
a
specific
just
a
host
builder
right.
C
I
wouldn't
go
as
fast
I
mean
micro
services
to
me
a
little
loaded
so
just
to
clarify
that
slightly
it's.
Basically,
you
want
to
run
an
app,
it's
a
console
app.
You
want
it
to
run,
and
you,
like
all
the
stuff
that
a
spirit
core
gives
you
up
into
the
point
that
you
have
to
accept:
HTTP
requests.
So
what
am
I
talking
about?
C
You
get
dependency
injection
and
a
host
that
manages
the
lifetime
with
the
app
like
it
responds
to
control
C
and
the
shuts
down
the
application,
for
example,
or
sig
term
on
Linux
configuration
logging
in
configuration
in
dependency,
injection
and
hosting
startup,
and,
like
those
type
of
things
you
want
that.
But
you
may
not
be
running
a
web
server.
You
might
be
D
queuing
stuff
from
a
rabid,
mq
or
an
azure
thing
or
listening
to
an
event
or
you
know
or
something
else.
So
it
is
that
type
of
thing
go
home.
Alright,.
B
So
I
had
talks
through
that
you
know
and
good
amount
of
sample
code
etc.
Talking
about
how
to
create
that
generic
host.
Okay,
here's
a
few
tips
from
David
pine:
these
are
asp
net
core
configuration
tips
a
little
bit
opinionated
the
ones
here
that
he
calls
out
here.
One
is
using
c-sharp
six
property
initializers
for
default
values.
So
this
just
in
terms
of
kind
of
cleaning
up
your
code
instead
of
you
know
doing
these
all
in
it
in
a
constructor
or
whatever
yeah
totally.
B
C
So
the
API
looks
like
it
does,
because
we
want
people
to
have
the
flexibility
to
do
what
makes
them
happy
say
if
you
like,
having
a
type
mapping
between
the
type
name
and
the
section
in
the
config
that
you
want
to
bind
to,
then
name
love
works
perfectly
reasonably.
But
if
you
don't
want
to
do
that,
then
the
API
is
such
that
you
can
do
whatever
the
hell.
You
want.
That's
why
we
did
it
that
way.
So
all.
B
Right
here
this
is
from
Ben
Foster
kind
of
curious.
Your
thoughts
on
this
here
he's
he's
using
URL
helper,
so
he
wanted
to
use
a
URL
helper
for
building
URLs,
so
not
having
a
hard-coded
URL,
and
so
he
said,
you're
a
helper
wasn't
available
by
default,
and
so
he
kind
of
went
and
you
know
then
then
you
sit
there.
I
just
wondered
if
that
was
intentional
or
that's.
C
Completely
intentional,
you,
the
the
usage
pattern
for
getting
a
URL
helper,
is
to
resolve
the
factory
and
then
ask
the
factory
to
create
one
for
you,
okay.
Otherwise
we
would
have
put
the
URL
hopper
in
the
eye
for
you
yeah
there
is.
There
is
a
reason
is
to
do
with
this
slight
variance
between
services
that
are
scoped
versus
ones
that
need
to
be
activated,
and
so
they
have
slightly
different
patterns
and
what
we
consider
one.
You
know
what
we
consider
an
activated
type
versus
a
like
a
service,
and
it's
not
always
clear-cut
but
effectively.
C
When
you
get
the
factory
out,
which
is
a
single
thing
that
lives
in
the
NDI,
it
has
a
create
method
on
it
and
you
have
to
pass
it
some
some
context
and
you
can
see
there.
It
needs
an
action
context,
so
action
contexts
in
DI
by
default,
and
you
can
see
that
he
had
to
add
the
action
context
accessor
in
order
to
get
the
current
one,
because
it's
ephemeral
there
is
an
overhead
and
doing
that,
so
you
actually
pay
slightly
to
to
be
able
to
do
what
he's
doing
there.
C
B
C
Yeah
I
mean
yeah,
exactly
I
mean
there's
two
different
mean
it
depends
on
what
you're
doing
it
so
really
what
it
comes
down
to
is
like.
Where
is
it
that
you're
trying
to
generate
the
URL
like
at
what
point
in
the
application
and
what
code
path
are
you
trying
to
get
a
URL,
because,
obviously,
if
you're
in
anywhere
control
by
MVC,
it's
fairly
straightforward
to
get
a
URL
helper?
If
you're
in
a
view,
then
we
obviously
just
give
that
to
the
base
type.
If
your
controller,
the
base
type,
has
URL
helper
on
it.
C
A
C
More
nuanced
and
that
it'll
only
work
if
you're,
in
the
context
of
something
that
Vesey
has
already
resolved,
because
there
won't
be
in
action,
context,
right
action,
be
executed,
and
so,
depending
on,
where
you
try
to
do
this,
you
can
it'll
explode
horribly.
So
you
do
have
to
be
aware
of
all
that
stuff.
This
is
something
that
we're
looking
to
is.
It
is
a
gap
like
if
you
want
to
generate
a
URL
today.
That's
that's
all
in
the
scope
of
MVC.
C
If
you
want
to
have
this
idea
of
knowing
how
to
get
URLs
from
outside
of
MVC,
we
have
a
plan
in
the
longer
term
to
look
at
that
using
something
called
dispatching.
This
basically
means
of
removing
routing
to
earlier
in
the
pipeline
and
then
adding
new
features
to
it,
so
that
it's
not
as
tied
to
MVC.
So
we
have
some
prototypes
of
that
and
you
have
to
do
that
in
a
future
version.
Oh
look
at
is
that
it
was
that
the
plug-in
chain
that
just
caused
you
to
to
remove
that
yeah.
B
B
B
And
then,
as
far
as
if
you
want
to
read
more
about
that,
that's
in
the
release
notes
those
switches
here
so
yeah
just
kind
of
listed
those.
So
those
are
in
the
links
I'll
share
for
this
week.
If
you
want
to
read
more
so
just
kind
of
handy
to
have
those.
So
john
hilton
has
been
doing
a
series
here,
and
this
is
the
most
recent
one.
This
is
on
complex
security
policies
for
an
asp
net
core
app.
B
So
he
talks
about
you
know
creating
a
policy,
a
custom
policy
in
this
one
he
goes
through,
and
this
example
is,
somebody
can
only
has
access
to
something
when
they
have
completed
their
basic
training
and
have
three
months
of
employment,
so
kind
of
an
arbitrary.
But
you
know
it
is.
It
is
pretty
often
in
a
business.
You
have
kind
of
complex
requirements
and
it's
instead
of
kind
of
hard
coding
that
into
a
controller
or
a
filter
or
whatever.
B
This
is
this,
is
you
know,
you've
done
by
creating
a
policy
and
then
setting
that
up
mmm-hmm
cool,
and
if
you
want
to
look
more
into
this,
this
is
in
one
of
the
sessions
that
Jeff,
Ritz
and
I
did
in
our
MBA.
We
Jeff
built
out
a
custom
policy
there
all
right,
quick
couple.
Few
on
this
is
on
bootstrap
for
and
this
this
is
a
look
by
Ignacio
talking
about
telluric
UI,
with
asp
net
core
and
bootstrap.
B
For
so
I've
been
kind
of
keeping
an
eye
on
bootstrap,
for
you
know
the
work
involved,
so
here's
kind
of
a
look
at.
What's
you
know
what's
new
in
bootstrap
for-
and
you
know
talking
about
like
so
you
know
some
big
things
like
flexbox
using
ram
instead
of
pixel,
which
is
great,
so
you
know
some
some
nice
kind
of
updates,
so
here
Ignacio
walks
through
updating
a
project
and
and
using
bootstrap
for
with
that
and
then
just
kind
of
in
the
links
I'll
just
kind
of
flip
through
these.
B
But
if
you're,
if
you're
interested
telluric
UI
phrase
peanut
core
is
a
paid
product,
just
you
know
which
is
cool.
That's
fine!
We
like
that
too,
and
here's
here's
the
code
for
that.
It
includes
a
you
know:
here's
how
to
do
it.
Here's
so
you
know
just
looking
at
I
dug
through
his
views
and
I'm
linking
to
that.
Also.
Okay,.
C
We
do
have
been
tensions
by
the
way
to
move
our
own
templates
to
bootstrap
for
at
some
point,
it's
just
work
is
that
we
need
to
do
and
we
have
like
we've
kind
of
fully
booked
for
two
one
already,
and
it's
not
just
the
templates.
We
have
to
actually
update
scaffolding
as
well,
because
we
have
lots
more
than
that
now
because
into
1
we
have
UI
in
class
libraries
as
part
of
the
razor
feature.
C
We
would
have
to
update
many
things
in
order
to
move
to
bootstrap,
for
so
we
have
to
do
it.
We
know
we
want
to
do
it,
but
it's
not
just
as
simple
as
updating
a
template.
We
have
to
ensure
that
our
scaffolding
knows
when
to
do
use
bootstrap
three
verses
before,
depending
what
your
scaffolding
into,
and
we.
B
B
All
right,
here's
here's
an
interesting
one
from
catch,
our
wiring-
and
this
is
talking
about
easy
caching.
So
this
you
know
from
the
two
posts
on
this
that
I've
looked
at.
This
is
kind
of
a
wrapper
around
a
caching
pattern.
So,
instead
of
kind
of
have
to
create
a
you
know,
checking
if
something's
in
a
cache,
if
not
go,
get
it
he's
kind
of
built,
a
library
on
top
of
that
with
some
utility
methods,
so
yeah
and
then
what's
also
kind
of
nice
with
this
is
they're
swappable
providers.
B
So
there's
a
built-in
memory
provider,
but
it's
also
possible
to
point
it
at
Redis.
He's
got
one
for
memcache,
so
some
different,
you
know
kind
of
swappable
sequel
light.
So
the
idea
there
is
it's
kind
of
a
generic
caching
provider
and
then
you
could
earn
caching
system
and
then
you
can
switch
your
provider.
That's
kind
of
neat.
You
know,
maybe
for
testing
use
your
caching
against
memory
or
whatever,
and
then
you
know
in
production
you're
using
against
Redis.
B
So
this
is
I
believe,
is
this
Pippen
Joshi
I
believe
it
is
just
a
sec
because
I
want
to
get
the
name
right.
It
is
so
anyhow
this
this
one
is
server,
sent
events
in
asp
net
core,
so
this
is
yeah
so
interesting
and
of
course
you
could
do
signal
are
still
support,
servers,
an
event.
That's
right.
Yeah.
B
If
you
want
to
do
actually
yourself
implement
this,
this
is
a
look
into
doing
that
so
creating
server
sent
events,
and
then
this
is
you,
know,
kind
of
sending
events
out
based
on
database
updates,
cool
stuff
and
actually,
in
this
case
it's
you
know,
it's
implementing
he's
got
jQuery
code
cool,
so
this
one.
This
is
interesting.
Missus
learned
lasercom.
This
is
a
community
thing
and
it's
it's
clear
here
in
the
about
there.
This
is
just
some
excited
community
folks
that
are
going
through
and
and
kind
of
writing.
B
Through
here
this
is,
you
know,
this
is
kind
of
you
know,
documentation
on
here's,
how
to
do
it,
some
looking
at
all
kinds
of
stuff
data,
binding
and
all
kinds
of
stuff.
So
this
is
and
it's
clear
on
the
front.
You
know
it's
like
they
talked
about
that
hey.
This
is
this
is
improve
each
experimental
preview
etc,
but
yeah
pretty
pretty
pretty
interesting
and
being
built
out
by
the
community
there.
So
good
stuff,
okay.
B
B
This
is
this
is
kind
of
turning
on
things
like
service
workers
manifests
and
all
that
stuff,
so
that
your
app
works
well
kind
of
your
website
works
well
as
an
application,
so
this
is
kind
of
just
a
nice
kind
of
a
quick
look
at
doing
that
configuring
services
and,
and
all
that
and
built,
you
know
kind
of
looking
at
it
from
the
point
of
view
of
asp
net
core
mm-hm
yeah
interesting
stuff.
That's
definitely
something
I'd
love
to
hack
around
with
more
oh.
This
is
from
sky
addy.
B
Scott
addy
is,
is
posting
on
creating
a
link
button.
So
basically
Scott
misses
the
link
button
that
was
in
and
web
forms
and
so
here's
how
to
create
something
similar
in
asp
net
core
in
raiser
pages
so
going
through
and
whoops.
Sorry,
I
scrolled
a
little
faster.
So
there
it's
yeah
he's
got
a
tag
helper,
so
simple
button
tag
helper
nice.
B
This
one
is
a
kind
of
a
repost.
This
is
one
that
came
up
recently.
We
featured
it
on
the
stand
up
at
least
once
before,
but
somebody
asked
Scott
about
this
on
on
Twitter
and
and
I
mentioned
it
and
we
got
tons
more
people.
You
know
looking
at
it
and
tons
more
attention
for
it.
So
I
just
want
to
make
sure
people
are
aware
of
this.
This
is
a
free
asp,
net
core
book.
It's
Josh,
I
think
it's
150
or
more
pages.
B
D
B
To
be
aware
of,
this
is
a
new
thing
that
it
was
just
just
released.
This
is
built
by
cloud
developer,
advocates
I,
believe-
and
this
is
this
is
a
function
library.
So
a
lot
of
people
are
using
functions
and
logic
apps.
You
know,
in
the
background,
to
do
services
to
do
things
for
ASP,
NetApp's
and
so
I
just
thought
this
was
worthwhile
to
share
this
is
a
function
library,
so
it's
function
library
as
your
websites
dotnet,
and
it
has
all
kinds
of
different.
B
You
know
a
little
bit
more
than
this
sort
of
simple
samples,
you'll
see
for
functions,
a
nicer
Keable
library
there,
so
you
know
like,
for
instance,
email
right,
so
you
know
you
can
go
through
FTP,
you
know.
So
this
is
this
Bernini
very
cool
Oracle,
so
Oracle,
the
ODP
net
core
beta
2,
is
now
available.
So
this
release
the
beta
2.
B
Yep,
okay,
so
there's
asp,
net,
boilerplate
and
asp
net
and
make
sure
I
say
it
right
and
there's
asp
net
core,
boilerplate
and
I.
Just
there's
there's
two
different
things:
they
both
had
releases
and
just
to
make
sure
that
I'm,
clear
and
not
to
you
know
like
just
being
straight
upfront
with
everybody.
This
is
there's
two
different
ones
and
this
one
we
featured
recently.
This
is
the
asp
net
boilerplate.
B
C
C
C
B
This
one
has
4,500
commits,
so
you
know
yeah,
so
I
would
just
say
both
of
these
are
out
there.
Both
of
them
are
active
and
being
update.
So
there's
a
lot
of
stars
too
yeah,
so
this
this
is
very
useful.
This
this
three
five
update
includes
some
updates,
for
instance
like
quartz
core,
so
of
quartz,
the
kind
of
timer
schedule
library
update.
B
So
you
know
several
updates
and
enhancements
and
then
not
long
ago,
Jan
ended
and
there
is
a
three
for
release,
so
pretty
active,
so
I
just
wanted
to
congratulate
them
and
call
out
the
three
five
released
there
and
then
for
the
asp
net
core
boilerplate.
What
what
muhammad
rohan
sahib
has
been
updating
is
the
core
API
stuff.
So
this
is
focusing
on
the
api,
including
things
like
you
know:
pre-baking
some,
you
know
like
swagger
and
and
stuff
in
and
so
in.
B
B
B
C
C
B
C
Please
please
PLEASE
the
game
like
the
the
the
viewing
numbers
on.
This
are
pretty
good,
that
there's
not
quite
as
high
as
I
would
expect
yeah,
given
our
customers
we
have.
We
do
suspect
that
our
SEO
on
this
blog
might
not
be
as
good
as
it
should
be,
and
we're
looking
to
fix
that
like
if
you
type
a,
is
B
net
blog.
You
do
not
get
this.
D
C
If
you
go
to
dub
dub
dub
that
asp.net
and
follow
the
blog
link
at
the
top
of
that
website,
you
will
get
here,
but
we
did
like
a
dozen
posts,
I
think
four
to
one
preview,
one
with
each
post
being
an
overview
of
a
particular
feature.
We
have
a
couple
lined
up
still
too
so
yeah.
Please
do
read
those
again
in
and
tell
the
preview
and
try
and
give
us
feedback.
Okay,
I'm
in
the
final.
B
A
B
He
just
wanted
to
get
asp
net
core
running
on
GoDaddy.
It's
not
supported
there,
so
he
was
like
well,
heck
can
I
just
get
in
there
and
install
it
anyhow.
So
he
SSH
is
in.
He
goes
through,
you
know,
starts
creating
stuff
runs
into
permission,
problems,
the
craziest,
maybe
saddest,
or
maybe
best
I'm,
not
sure,
but
there's
a
part
here
where
it
actually
wasn't
mapping
the
port's
correctly.
So
he
used
PHP
as
a
proxy
I.
Think.
C
B
That
individual
in
this
is
this
is
some
horrible
stuff,
but
I
think
part
actually
of
what
what
he's
doing
here
is
saying
like
look
I
can
do
this
hacking
around
on
here.
We'd
love
from
our
hosters
to
actually
support
this
out
of
the
box,
and
you
actually
can
admin
your
machines
and
you
can
map
the
ports
yeah.
A
B
C
He's
podcast:
this
is
Hanson
minutes
and
he
updated
it.
He
converted
it
from
the
old
web
pages
framework
to
razor
pages
a
few
weeks
ago,
yeah
and
then
well.
I
may
as
well
use
two
one
preview,
one
as
well,
so
he
did
that
and
then
he
was
like.
Oh
it's
on
Azure.
Let
me
add
application
insights,
because
why
wouldn't
want
to
do
that?
So
he's
never
really
had
this
level
of
time
until
much
on
the
site
before
and
so
after
doing
that,
he
realized.
He
has
a
whole
bunch
of
work
because
he's
like
oh
I've,.
A
C
These
errors,
and
oh
everyone
in
Bangladesh,
is
taking
like
three
days
to
see
my
site,
like
he
didn't
realize
how
bad
things
were
so
he'd.
Never
visualized
he's
issues
before
right,
which
is
the
first
rule
you
to
visualize
everything
and
then
once
he
done
that
he's
like
oh
I,
should
probably
go
and
fix
this
so
I'm,
assuming
he
probably
added
the
CDN,
and
then
he
started
looking
at
least
500
areas
like
why
am
I
getting
500
errors,
see
there
you
go
so
like,
and
so
I
we've
been
working
with
him
to
help
rectify
or
see.
C
C
He's
still
fixing
this
right
now
today,
right
so
there.
This
is
a
characterization
of
how
page
based
frameworks
work
right.
So
if
you
think
back
to
classic
ASP
right,
you
have
a
page
which
you
know
implies
a
URL
and
then
every
request
that
URL
goes
to
their
page
right,
we'll
go
to
theirs.
P
file.
Okay,
you
with
me
so
far
yep,
regardless
of
the
verb
like
it
doesn't
matter,
if
you
don't
again
or
ahead
or
adoptions
or
a
post
it'll
all
just
go
to
that
page.
Okay.
C
So
if
you
write
your
page
verb,
add
gnostic,
then
the
page
will
probably
execute
just
fine.
If
you
write
your
page
to
inspect
the
verb
coming
in.
Is
this
a
get
request?
Is
this
a
post
request
and
then
do
different
things?
Then
you
might
imagine
if
you
send
a
request
to
that
page
with
a
different
verb
that
the
author
wasn't
expecting,
then
you
might
get
a
result
that
you
weren't
expecting
right,
because
you.
C
Right
if
someone
had
someone
sets
ahead
request
to
an
options
request
which
is
quite
common
for
search
engines
to
do.
You
might
find
that
you
are
running
that
page
because
the
route
matches
and
then
but
you
don't
run
a
quote-
a
block
of
code
that
is
used
to
initialize
some
variable
or
some
property
that
you're
then
reading
later
on
right.
So
then
you
get
a
500,
because
you
get
a
null
reference
exception,
which
is
exactly
what
was
happening
here
now
in
traditional
MVC,
with
pattern
based
routes
for
serving
views
in
pages.
C
This
wasn't
typical
right,
because
all
verb
requests
for
a
page
for
a
review
would
go
to.
This
would
go
to
one
action
method
generally
right.
You
would
just
so
here's
the
pattern
go
here
and
then
you
would
have
me
you
might
have
a
specific
post
handler,
but
it
would
be
constrained,
and
so
the
route
system
would
pick
the
first
one
unless
the
other
one
was
more
specific.
If
that
makes
sense.
So
basically
course
we
get
selected
for
post,
but
then
all
other
verbs
will
go
to
the
the
pattern
based
route
for
attribute
routing.
C
Obviously,
that's
not
quite
how
it
works
like
you
decorate
it
with
HTTP
GET,
and
if
it's
not
again,
it
doesn't
execute
right.
So
that's
the
first
you
know,
inclination
of
where
we're
going
here
is
that
you
might
now
get
into
situations
where
you
hadn't
considered
the
non
get
requests
going
to
your
site.
Then,
with
raiser
pages
we
kind
of
took
the
attribute
routing
approach,
which
is
you
have
a
handler
called
on,
get
and
maybe
have
one
called
on
post,
and
we
just
do
that
verb.
Matching
we
find
look
at
the
verb.
C
Look
at
on
blah
is
your
handler
method.
We
execute
that,
but
we
run
the
page
regardless
right,
because
it's
a
page
based
frameworks
and
the
route
match,
and
so
we
run
the
page.
So
if
we
don't
run
your
handler,
but
you
we
run
your
view,
we
run
the
HTM,
the
CS
HTML,
then
you'll
probably
end
up
with
a
thumb
with
null
refs.
If
you
were
relying
on
some
code
running
in
your
handler
in
order
and
that's
what
happened
to
Scott
here,
so
we
have
worked
with
him
to
give
him
a
tweet,
page
method.
C
Selector
factory
thing
like
it's
a
type
in
the
framework
they're.
Basically,
given
the
requests
coming
in
figures
out,
which
page
method
to
call
on
your
page,
that
will
basically
do
a
fuzzy
logic
search
which
is
like
it
says,
is
literally
says
in
the
comments,
like
fuzzy
logic,
which
basically
means.
Oh,
we
give
the
current
handlers.
C
We
give
all
the
handlers
a
score
based
on
the
incoming
verb,
but
it
allows
for
situations
where,
if
like
a
head
request
comes
in,
but
you
don't
have
a
handler
for
head,
but
you
have
on
forget:
it'll
run
the
get
request
handler
right.
But
if
you
had
one
for
head,
it
would
get
a
higher
score,
so
it
would
be
the
one
that
actually
gets
executed.
C
That
will
look
at
making
this
change
in
2.1
so
that
you
know,
if
you
don't
have
an
exact
matching
handler
based
on
the
verb,
will
do
this
fall
back
to
the
closest
matching
handler
based
on
some
logic,
that
will
event
basically
will
fall
back
to
a
get
or
a
post.
So
if
it's
an
options
or
a
head
or
a
get
it'll
run
get
unless
you
have
a
specific
options
or
a
specific
head
handler
and
for
posts.
C
C
I'm
gonna
give
a
tech
empower
update,
so
we
had
MVP
summit
last
week
right
for
those
who
don't
know
it'll
all
they're,
our
our
highly
regarded
community
members
who
do
so
much
for
us
in
promoting
the
product
and
whatnot
in
supporting
customers
and
we're
in
Redmond
last
week
to
meet
with
the
product
team
and
yeah
discuss.
You
know
their
feedback
and
all
the
type
of
stuff,
which
was
great.
C
The
some
stuff
would
good,
and
some
of
you
may
have
seen
some
posts
that
came
out
of
that
and
stuff.
But
what
I'm
more
interested
in
is
we
had
some
discussions
around
performance
last
week
and
I
want
to
give
people
some
updates
about
some
of
the
things
that
we'd
be
doing.
So
you
ever
spoken
about
to
compare
in
a
while,
so
I
think
I
said
round,
15
came
out,
I
think
we
may
have
said
this
a
couple
weeks
ago
and
it
was
the
first
results
with
two.
Oh
and
yeah.
C
We
did
okay,
but
there's
a
lot
going
on
in
tech
and
power
right
now
and
I
want
to
give
folks
an
update.
So
we
have
been
doing
a
lot
of
work
on
Fortune's,
which
is
the
data
access
test
that
does
HTML
rendering
and
we
have
some
new
numbers.
So
our
new
numbers
which
aren't
in
the
they're
not
in
the
official
framework
yet
but
tech
and
power,
now
does
continuous
continuous
stuff.
How
do
you
say
continuous
results?
C
B
C
We
go
all
right,
so
this
is
the
continuous,
basically
tech
empower
framework
benchmark
results.
So
whenever
you
do
a
check-in,
they
run
the
change
tests
and
you
can
get
those
results
immediately.
So
you
can
say
visualize
this
run
on
the
TFP
website
and
it'll
actually
render
the
website,
with
the
results
that
you
asked
for
right.
So
this
one
didn't
include
any
fortunes,
but
I
know
it
included
plain
text,
so
you
can
see
here.
C
Their
physical
environment
they've
got
some
nice
new,
fast
numbers,
so
seven
million,
which
is
like
insane
I,
want
to
say
and
a
spirit
core
is
down
here
at
1.5.
We
know
this
number
is
wrong.
This
is
a
result
of
them,
Daka
rising
all
tests.
So
that's
the
first
thing
going
on
right
now
is:
second
power
is
standardizing
on
docker,
which
is
should
do
wonders
for
enabling
people
to
reproduce
results
better.
C
There
has
been
some
historical
challenges
and
even
tech
empowers
own
results,
replicating
numbers
from
run
to
run
and
we're
hoping-
or
we
have
a
firm
expectation
that
the
move
to
docker
will
allow
us
to
have
a
much
cleaner
separation
between
testing
framework
A's
would
be
framework.
Seen
any
given
run
with
much
less
chance
of
kind
of
them,
impacting
each
other
so
like
here
having
some
latent
bit,
rot
lasting
from
running
Neddy
and
then
running
a
spin,
a
second
that
affects
the
second
result.
C
If
that
makes
sense
right,
because
they're
running
on
the
same
machine,
they
literally
just
have
like
a
setup
script,
they
run
they're
a
result
or
a
framework,
and
then
they
have
a
teardown
script.
Did
I
reboot
the
machines
or
re
image
of
them,
or
anything
like
that
when,
in
between
everyone,
it
would
take
weeks
if
they
were
to
do
it
that
way,
and
so
moving
to
docker
should
hopefully
help
encapsulate
every
framework
so
that
if
it
is
doing
something
bad,
it
all
gets
torn
down
anyway,
when
the
docker
container
goes
away.
So
that's
good!
C
So
one
point:
five
million,
isn't
great:
we've
already
sent
a
PR
that
in
the
next
continuous
run,
we
expect
that
number
to
go
up
to
well
over
two
million
okay,
so
that
we
we
should
be
able
to
get
back
to
where
we
were
prior
to
the
docker
ization
stuff.
So
that's
good!
It
was
just
some
CPU
thread
tuning
stuff
that
we
had
to
do
now.
That
still
puts
us
near
it's
it's.
C
You
know
it's
above
the
fold,
it's
within
the
top
fifteen
or
twenty,
which
is
fine
and
there's
more
results
to
come
yet
so
this
isn't
finished,
but
we
would
like
to
be
up
here.
We
referred
where
this
before
this
classification
column
down
here
that
talks
about
what
type
of
test
this
is
houses
classified
as
a
micro
test,
but
you
can
see
all
the
top
ones.
In
fact,
the
top
ones
within
sort
of
fifty
percent,
a
marked
as
platform
tests
which
but
effectively
means
they're
just
testing
the
h-2b
server
and
nothing
else
right.
C
There's
no
framework,
there's
no
HTTP
method,
dispatching
there's
no
path!
Dispatching
like
you,
don't
get
handlers
right.
You
basically
get
a
request
came
in
here.
It
is
have
fun
and
it's
up
to
you
to
write
all
the
code
to
figure
out
what
you
actually
want
to
do.
Uris
statement
right.
So
this
is
a
raw
effectively
R
or
HTP
server.
So
we
are
working
on
one
of
these
tests
for
a
spinet
core
and
I.
Have
that
pull
request?
C
Open
right
here,
it's
up
on
the
benchmarks,
repo
github,
that
asp
net
I
didn't
get
on
slash
a
spear,
slash
benchmarks,
pull
request
for
one
two,
but
there's
more
than
one.
This
is
Ben's
pull
request
on
top
of
David's
pull
request
so
David
as
a
pull
request,
which
is
this
one,
which
adds
the
platform
scenario,
and
then
Ben
Adams
saw
this
and
he
is
now
made
his
own
improvements
right.
So
the
number
that
we've
gotten
to
now,
as
you
can
see
here,
is
6.4
million
requests
per
second
and
we're
not
done
yet.
Okay,
now.
A
C
If
you
look
at
this
code,
this
is
very,
very
raw
I
mean
it's,
it's
real,
like
a
were
using
Kestrel,
you
get
a
callback
per
request
and
you
get
a
queue.
Basically,
whenever
any
header
gets
read
on
the
request,
you
get
a
call
back
saying:
here's
the
memory
pointer
using
the
new
span
of
T
stuff,
so
here's
the
span
of
T
for
the
header
name
and
here's
the
span
of
T
for
the
header
value,
so
it
doesn't
allocate
any
memory.
C
It's
up
to
you,
whether
you
want
to
turn
that
into
a
string
or
anything
like
that
right.
So
this
is
as
high
performance
as
you
can
get
in
net
today,
right
and
so,
as
you
can
see,
the
benefits
are
quite
large,
and
so,
with
this
level
of
test
this
platform
level
of
test.
If
we
just
took
the
current
results,
we
would
be
up
here
at
number
three.
Okay.
Well,
we're
not
done
yet,
though
we
think
we
can
get
further
and
Ben
is
on
the
job
now
and
so
I
wouldn't
be
surprised.
C
If
we
see
some
improvements
in
this
fairly
rapidly-
and
then
very
you
know
hopefully
we'll
get
this
checked
in
very
soon,
we
can
get
it
submitted
to
tech
and
powers
repo
and
then
we
can
get
it
as
part
of
their
continuous
results
for
the
next
week
or
two,
and
then
we
can
actually
see
for
real
where
we
land
in
their
environments.
So
that's
really
cool
right.
It's
a
really,
really
good
improvement.
C
It
helps
understand
the
difference
between
using
Kestrel.
You
know
to
R
or
s
form
as
it's
just
a
pure
HTTP
server.
That
calls
you
back
whenever
it
has
data
and
says
well,
I
parse,
HTTP,
here's
a
header
what
you
want
to
do
with
it
versus
having
all
the
niceties
of
a
framework
like
a
snake
or
where
you
get
di
in
a
middleware
pipeline
and,
like
you
know,
a
nation
to
be
context
that
has
a
cookie
collection,
all
that
goodness
on
it.
There
is
a
cost
of
doing
that.
C
Now
the
cost
is
quite
high
right
now,
I
will
admit
and
I
would
like
it
to
not
be
that
high
and
we're
still
looking
at
ways
that
we
can
obviously
reduce
the
overhead
of
all
that
nicety.
But
it's
good
we're
just
visualizing
more
and
more
stuff,
the
other
one
is
fortunes,
and
so
there's
no
fortunes
results
in
this
run.
If
I
choose
a
previous
run,
no
fortunes
in
bear
I
think
it's
two
runs
ago:
let's
throw
this
one.
Okay,
so
fortunes
is
kind
of
the
the
kitchen
sink
test.
It
accesses
a
database.
C
Typically
Postgres
you
can
see
most
of
them
in
the
top
results
here.
Postgres
and
Sebastian
is
like
I
am
in
me
to
do
a
tech
empower.
You
can
see
right
now
and
I'm,
not
I,
probably
I,
shouldn't
click
that
it
could
be
something
super
secret
right,
so
I
right-click
that
Sebastian.
If
you're,
watching
the
stream
right
now
and
you're
trying
to
get
me
to
read
this,
come
round
to
the
room
and
like
knock
on
the
door
and
tell
me
it's
okay,
so
you
can
see
the
numbers
here.
C
We
have
traditionally
not
been
great
in
this
test.
In
fact,
I
think
we
have
a
result
in
here
currently,
if
I
scroll
all
the
way
down
here,
it
is
a
spirit
core
middleware
60,000
requests
per
second
okay,
which
is
still
better
than
it
used
to
be
there.
He
is
okay,
he's
telling
me
it's
okay
and
it's
at
16
and
a
half
percent
of
the
top
number.
We
have
new
results
that
will
hopefully
will
go
live
as
part
of
the
2.1
release.
That's
are
quite
a
lot
better
than
that.
C
If
I
have
an
email
from
Sebastian
here
that
show
what
they
are,
they
would.
The
new
results
put
us
at
about
two
hundred
and
seventeen
thousand
requests
per
second
for
fortunes.
That's
using
dotnet
core
2.1
on
top
of
the
new
soccers
transport,
with
the
new
npg
sequel
dotnet
driver
for
Postgres
on
linux
will
get
us
to
around
217.
That
would
put
us
at
number
seven,
so
top
ten.
So
that's
a
massive
jump.
C
You
know
we
had
previously
been
not
even
in
the
story
when
it
comes
to
data
access,
but,
as
you
can
see
here
with
2.1,
we
are
gonna
leapfrog,
a
whole
bunch
of
other
frameworks,
and
it
will
end
up
about
there
right
just
above
undertow,
assuming
nothing
else
appears
in
here,
but
you
know
still
amazing,
perform
amazing
improvements.
So
we're
very,
very
happy
with
the
progress
that
we're
making
there
now
Jason.
C
We
still
have
a
ways
to
go,
so
we
are
currently
focusing
a
lot
of
effort
into
trying
to
understand
the
performance
discrepancy
between
where
we
do
on
plain
text
versus
where
we
do
and
Jason
it's
the
difference
between
pipelining
a
should
be
request
and
not
pipelining
them
for
very,
very
low
overhead
request,
processing
which
Jason
is
like
it
doesn't
do
a
lot
of
work,
and
so
we
have
a
lot
of
very
smart
people
at
Microsoft
at
the
moment.
Talking
about
and
theorizing
about
what
might
because-
and
this
we
also
have
people
in
the
community.
C
Of
course,
like
men,
is
at
the
forefront
of
this
as
well,
and
we're
working
diligently
to
try
and
really
understand
what
it
is.
That's
holding
us
back
in
the
Jason
benchmark
in
particular,
so
you
can
see
here
for
Jason
the
top
performers,
like
you
know,
but
starting
to
get
up
towards
to
million
requests
per
second
for
non-pipelined
HTTP
right,
which
is
just
crazy,
yeah
only
512,
clients,
I.
Think.
If
you
look
here,
it's
like
the
top
performer
is
actually
hitting
his
top
result
with
only
256
clients.
C
So
that's
256
levels
of
concurrency
they're
there
they're
starting
to
bump
up
against
two
million
requests
per
second,
we,
if
I
scroll
down
and
down
here
right
at
575
and
so
we're
gonna,
be
submitting
a
platform
level
test
for
this
as
well,
which
is
again
is
that
is
basically
kestrel
as
a
raw
server
with
just
and
then
we
just
bolt
on
the
JSON
bit.
But
we
do
not
expect
at
this
point
that
doing
that
will
will
give
us
a
really
good
result.
C
In
fact,
I
personally,
don't
even
think
it'll
get
us
close
to
a
million
which
is
good
because
it
means
that
all
their
theories
about
where
the
problem
is
right
now
will
be
true,
and
we
know
where
to
focus
our
effort,
but
we
need
to
figure
out
why
it
is
that
a
platform
level
test
for
us
doesn't
put
us.
You
know
up
in
this
epsilon
because
it
does
for
plain
takes.
You
know
up
toward
unity
and
rapid
lloyd,
which
is
where
we
really
want
to
be
alright.
C
Let's
open
up
this
secret
I
am
from
Sebastian
who
says
it's
a
good
thing.
Oh
maybe
there
is
a
new
run.
Look
at
that.
So
let
me
click
on
this.
Let's
see
if
it
opens
in
the
right
browser,
yeah
that'll
do
so
we
are
visualizing
I!
Guess
it's
plain
text,
severally
I!
Don't
see!
Oh
no,
that's
the
same
result!
No
he's
gonna
ping
me
now
and
say:
you're
looking
at
the
wrong
one.
No,
what's
he
what
you
want
me
to
look
at
nothing
secret,
just
a
good
link!
Oh.
D
C
That
it
said
all
you
wanted
me
to
show
is
just
the
link.
Okay,
I
already
showed
this
nevermind
I.
Think
I
did
so.
This
is
basically
the
same
as
me
clicking
through
from
here
right,
so
I.
He
gave
me
whatever
one
idea
was
a
five
six
see,
which
is
yet
this
one
here
alright,
so
we
click
on
visualize
this
one
on
the
checkup
area
and
drag,
and
you
get
this
and
you
get
a
plaintext,
and
this
is
the
the
faster
or
the
most
recent
plaintext
result,
so
yeah.
So
now
we
get
to
the
concrete.
C
You
know
things
have
happened
here
is
taken,
power
is
moving
to
docker,
which
is
fantastic.
It
means
the
test
should
be
more
repeatable
for
everyone
for
both
the
cloud,
which
is
a
sure
you
know,
for
core
I,
think
there
F
four
machines
or
D
four
machines
or
something
I
think
they
do
say
they
say
them.
The
blog
post,
when
they're
released,
so
that
there
is
for
people
folks
to
replicate
and
for
their
physical
environments,
are
moving
to
docker
as
well.
So
you
should
just
be
able
to
like
clone
the
repo
grab.
C
C
It
makes
it
much
easier
to
compare
results
and
then
for
all
the
framework
owners
to
ensure
that
they
can
replicate
results,
so
they
can
submit
fixes
and
feel
confident
that,
when
the
next
tech
empower
continuous
run,
which
is
the
other
big
improvement,
they've
done
when
the
next
continuous
run
occurs,
which
is
running
basically,
you
know
all
the
time
that
you
can
very
quickly
verify
that
the
fix
that
you
made
is
actually
being
represented
in
their
infrastructure.
So
that's
great,
really,
really
good.
C
B
C
So
you
can.
You
know
the
JSON
test
simply
represents
the
a
realistic
service
scenario
where
you're
doing
non
pipeline
requests,
which
means
that
every
client
only
ever
sings
that
ever
sends
a
single
request
at
a
time
and
the
server
does
some.
You
know
non-trivial
work
news
up
an
object.
It
C
realises
that
object
via
JSON
serializer
out
to
the
response
stream.
Okay,
so
it's
not
much
work,
but
as
soon
as
you
introduce
any
type
of
latency,
you
get
an
understanding.
C
Now-
or
you
know,
some
CPU
overhead
of
doing
so,
like
serialization,
there
will
be
ripple
effects
throughout.
You
know
the
system
on
how
well
that
scales
I'm.
You
know
there
are
many
real-world
scenarios,
that's
all
the
service
doing
if
you're
running
a
front-end
proxy,
that's
doing
something
very
simple,
like
investigating,
looking
the
incoming
payload
or
looking
at
the
incoming
requests
and
then
making
a
decision
based
on
the
path
and
the
verb
about
where
to
send
that
somewhere
else.
C
C
Then
you
move
through
the
data
access
scenarios
where
now
you're
introduced
in
latency
injured
this
into
the
benchmarks,
and
now
you
want
to
ensure
that
the
server
continues
to
scale
and
perform
well,
while
it's
waiting
for
IO
operations
on
a
different
machine
to
take
place
as
well
as
obviously
exercising
just
how
well
their
data
stacks
are
written
like
what
is
the
database
driver
like
is
the
connection
pooling
good?
What
does
the
HTML
rendering
stack
look
like?
Is
it
memory
efficient?
Is
it
fast?
How
does
it
scale
so?
C
I
think
the
suite
that
taken
power
currently
has
is
a
good
suite
it
just
by
no
means
complete,
because
no
tests
no
benchmarks
whatever
it
is,
and
they
are
actively
working
with
the
community
to
add
new
scenarios.
The
two
that
I
am
very
keen
on
myself
to
get
added
is
basically
adding
HTTPS
to
all
of
the
existing
scenarios
so
that
you
can
run
every
scenario
with
HTTPS
and
without
it.
C
So
you
understand
the
overhead
of
doing
HTTPS
on
your
stack,
because
that's
basically
should
be
the
default
today,
I'm
very
happy
to
say
that
HTTPS
on
kestrel
into
one
is
miles
ahead
better
than
it
used
to
be
back
in
the
old
1x
days.
So
the
the
pure
overhead
of
doing
HTTPS
on
the
sockets
transport
is
much
much
much
lower
than
it
used
to
be.
So
that's
a
really
really
good
like
to
the
point
that
plaintext
approaches
a
million
four,
even
with
HTTPS
enabled
which
is
fantastic.
You
know
versus
1.8
without
it,
then
the
database
stuff.
A
C
The
other
scenario
I'm
interested
in
is
the
is
like
the
proxy
scenario
right.
You
got
two
servers
like
a
micro
service,
you
type
thing
a
front-end
server
that
takes
a
payload,
makes
a
decision
based
on
the
route
effectively.
Maybe
does
some
very
light.
Processing
of
the
payload
sends
that
to
the
backend
asynchrony
waits
on
it
to
be
processed
by
the
backend.
It
gets
it
back
and
then
sends
it
back
to
the
client
right.
C
So
it's
similar
to
data
access,
but
you're
testing
the
outgoing
HTTP
stack
rather
than
the
outgoing
data,
so
they're
critical
pieces
of
infrastructure
that
you
need
to
measure
when
making
an
assessment
of
any
you
know.
Web
frameworks
stack,
obviously
I
think
in
my
mind,
cuz.
They
are
very
common
things
that
everyone
does
so
yeah.
Hopefully
that
answers
the
question,
but
you
know
ultimately,
this
is
data.
B
B
C
So
there's
we
there,
the
one
that's
submitted
a
second
power
users,
Jason,
don't
nap,
because
that's
the
bus,
the
the
most
common
one.
We
have
done
test
with
Jill
as
well,
and
it
shows
a
slight
improvement.
It's
not
massive.
Just
given
the
type
of
object
being
serialize,
it
doesn't
make
a
huge
difference.
C
We've
also
done
a
test
using
a
non
allocating
serializer,
which
a
community
member
wrote
and
published
it
is
on
new
get,
but
we
wouldn't
submit
that
to
take
in
power
because
we
wouldn't
class
that
as
production
ready
or
production
grade,
which
is
a
prerequisite
for
the
tech
and
power
tests,
so
you
can't
just
submit
really
nearly
code.
It
does
say
that
anything
you
submit
should
be
production
grade
now.
Obviously,
that's
open
to
interpretation.
There
are
frameworks
on
there
that
I
believe
no
one
on
the
planet
uses.
C
You
could
argue
that
they're
not
really
production
grade
if
no
one's
ever
used
them,
but
you
know
it
is
what
it
is,
but
when
it
comes
to
dotnet
I,
don't
think
it
would
be
particularly
fair
of
us
to
just
take
a
new
get
package
that
has
a
single
release.
That
is
an
experiment
that
is
demonstrably
faster,
but
you
know,
I,
don't
think
wouldn't
meet
would
would
meet
the
sniff
test
for
production
ready
in
the
normal
sense
of
what
net
means.
C
Jill
certainly
would
and
as
I
say,
we
have
done
that
before
the
platform
tests
that
we're
going
to
submit
for
Jason
will
we'll
use
Jill
because
it
will.
Let
us
eat
gout
a
little
bit
more
performance,
but
it
is
very
much
not
the
bottleneck
there.
Right
now.
We
need
to
do
work
deeper
down
in
the
stack
cool.
B
C
No,
that's
actually
one
of
the
great
things
that
came
out
of
this
entire
effort,
and
we
just
happened
that
a
long
time
ago,
people
think
back
like
three
years
ago
when
we
started
this
show.
We
did
a
lot
of
talk
about
this
from
the
very
beginning
right
and
we
showed
the
periphery
the
very
the
very
first
perfect
that
we
had
that
I
used
to
run
and
one
of
the
things
we
did
was,
let's
just
build
a
server
that
isn't
really
a
server.
Let's
just
build
a
c-sharp
app
that
pretends
it's.
C
An
HTTP
server
opens
a
socket
and
just
cheats
like
just
wildly
cheats
right
to
get
the
fastest
possible
thing
that
we
could.
You
know,
just
basically
counts
the
bytes
coming
in
when
it
sees
two
carriage
return
line,
feeds
assumed
just
mean
of
an
HTTP
request
and
then
returns
a
pre
cached,
pre
buffered
result,
and
at
that
point
we
were
able
to
very
quickly
see
that
we
could
get
millions
of
requests
per
second
in
terms
of
doing
Network
IO
by
poking
into
libuv,
without
an
awful
lot
of
optimization.
So
we
were
convinced
very
early
on
that.
C
The
dotnet
and
c-sharp
were
not
the
problems,
with
the
caveat
that
doing
some
of
the
things
in
the
c-sharp
that
you
had
that
you
have
to
do
to
reach
this
type
of
performance,
previously
weren't
fun
that
is
before
span
and
memory,
came
a
mod
in
2.1,
200,
first
internally,
in
kestrel
and
now
into
one
for
public
use
and
the
pipeline
system
borrowed
our
pipelines
namespace.
Without
those
types
riding
this
type
of
curtain,
c-sharp
wasn't
particularly
fun.
C
Okay,
we
have
taken
all
those
learnings
and
we
have
codified
them
into
span
and
memory
and
pipelines,
and
these
supporting
infrastructure
compiler
checks
the
serialize
of
types.
You
know
those
type
of
things
so
that
writing
this
type
of
code
now
is
easier
for
everybody.
Okay,
so
it
has
always
been
possible.
It
just
used
to
be
a
lot
harder
now
that
said,
yeah
let.
C
C
Equivalent
for
strongly
pollute
back
on
the
agent
core
for
rapid
rest.
I
sorry
I,
don't
know
what
that
is,
I,
don't
either
so
I
can't
tell
you
if
there's
an
equivalent
I
will
say
that
we've
done
some
experiments
about
yeah
utf-8
Jason
is
the
non
allocating
serializer
now
someone's
asking
we
have.
They
make
some
experiments
with
what
it
would
look
like.
You
know
when
you're
building
HTTP
API
is
what
are
the
type
of
things
you
want
to
do?
C
You
might
often
have
something
like
postman
up
or
you
might
be
using
curl
at
the
command
line
or
something
like
fiddler
to
issue
requests
right,
because
you
may
not
have
a
strongly
you
might
be
using
swagger
right.
You
might
be
using
something
like
that.
That
gives
you
a
test.
Client.
We
thought
to
ourselves,
you
know
what
would
it
mean
to
as
a
command-line
developer
or
you
know,
even
an
ID
developer
had
that
had
a
console.
C
What
again
that
type
of
idea,
but
for
HTTP,
like
we
had
an
interactive
command-line
experience
that
once
you've
booted
your
server
and
you
can
code,
your
API
is
in
one
window
and
then
you
can
go
to
the
console
window
and
issue
requests
to
those
api's
to
the
point
that
you
treat
it
like
a
resource.
I
can't
very
resti.
You
can
change
directory
around
your
URL
structure
and
you
know
use
the
metadata
system
that
the
server
has
the
MVC
has
to
understand
at
a
given
endpoint.
What
parameters
should
be
exceed?
C
What,
if
that
metadata,
could
drive
a
command
line
experience
so
that
you
could
you
know,
do
a
question
mark
slash
question
mark
at
any
point
in
your
API
and
it
would
print
out
the
parameters
for
you
like
it
was
a
command-line
interface.
We
actually
have
prototypes
of
those
type
of
things
that
we
hacked
up
quite
a
while
ago.
Now
that
are
quite
compelling,
we
have
shown
them
to
a
few
people
and
the
other
eyes
are
not
go
home
I
can't.
That
would
be
an
amazing
experience.
C
So
we
have
some
ideas
around
how
we
can
improve
that
type
of
feedback
loop.
As
an
API,
developer
and
I
wouldn't
be
surprised
if
you
see
something
like
that,
get
invested
in
in
a
future
version.
Oh
yeah,
that's
all
the
questions.
Wow,
okay,
well,
I
will
tell
everyone
that
we
are
hard,
as
ones
all
the
pur
stuff.
We
are
hard
at
work
on
the
second
preview
for
2.1.
We
will
do
a
second
preview,
I
think
I
said
in
the
blog
post,
where
we
plan
to
do
a
second
preview
than
we
plan
to
do
an
release.
C
Candidate
it'll
be
a
true
release.
Candidate
with
a
go,
live
license
that
you
can
put
in
production
and
we
will
support
you
followed,
obviously,
by
a
final
release.
Hopefully
we
won't
need
more
than
one
release
candidate.
We
aim
to
be
feature
complete
in
preview
2.
So
the
idea
is
that
when
you,
when
preview
2
comes
out-
which
we
hope
will
be
within
a
month
from
now
that's
hard
time
frame
to
give
you
an
idea,
it's
not
months
away.
It's
not
next
week.
C
It's
kind
of
you
know,
four
to
five
weeks
away
for
the
weeks
way
that
it'll
be
preview.
You
know
the
preview
will
be
feature
complete.
So
you
know
we'll
roll
in
feedback
that
we
got
from
preview.
One
will
finish
any
feature,
work
that
we
hadn't
completed
for
preview,
one
so
that
all
features
should
be
well
enough.
C
Baked
that
you
can,
you
can
use
them
and
give
us
feedback
on
them
and
then,
but
there
may
be
bugs,
and
then
we
would
fix
all
the
bugs
and
tighten
down
anything
that
we
need
to
for
the
release
candidate,
so
you'd
be
ready
for
you
to
use
and
then
shortly
after
the
release
candidate,
hopefully
very
quickly.
If
we
get
enough
usage
and
we
feel
confident
about
it,
we
can
roll
out
the
final
release
so
and
again
that
that
will
likely
be
some
way
towards
mid-year,
so
we're
already
halfway
through
March.
So.
C
Yeah,
so
everything
we've
talked
about
your
regarding
performance
is
in
the
scope
of
2.1
and,
of
course,
the
other
performance.
That's
a
huge
and
2.1
that's
improved
is
the
build
time
performance
and
the
preview
1
to
preview
to
improvements
are
fairly
dramatic,
so
preview
1
was
always
was
already
much
faster
than
2
hour
or
the
you
know,
sdk
to
1/4,
which
is
the
current
RTM
sdk,
and
the
preview
is
to
1
300
preview
1
preview.
C
2
of
that
sdk
will
be
almost
twice
as
fast
again
in
some
scenarios
and
we
expect,
by
the
end
of
2
1,
that
the
performance
of
the
of
the
sdk
the
build
time
scenarios
should
be
very
close
to
on
par
in
terms
of
how
it
feels
and
using
it
with
what
you
are
used
to
with
the
traditional
sort
of
full
framework,
dotnet
asp,
net
or
a
class
library,
or
a
console
app.
So
you
shouldn't
be
paying
any
cost
for
doing
those
things
in
some
is
actually
faster.
So
we've
made
fantastic
throughput
of
that
cool.
B
C
C
D
B
C
Christoph
disaster
will
the
tear
compilation
thing
be
active
by
default
into
one
if
they're
referring
to
sort
of
Roslin
like
CST
compilation,
the
answer
is
no
razor
is
having
compilation
changes,
however,
into
one
in
preview.
We
will
enable
razor
compilation
on
build,
so
the
default
experience
will
be.
When
you
build
your
project,
we
build
all
of
your
razor
views
right
all
your
razor
files,
as
opposed
to
only
doing
your
dream,
publish
that
yields
a
whole
bunch
of
benefits.
It
means
that
you'll
find
any
errors
in
your
razor
files.
C
When
you
build
your
project,
which
is
great,
it
is
faster
overall,
because
we
can,
we
will
have
a
build
server
in
the
background
when
you're
doing
that,
so
that
the
second
time
you
build,
we
get
to
cache
a
whole
bunch
of
stuff.
We
did
the
first
time
so
incremental
changes
which
result
and
build
will
now
be
much
faster
when
doing
that,
which
is
great.
And
thirdly,
it
also
helps
us
do
the
whole
embedded
razor
stuff
in
the
class.
So
everything
a
lot
better
by
having
a
build
server.
So
all
good
things.
B
So
we
do
have
some
swag,
although
they've
actually
shut
down
a
lot
of
the
stuff
we
use
sticker
more
for
stickers
and
that
actually
they
don't
allow
print
on
demand
for
stickers
anymore.
So
we're
switching
that
over
yeah,
so
not
too
much
more
to
say
there,
although,
if
you're
having
trouble
getting
something
I
can
order
it
or
I,
can
you
know
get
one
off
links
for
people?
We
have
another
spam,
that's
exciting.
Some
people
asking
about
EF
core
fef
core
is
on
the
same
release.
Yeah.
C
So
that
everything
here
is
on
the
same
release
cycles,
so
II
of
call
signal
our
HT
net
cord
net
core,
they
all
ship
on
the
same
cycle
they
all
ship
in
the
same
vehicle
right.
So
yes,
you
can
reference
e
if
core
packages
or
other
packages
from
your
you,
WP
apps
or
don't
ephemera
caps,
but
don't
net
everything
call
ships
on
one
cycle.
Okay.
C
So
when
I
talk
about
preview
to
I'm,
talking
about
the
whole
thing,
so
there'll
be
EF
stuff
that
we
don't
make
or
stuff,
although
most
of
the
stuff
now
is
in
the
EF
layer
and
the
asp
net
core
layer.
So
so
lewis
asks
is
the
idea
of
having
a
spinet
team
create
a
tool
for
load
testing
application.
I
know
there
are
many
out
there,
but
I
think
that
he
might
be
able
to
come
up
with
something
great.
C
C
Tool
yeah,
which
is
still
around
you,
can
still
use
it,
but
it
is
very
Windows
centric
and
it's
written
in
native
code.
It
supports
plugins,
but
you
have
to
write
them
in
C++.
The
reality
is
that
we
have
other
parts
of
the
company
that
deal
with
those
aspects
of
building
applications
for
more
than
a
am
aspect.
So
there's
the
Azure
load
test
tool,
which
is
pretty
cool.
It's
built
on
the
vs
load,
testing
infrastructure
or
the
VST
s-slow
tester,
which
you
can
get
in
the
enterprise
SKUs
of
Visual,
Studio
I
believe.
C
But
if
you
want
a
very
basic
tool
for
load
testing,
we
just
use
work
which
runs
on
Linux
and
you
can
run
it
on
the
windows
subsystem
for
Linux.
It's
w
RK,
it's
a
very
low
overhead,
all
the
benchmark
stuff.
I've
been
talking
about
users
work.
If
you
want
something
even
simpler,
that
doesn't
need
to
be
quite
as
fast
but
perhaps
has
more
features.
C
You
can
use
something
like
a
p
AV,
sorry,
which
is
apache
bench,
which
is
quite
popular,
and
the
node
ecosystem
has
a
bunch
of
quite
popular
load
tools
as
well
that
are
in
NPM.
It
really
comes
down
to
what
features
you
want.
If
you
want
something,
that's
very
high
featured,
and
you
know,
has
an
Orchestrator
and
condemned
all
can
organize
multiple
machines
to
generate
load
and
then
collect
data.
W
cat
does
all
that
stuff,
as
does
like
the
the
visual
studio
enterprise
web
load
test
tool.
C
C
C
Someone
asked
about
multi-tiered,
you
can
cause
he
like.
Yes,
no,
there
was
no
plans
to
ship
multi-tier
at
all.
Right
now
he
was
it
was
announced
as
a
experiment
like
I,
think
on
Twitter
awhile
back
I.
Don't
even
think
there
was
a
blog
post.
I
might
be
wrong,
there
might
be
a
markdown
file
and
it
get
up
somewhere,
but
I'm
not
aware
of
any
firm
plans
to
actually
ship
that
at
some
point
it
was
something
that
we
were
interested
in.
C
Looking
at
I
personally
feel
that
would
be
something
would
be
good
for
us
to
invest
in
more
I
think
it
would
help
with
those
sort
of
inner
loop,
I'm
just
doing
development
scenarios,
where
I
just
want
the
fastest
startup
time
possible.
If
I
make
a
code
change
and
hit
save
I
want
that
to
be
reflected
in
my
app
in
the
shortest
amount
of
time
like
in,
like
hundreds
of
milliseconds,
not
two
or
three
seconds.
An
interpretive
or
multi-tiered
JIT
in
theory
should
help
those
scenarios.
C
So
I
would
like
to
see
us
do
that
at
some
point.
Let's
have
a
look:
here's
TLS
into
one
only
for
the
localhost
Giovanni,
I'm,
not
sure
I,
follow
if
you
mean
in
the
templates.
Yes,
the
templates
and
the
defaults
that
kestrel
has
is
that
it
has
special
scenarios
around
detecting
configured
endpoints
with
the
presence
of
our
new
dev
cert,
which
is
obviously
for
localhost.
Only
so
that's
setup,
but
it
also
supports
simplified
configuration
for
TLS
in
non
development
scenarios
as
well.
C
So
we
have
that
in
the
blog
post
that
we
talked
about
earlier,
you
can
go
and
check
that
out
there.
That
would
be
good
to
get
some
feedback
on.
We
also
paged
in
support
for
sni,
which
we
have
an
announced.
But
last
week
we
had
a
customer
explicitly
reach
out
around
adding
s
and
I
support
two
Kestrel.
We
had
planned
to
do
this
eventually,
but
it
wasn't
in
the
scope
four
to
one,
but
we
have
now.
We
are
now
attempting
to
try
and
fit
that
in
the
to
one
schedule.
C
So
s
and
I,
for
those
who
don't
know,
is
what
allows
you
to
have
a
single
web
server
listening
on
one
port
but
serving
HTTP
for
multiple
hosts.
Okay,
so
I
think
it's.
Nice
stands
for
server,
name,
identifiers
or
server,
name
identification
or
something,
and
so
as
opposed
to
having
have
a
separate
IP
address.
Sorry,
one
IP
address
not
one
point
in
the
old
SSL
Dasia
to
have
a
separate,
IP,
separate
IP
address
for
every
host
that
you
were
trying
to
serve
HTTP
traffic
for
with
s
and
I.
You
don't
have
to
do
that.
C
This
have
been
supported
in
is,
since
version
eight
seven
point:
five,
maybe
something,
but
if
you're
doing
cast
your
self
hosting,
not
behind
a
reverse
proxy,
like
you're
running
cached
on
the
edge,
you
might
want
to
do
s
and
I
there.
So
we
are
looking
aggressively
and
adding
that
into
one.
So
that
would
be
good
for
the
folks.
You
want
that.
What
do
we
got
any
good
good
tutorial
on
the
new?
What
is
that
our
pipeline's
API
from
vladislav
I,
don't
know
ever
tutorial?
C
There's
been
a
couple
of
blog
posts
that
it
might
all
be
wildly
out
of
date
because
they're
from
before
it
got
merged
into
core
FX.
If
anyone
on
the
team
is
listening,
I
mean
obviously
we're
gonna
have
to
get
all
that
stuff
dot,
because
it's
part
of
the
two
one
train
now
Nick
I,
don't
know
if
Mark
is
planning
on
doing
an
updated
series
of
posts,
because
I
know
he
did
some
one
pipelines
previously.
But
we've
had
quite
a
bit
of
interest
in
that,
but
it
is
not
trivial
to
get
started
in
pipelines.
C
It
does
need
some
good,
solid,
I,
think
guidance.
So
yeah
we've
had
that
feedback
a
couple
times
in
the
last
couple
of
weeks,
but
I'm
not
aware
of
one,
but
we
should
you
know
we
should
fix
that.
So
did
you
do
anything
else?
I'm
scrolling,
I'm
scrolling
I'm
scrolling
any
more
interesting
change
to
the
spa
templates.
We
should
know
I've
tried
the
preview
ones
Brian,
not
that
I
know
of
other
than
by
the
preview
ones.
I'm,
assuming
you
mean
the
new
ones
that
use
the
new
CLI.
C
C
Then
Tim,
that's
like
saying
just
use
work
work,
all
the
things,
that's
what
you
want
to
use.
So
I
don't
disagree
with
him.
It
depends
on
what
you
want
ice
express
locks
and
localhost.
You
can
actually
make
ice
express
work,
not
on
localhost,
it's
my
most
popular
Stack,
Overflow
answer
ever
it's
where
all
my
stack
overflow
points
come
from
me
pointing
to
the
blog
post
that
someone
else
wrote
on
how
to
make
is
Express,
listen
on
non-local
host.
So
if
you
want
to
do
that,
there's
a
command.
C
You
can
run
as
an
administrator
at
the
command
line
to
set
up
the
net
TCP
bindings
with
an
SH
binding.
Sorry,
so
that
HTTP
sis
will
allow
that.
So
it
is
possible.
You
can
do
that
right
now.
If
you
want
to
do
docks
on
server
side
now,
I
don't
know
any
stuck.
Sorry
mastic
talked
to
Steve
about
that.
So
ever
name
indication.
They
go
correction
of
my
as
SN
I
before
and
weekend
at
the
bottom.
This
probably
should
be
what
about
OData
new
or
a
graphic
QL
support.
C
I,
don't
have
anything
to
add
there
other
than
I.
Think
I've
said
previously.
The
idea
team
has
stated
that
they
are
adding
the
a
spinet
core
or
data
stuff.
I
have
nothing
to
add
on
graph
QL
I
think
there
has
been
some
discussions
but
I'm
not
aware
of
any
news
or
anything
to
report
there.
So
if
you
want
graph
QL
and
have
very
compelling
needs
for
it,
tell
us
log
an
issue
on
the
home
repo
so
that
we
have
data
that
we
can
use
to
help
us
prioritize.
C
These
things
is,
you
know,
we
only
have
a
limited
man
of
resources.
Unfortunately,
people
and
we
have
to
choose
what
to
work
on
next,
so
to
do
sequel,
always
encrypted
an
entity
framework
or
Jeff.
That's
the
best
better
question
for
them:
I
always
get
confused
between
the
different
sequel,
encryption
features.
I
know
one
of
them
works
just
fine
and
it's
basically
transparent
to
anyone
connecting
to
it
sequel
always
encrypted
I'm.
Not
I,
can't
remember.
If
that's
the
one
that
has
impact
on
the
clients,
I,
don't
I,
don't
know
my
hunch
says
it
just
works.
C
I,
don't
think
it
has
any
impact,
but
I
might
be
wrong
so
best
to
ask
that
question
on
the
entity
framework
core
repo
on
github
mark,
a
secret
blog
post
and
Doc's
I
might
have
seen.
Tim
is
saying
that
mark
and
then
Nick
is
realizing
that
now
the
chat
saved
that
mark.
C
C
Yep
yep
very
good,
very
good,
that's
it
and
I
will
the
blaze
a
team.
Do
a
similar
stand-up
thing.
That's
a
great
question:
I
will
ask
tend
I
mean
blazer
is
obviously
still
officially
an
experiment.
There's
no
commitment
that
blazer
will
ever
ship
a
supported
version.
So
I
just
have
to
reiterate
that
every
time
that
comes
up-
but
there
are
many
folks
interested
in
it.
If
you
think
a
show
like
this
regularly
for
blazer
like
weekly
or
every
two
weeks
fortnightly,
they
say
where
I
come
from.
B
B
C
Wouldn't
be
surprised
if
folks
are
looking
for
something
a
little
more
regular
way.
You
know
they
could
highlight
community
work,
that's
happened.
They
could,
you
know,
inform
people
of
any
takers,
because
there's
such
a
rapidly
moving
landscape
at
the
moment
with
blazer,
because
it's
you
is
it
it's
very
much
an
experiment.
Bits
are
only
just
becoming
available
for
people
to
try
out,
and
things
can
wildly
change
from
week
to
week.
C
It's
kind
of
the
same
reason
we
started
this
show
in
the
first
place,
right
like
we
were
in
the
middle
of
beta
3
and
like
there
we
had
like
another
5
there
just
to
go
and
things
are
changing
every
week.
We
wanted
to
make
sure
that
we
could
stay
ahead
and
get
the
message
out
there.
So
I'll
suggest
that
to
Dan
and
see
if
he,
if
he
likes
that
idea
so
yeah
time
spent
a
fortnight
mi
look
forward
to
that.
Yes,
expansion.