►
From YouTube: .NET Design Review: GitHub Quick Reviews
Description
00:00:00 - Approved: Add UnderScoreCase support for System.Text.Json https://github.com/dotnet/corefx/issues/39564#issuecomment-532341171
00:09:39 - Needs Work: Add mechanism to handle circular references when serializing https://github.com/dotnet/corefx/issues/41002#issuecomment-532349735
00:32:28 - Approved: HashAlgorithmName.FromOid https://github.com/dotnet/corefx/issues/40558
00:45:49 - Approved: Add RemoveIfValue to ConcurrentDictionary https://github.com/dotnet/corefx/issues/24770#issuecomment-532361229
B
How
people
have
asked
for
in
the
past
and
B
is
like
okay,
we
just
stick
with
camel
and
Pascal
for
now
and
then
once
we
have
500
open
again,
we
work
on
this
feature,
so
that's
what
they
proposed
and
how
many
members
are
eager
to
implement
it,
which
is
why
I
want
to
get
it
reviewed
now
to
avoid
I'm
going
to
unlock
them
and
has
to
look
at
other
endpoints
like
Facebook
and
Twitter
and
other
endpoints.
They
all
use
campus
snake
casing.
So
it's
it
JEP
is
why
Danny
becoming
a
force
lad?
It
went
in.
C
C
Mean
because
there's
just
that
I
was
saying
that
you
know
when
you
don't.
The
lower
number
is
fine
and
works
well
for
English,
but
then
upper
to
upper
like
when,
when
do
you,
do
it
from
upper
topper,
always
never
after.
B
C
B
B
D
G
I
C
It's
carried
in
the
operating
system
for
the
most
part,
but
it's
it's
defined
by
the
initial
standard,
like
Jeremy
said
as
never
changing
ever
under
any
circumstance.
The
only
way
you
could
ever
get
into
a
situation
as
if
he
is
undefined
characters
and
your
property
names,
but
please
don't
do
that.
I.
B
C
A
A
B
B
E
C
B
I
mean
the
Sears
or
options
has
two
concepts.
One
is
for
properties
where
you
can
use
the
Navy
policy
and
the
other
is
for
extreme
key,
which
you
can
also
use
a
very
possible.
So
you
can
set
forth
a
dictionary
key
policy
and
property
in
imposing.
What
do
you
configure
this
son
Jesus
it
as
a
reference?
B
M
N
N
F
A
B
So
I'm
gonna
let
David
drag
this.
One
is
nutrition,
reminding
this
effectively
the
gist
of
it
is
we
have
we
don't
have
the
reference
loop
handling
and
reservations,
yeah
and
you
know,
Jason-
has
it
and
customers
with
it
and
for
Spartacus
was
the
EF
and
etcetera
and
laser
the
wireless
feature,
and
if
you
find
a
box
for
quite
a
bit.
O
Well,
I
think
some
tomato
one
of
the
last
things
that
bin
Laden
is
that
which
will
check
that
the
purpose
of
the
property,
the
property
lynnster
is
getting
from
the
from
the
reflection
is
no
doubt
it's
not
deterministic.
So
the
way
we
should
probably
if
they
ignore
option,
because
not
always
will
give
this
at
there
in
the
same
result,
I
don't
know
which
will
check
that
there.
You
can
hear
that
in
order
to.
B
O
Okay,
so,
basically,
if
you
have
two
objects
and-
and
you
have
a
reference
to
at
some
point-
the
reference
goes
back
to
the
first
object.
It
would
create
a
look
right
now.
We're
just
checking
against
maxstep
I
went
to
we
reach
that
yeah
the
cellulite
approach,
and
why
we're
trying
but
I
we're
trying
to
achieve
knowledge.
I
like
for
a
mechanism
to
palliate
up
front
in
order
to
check
for
the
reference
loop
handling,
so
you
don't
have
to
rely
on
the
Mac
step.
P
R
G
C
Want
to
you
don't
want
to
interpret
on
D
serialization
by
default,
anything
that
goes
against
the
Jason
specification,
including
circular
references,
because
what
happens
is
we've
seen
cases
where
somebody
will
send
a
an
object
graph
with
cycles
to
a
server
the
server
will
attempt
well,
the
server
will
successfully
deserialize
it.
One
of
the
properties
might
be
bad,
an
exception
will
get
thrown
inside
of
the
exception,
handling
logic
they
try
to
write.
The
object
to
a
log
generally
involves
a
called
a
two
string
thought
to
string
now
stack
dives
and
you
crash
the
process.
O
Also,
there's
no
like
a
correctable
behavior,
because
I
checked
many
other
she
devices
and
languages
and,
for
example,
is
during
a
fight.
It
was
not
the
event
and
all
this
kind
of
stuff
it
is
drugs
and
any
other
and
languages,
for
example,
one
of
the
libraries
from
Java.
Yes,
that
you
should
just
pass
a
property
to
ignore
their
sorry,
sorry
to
pass
on,
naturally
to
another
property.
So
you
can
also
avoid
the
serialization
of
that
specific
property.
O
O
Q
B
B
J
J
O
J
O
C
B
B
These
are
so
if
you're,
an
employer
manager
object.
If
you
had
a
customer
order,
object
that
people
want
this,
just
as
a
bow,
the
reference
ignore
and
your
hands
Praetorian.
So
it's
like
one
of
the
most
highly
requested
feature
that
we
ended
up.
Chef
is
act
when
six
thumbs
up
so
yes,
our
answer
is
yes.
I
mean.
B
C
I
feel
like
I,
understand,
preserve,
preserve
as
I
have
a
I
have
a
complex
graph
make
the
graph
materialize
on
the
other
side.
Ignore
seems
like
a
thing
that
you
succeed,
but
can
you
reason
about
the
thing
that
succeeded
when
you
rehydrated
on
the
other
side,
you've
lost
it
I'm
just
wondering
if
that's
actually
useful.
A
B
P
B
B
C
Like
in
this
scenario,
what
you
really
would,
if
you
would,
you
would
want
to
not
realize
reports
do
or
manager
whatever
as
the
moment
that
you
start
see
realizing
you
know,
you
say
realize
me
and
now
you
get
Dan,
and
so
now
you
get
Steve
and
like
and
now
like
you're
serializing
me
why
I
saw
SIA
in
this
rap.
That
doesn't
make
sense.
So
you
that's
a
place
where
you
don't
want
to
walk
up.
You
may
want
to
have
your
object,
know
how.
A
Is
that,
as
far
as
earnest
and
ignore
is
basically
a
function
of
what
your
data
is
versus
the
the
one
that
I
proposed
the
base
is
just
a
function
of
your
scheme,
all
right,
your
schemer
stakes,
you
can
reason
about
the
schema.
You
just
never
get
reports
to.
For
instance,
all
you
get
reports
to
it
all
in
the
first
guy,
because
the
second
guy
below
just
know
I
have
it
like.
How
do
you
reason
about
data
like
that?
Well,.
C
The
reference
graph
up
here
but
ignore
like
emo
and
I,
are
saying
the
same
things
like
yeah
you,
you
had
a
thing
and
now
you
send
it
somewhere
and
it
either
has
all
the
data
you
wanted
or
because
there
was
a
cycle,
it's
abruptly
cut
off
and
now
you
don't
know
what
you
have
on
is.
So
that's
why
I
was
really
asking
yeah
you
people,
the
people
who
want
ignore
understand
they
want
ignore.
They
want
to
not
have
an
exception,
because
maybe
what
they
want.
This
Christmas
the.
B
C
Is
that
lately
so
I'm
a
little
confused
of
theatrical
scenario,
because
if
you're
trying
to
preserve
the
entire
object
graph
and
you're
trying
to
do
it
over
Jason,
which
is
a
payload
format
that
clearly
cannot
support
it?
Why
do
people
not
want
metadata
as
part
of
the
payload
and
then
as
one
like,
like
dollar,
sign
ID
dollar
genre,
like
you're
you're,
asking
someone
to
serialize
something
in
a
payload
format?
That
clearly
is
not
expect
to
be
able
to
do
the.
A
Only
scenario
is
something
like
you
know:
you
basically
have
you
basically
have
a
payload
that
UDC
LS
did
somebody
else,
control.
It's
yes
and
then,
on
your
end,
when
you
actually
define
your
classes,
you
actually
have
back
pointers
or
other
things
that
make
your
life
easier,
but
but
it
still
has
it
back
out
that
will
not
work,
because
no,
you
have
cycles
right,
but
I
see
the
way
you
leave
with
this.
A
Think
the
the
only
case
where
what
makes
sense
is,
if
I
do
what
we
said
we
shouldn't
do,
which
is
you
see
you
soon
as
the
schema
that
you
don't
actually
control
you
try
to
butcher
it
into
submission
right,
in
which
case
it
just
make
it
work
by
saying.
Oh
yeah
give
me
the
top
as
payload,
but
I,
don't
know
how
you
would
reason
about
the
end
result
when
you
know
the
first
employee.
K
A
I
think
people
that
would
want
those
ID
various
thing
as
people
that
basically
control
the
scheme
on
clients
or
everyone
just
want
to
use
Jason
for
whatever
reason,
and
they
based
use
it
as
a
glorified
civilization
technology
right
versus
the
other.
One
is
more
like
well,
there's
a
service,
this
service,
it
has
the
schema,
so
I
just
have
to
make
that
scheme
ever
no
matter.
What,
but
yeah
I
don't
ignore,
seems.
C
A
T
L
C
Yeah,
like
you
could
have
you
know
if
you
have
a
an
array
that
has
the
same
thing
in
at
twice
right
like
if
it's
this,
if
it's
reference
to
the
same
object
and
with
reference
preservation,
you
would
want
it
to
say.
The
second
thing
was
a
reference.
Was
reference
equal
to
the
first
thing
with
don't
preserve
references,
it
would
flip
them
both
out
as
objects
during
deserialization.
O
During
this,
is
you
serious
ocean?
Yes,
well,
on
on
kids
on
the
net,
there
are
some
people,
I
mean
I,
found
in
a
stock
overflow
with
you
that
they
that
the
guy
was
asking
for
something
to
disable.
This
thing,
when
his
rate
is
destabilizing
and
also
place,
another
thing
you
can
it,
for
example,
if
your
ID
is
not
on
the
same
or
on
the
on
the
above.
C
O
P
T
O
C
If
I
did
clear,
dollar
sign
wrapped
twice
level,
also
I'll
have
you
know
the
reason
I'm.
Bringing
this
up
in
API
review
is
because
we
should
also
be
considering
behaviors
and
during
API
review,
and
it
is
like
the
D
serializer
behavior
has
to
be
really
well
defined.
What
it
sees
something
like
this
yeah
I
would
definitely
need
opt
in
to
yeah.
Oh,
it
absolutely
has
to
be
often
like.
Unquestionably,
deserialize
should
not
affect
yeah.
Oh
yeah,
it
is
a
can't,
execute
a
holder.
Have
this
one
yeah.
B
C
The
reason
the
reason
something
like
this
can't
be
often
by
default
is
because
say
that
say
that
you
modify
Angela
to
say:
Angela's
manager
is
Angela,
so
Angela
manager
rough
one.
Now
the
problem
is
to
submit
that
to
the
server
the
server
tries,
displaying
the
the
organizational
hierarchy
of
Angela.
The
server
now
gets
a
Stack
Overflow
exception
in
the
process.
I'm
like
we
can't
have
stuff
like
this
on
by
default.
U
C
Be
wherever
you
know,
but
you
know
whenever
you
handled
for
the
properties
or
values
from
the
payload,
that
don't
map
to
properties
on
your
time,
yeah
yeah.
So
if
you
didn't
put
an
overflow
bucket,
it
just
drops
it
dude.
We
did
put
an
overflow
bucket,
it
says:
I,
don't
know.
There's
this
thing
called
dollar
ID
here.
Q
O
You
hit
that
mess.
That's
why
the
option
we
have
exist
to
first
check
what
reference
do
you
have,
so
you
mean,
for
example,
if
you
have
on
at
least
reference
ID.
Well,
sorry
reference
to
it.
Yes,
sir
numbers
22
and
then
you
have
the
ID
1.
You
know
and
we
in
the
ID
1
your
yes,
you
are
the
finding
the
ID,
but
you
you
are
ready.
First,
the
ID
yeah
that
thing's
handled
by
their
I
have
property,
but
that
we
we
didn't
want
to
add
that
option,
because
no
one
is
asking
for.
B
C
K
T
K
T
K
C
L
Think
we
need
to
understand
the
use
cases.
Also
primarily
I
would
expect
this
to
come
up
when
you're
just
serializing
a
random
object
graph,
and
it
may
be
far
less
frequent
when
you're
using
the
Dom
is
with
the
Dom.
You
actually
have
to
kind
of
think
about
what
you're
putting
into
it
right,
but
I
mean
I,
guess
this
would.
C
F
B
C
That
I
think
before
you
can
say
that
you
actually
know
what
the
API
that
you
want
to
look
like
is
we
need
to
have
the
spec
that
defines
what
behaviors
you
so
I
think
that
there's
a
cycle
right
now
you
have
us
back
from
the
spec.
We've
said
there
a
lot
of
where
the
Eggers
would
like
to
know
the
one
you
have
written
down,
and
then
we
can
talk
about
the
second.
A
C
You
start
talking
about
what
are
all
the
fail
modes:
unbound
references
references
with
additional
properties
to
find
forward.
Look
ahead
like
all
these
making,
they
influence
the
feature,
totality
API,
and
then
you
can
cut
feature
once
you
define
what
the
whole
thing
should
look
like.
But
if
we
start
with
just
a
simple
thing,
then
it
may
end
up
worse
over
time.
Yeah.
A
I
think
it's,
the
kind
of
the
officer
problem
was
doing
language
design,
but
you
can't
have
the
syntax
before
you
have
the
behavior
right,
like
I,
think
once
we
had
the
behavior,
the
syntax
should
be
easy
to
make
a
the
API
should
be
really
straightforward,
but
we're
not
really
unblocking
anybody
very
improving
API.
So
we
don't
know
what
the
baby
is
when.
T
B
B
Whether
or
not
people
are
using
a
mercenary
whether
passing
data
have
cycles
in
the
wire
and
then
expected
it's
just
how
one
person,
one
group
of
people
use
this
Jason
stuff.
If
news
of
Jason
has
this
usage
and
people
are
using
it
a
lot
and
their
question
for
it,
how
do
we
say?
No,
you
need
to
copy
the
competition
air.
You
know
because
the
question.
C
Is
can
you
can?
Can
you
or
you
explain
to
me
when
I
want
it
when
it
makes
sense
and
why
I
would
use
it?
Because
if
not,
you
can't
write
documentation
at
a
sample
other
than
here's
a
convoluted
thing
we
came
up
with
so
without
actually
understanding
library,
adding
something
we
shouldn't
add
it.
So
that's
why
I
don't
care?
Who
has
the
concrete
request,
but
like
I,
want
a
very
specific.
This
is
why
this
makes
sense
yeah,
because
everything
I
can
think
of
is
this.
A
Having
weight
I,
don't
think
we
have
a
goal
to
make
porting
from
done
on
Jason
dotnet
to
assist
in
text
up
Jason
viable
for
everybody.
Right
like
the
goal
is
that
we
have
scenario
parity,
but
we
will
not
have
necessarily
identical
behavior
in
all
cases,
either
right,
but
I
think
the
question
is:
if
an
existing
implementation
has
will
be
considered,
ill-defined
behavior,
then
you
can't
really
match
that
anyways,
even
if
we,
even
if
we
had
ignored
that
we
have
been
no
signing
up
to
have
the
exact
same
corner
case,
behavior
Jason.
A
About
that
hand,
I
would
say
no
and
but
I
think
that
this
is
almost
impossible.
If
the
problem
is
well,
you
just
randomly
drop
some
stuff
on
the
floor
in
order
to
make
it
fits
a
certain
brand
right
and
then
becomes,
but
what
was
the
order
in
which
Jason
Lynette
reverse
the
object
grab,
maybe
towards
the
other
way
right
or
suddenly
we
make
different
decisions
where
we
lost
the
data.
I
didn't
know.
People
complain
again
that.
C
Oh
and
and
then
ideally
on
this
one,
and
we
find
the
Cory's
question:
can
we
find
anyone
other
than
James
who's
done
this
and
is
there?
Is
there
anything
that
looks
like
a
standard?
I
mean
I?
Read
that
there's
anything
yeah,
that's
a
thing
that
I
would
like
to
see
in
the
in
the
spec
light
of
that.
This
is
the
only
thing
that
does
this,
so
it
is
with
the
factors
as
opposed
before,
if
there's
three
different
competing
standards,
because
this
is
the
Internet
there
have
to
be
at
least
three
then
I'd
like.
A
Yeah
so
I
mean
then
the
question
becomes
a
menu.
We
can
know
one
second
after
design,
if
there's
only
two
modes.
Let
me
have
them
special,
that's
enum,
so
we
can
at
least
add
you
know
more
later
on.
If
you
really
want
to
and
I
think
that
would
be
fine,
but
I
think
somebody
would
have
to
define
what
that
actually
means
in
a
way
that
is
well-defined
and
not
randomly
dropping
data
on
the
floor
or.
A
I
F
D
C
A
F
A
Was
known
to
similar,
what
do
you
have
here,
some
sample
code
and
then
also
what
like
in
the
design
section?
Basically,
what
corner
cases
do
you
have
and
then
wholly
on
report
errors
like
it's?
It's
pretty
much,
maybe
a
document
slightly
larger
than
that,
but
not
by
much.
It's
just
more
focused
on
the
behavior
other
than
me.
Actually,
it's.
A
F
D
E
C
C
Inside
our
code
we
have
a
bunch
of
switch
statements
that
say:
switch
off
this
dotted
decimal
turn
into
algorithm
name,
whatever
Bevin's
written
that
a
bunch
of
his
own
code,
and
he
asked
why?
Isn't
it
just
public
API?
That's
a
really
good
question,
so
it
should
be.
We
do
have
a
rich
type
that
represents.
Oh,
it's
called
Owen,
but
it's
in
the
it's
currently
in
an
assembly
that
depends
on
this
assembly
and
it's
a
terrible
type,
because
everyone
thinks
it's
immutable,
but
it's
actually
mutable
so
I,
don't
like
actually
using
it
for
anything.
C
Semi
human
readable
string,
yeah,
so
11.3
no
1.28
60
died.
It's
the
atom
at
this.
At
this
point,
I
need
to
look
it
up
in
a
table,
become
shell
one.
Yes,
so
hash
algorithm
name.
You
can
actually
pass
into
other
api's
such
as
RSA
incorrect.
For
instance,
yeah
I
see
we
started
with
Mo
Kio
and
the
O
in
value,
and
we
hadn't
dropped
it
because
of
about
300
grams
yeah,
and
oh
it's
just
there
Jeremy,
so
not
really
a
good
type.
We
shouldn't
really
encourage
its
use,
yeah
or.
F
C
A
C
C
C
Strings
like
if
your
massive
I
think
that's
not
a
no
it'll
return
false.
If
you
pass
in
the
Oh,
would
representing
RSA
it'll
return
false
cuz,
that's
not
a
hash
out
growth,
so
the
case
where
it
will
return
true
is
the
six
things
that
we
know
about
the
case.
Where
will
return?
False
is
every
other
case.
Sorry.
A
C
A
C
C
J
C
P
B
C
Know
would
know
the
system
security
photography.
Oh
it
type
allows
you
to
work
both
from
would
numeric
dotted
decimal
value,
which
is
the
thing
that
everybody
really
talks
about,
and
then
it
also
lets
you
use
the
windows
friendly
name,
that's
associated
with
it,
and
so
in
all
the
API
is
that
I've
ever
proposed
the
take
that
that
a
decimal
we
call
it
old
value
to
be
very
clear.
We
are
not
going
to
try
friendly
neighbor,
accessible.
C
F
L
L
L
A
W
The
biggest
questions
here
also
are
early.
We
have
exclusively
rented
API
for
my
collection.
We
remove
a
game
developer
which
to
me
it
makes
the
most
sense
we
approve
this
API
is
we
have
another,
try
remove,
which
only
thing
all
your
a
believe,
the
key
which
it
takes
and
then
Eve
on
the
naming
is
confusing.
It's
when
you
accept
when
you
come
and
Eve
you
usually
have
a
Freddie
game,
Thank
You
Homer
right,
remove
it
Eve
the
predicates.
C
For
the
condition
dad
I
agree
with
Santi
if
we
want
to
have
this
is
a
public
method.
I
think
it
should
just
be
an
overload
of
try,
remove
and
either
take
a
Tiki
and
a
t-value
or
take
a
key
value
pair
of
Tiki
TIA
all
right.
So
the
Tiki
+
t-value
means
that
we
would
then
have
an
overload,
which
varies
on
graph,
which
violates
guidelines.
But
if
we
add
the
the
out
of
what
the
existing
value
was,
then
that
would
solve.
That
is.
Try,
remove
and
try
to
remove
with
an
extra
parameter,
because
so
what.
C
R
L
C
L
L
I
don't
know
it's
I
could
see
your
perspective.
It's
it
just
doesn't
seem
like
leading
from
having
two
ideas
that
are
so
similar.
We
will
give
you
that
they're
not
identical,
but
they're
similar
enough,
that
having
a
value
pair
on
one
and
not
on
the
other,
just
seems
kind
of
strange.
The
reason
my
perspective,
you
don't
you
can't.
C
Use
them
in
the
same
scenarios,
they
are
for
different
purposes.
One
of
them
is
removing
something,
but
a
key
value
pair
and
its
entirety.
One
of
them
is
replacing
a
value
that
there's
no
reason
for
like.
If
you
were
to
define,
try
update
with
a
key
value
pair,
it
would
look
I,
don't
even
know
what
a
little
I
guess
it
would
take
to
key
value
pairs
where
the
key
and
both
had
to
be
identical.
C
W
Agree
with
the
statement
that,
where
the
key
value
pair
you're
saying
we
remove
this
key
value
pair
and
when
you
iterate
on
atoms
on
the
dictionary,
you
usually
get
a
value
for
anything.
This
is
the
key
for
this
value
yeah
and
when
you
are
updating
you're
saying
on
this
key,
put
these
new
value
you're
not
saying
update,
update
this
whole
key
bumper
me
to
something
in
the
pool
yeah.
A
But
logically,
I
think
I'm
with
Korea
more
in
the
sense
that
what
you
do
really
is
you
give
a
key
and
the
comparison
value
right.
So
basically,
what
you
are,
what
you're
talking
about
is
a
version
key
right.
You
say
you
know,
update
this
key,
that
it
currently
has
this
value
and
you
either
remove
that
entry
or
you
added
or
you
set
it
to
a
new
value
wise.
It's
basically,
a
form
of
a
version
key
really
well,
both
like
I
may
try
after
base
it
says:
here's.
A
A
Basically,
it's
kind
of
a
of
a
versioned
key
right,
because
you
basically
use
the
value,
as
you
know,
as
the
as
the
words
met
it
currently
has,
and
so
in
that
sense,
remove
and
update
are
the
same
operation
and
from
the
point
of
view
that
or
you're
identifying
the
key
by
the
same
mechanism.
I'd
verify
its
its
key
and
its
current
value,
and
that's
the
thing
with
the
similarity
is
you're.
N
C
It
except
out
now,
if
it
doesn't
exist,
you
don't
pay
attention
to
the
values
it
does
exist.
The
key
is
sitting
there.
It
has
a
slot
associated
with
it.
We're
now
no
longer
going
to
touch
the
key
now,
just
looking
at
the
value,
slaw
yep,
and
then
we
can
be
say:
does
this
value
nothing
to
do
with
the
key
anymore
doesn't
match
this
other
value?
If
yes
place
it,
if
no,
no,
with
try,
remove
you're
saying
find
this
pair
in
the
dictionary
and
remove
it
if
it
exists.
B
L
D
L
O
C
Q
C
It's
an
elegant,
it's
obviously
it's
the
delegate,
but
not
a
delegate.
There
is
whatever
the
implicit
comparer
is
the
other.
Whatever
equality
compare
of
TDOT
default
gives
you
back,
and
so
you
know
it's
still
you're
still
potentially
executing
arbitrary
code.
The
difference
is,
in
one
case,
it's
very
explicitly
arbitrary
code
passed
at
the
call
site
that
could
reference
dictionary
itself
by
closure.
U
C
E
C
You
this
is
what
I
found
for
this
key
I
think
it
makes
sense
to
just
say
that
this
is
try,
remove
the
key
key
value,
comparison
out
value
existing
value,
and
now
we
don't
have
what's
the
over,
like
you
can
explain
why
the
overloads
different.
We
don't
need
the
suffix
because
we're
not
overloading
on
a
reduced
signature,
and
we
don't
need
to
keep
on
you.
There
we're.
C
C
There
that
we
removed
pretty
well
what
if
you
can
say,
evil
like
you've,
not
seen
that
depends
on
what
your
what
quality
compared
you
gave
the
concurrent
dictionary
anything
could
be
said
for
kids.
Well,
it's
it's
equality,
compare
of
T
value
default
in
all
cases,
because
it's
T
value
and
all
T
key,
all
right,
so
it'll,
so
it'll
just
be
reference
or
value
on
me,
okay,
nevermind!
So
it's
I
guess
so
what
it
would
be
is
the
key
value
out
to
key.
This
is
the
key
that
matched.
L
W
F
C
I
I
struggled
with
that
I
agree
with
you,
Santi.
On
the
one
hand,
on
the
other
hand,
we've
taken
some
liberties
with
try
in
in
concurrent
dictionary
and
some
of
the
other
concurrent
collection
types
in
general,
like
concurrent
dictionary,
doesn't
expose
ad.
It
exposes
triad,
there's
no
out
now
baby
we've
taken
too
many
liberties,
but
we
were
trying
to
drive
home
a
point
in
the
design
that
can't
trust
from
moment
to
moment
anything
about
the
state
and
we
wanted
to
avoid
people
doing
things
like,
if
not
contains
key,
add
yeah
I
mean
I.
I.
C
Think
that
you
know.
If
we
were
talking
about
the
span
and
memory
guidelines,
Chris
not
pointed
out
fit,
we
should
probably
take
what
we
currently
call
the
tryparse
pattern
and
make
it
the
try
pattern.
I
think
triad
actually
hits
that
as
the
I
return
false.
Instead
of
throwing
I'm
sorry,
this
already
exists.
So
I
I
think
that
we
found
the
correct
spirit
of
the
try
pattern
in
concurrent
dictionary
and
that
tryparse
is
a
subset
of
the
try
pattern.
C
D
V
C
J
C
It's
there
the
same
name
but
they're
they're,
not
actually
at
that
point,
they're
they're,
very
different,
so
try
remove
key
value
pair
that
try
reboot
key
value
pair
seems
to
be
the
least
controversial
do
and
then
it's
just
a
question
of.
Do
we
actually
want
it
and
I
don't
have
a
strong
opinion
for
it.
My
primary
my
primary
objection
to
the
issue
as
it
was
written
up
was
just
I've.
Seen
too
many
times
the
exact
pattern
that
was
used
to
motivate.
T
C
C
Certain
cases
I
just
want
to
make
sure
we're
not
doing
more
harm
than
good
and
to
be
clear,
the
example
the
current
example,
that
shown
is
fine,
where
it's
enquiry
updated
it
to
use
like
an
integer
pens,
breath
counting
a
ref
count
value
where
the
T
value
is
just
an
int.
The
previous
example
that
was
potentially
problematic
had
the
T
value
being
some
type
that
had
a
count
property
on
it,
and
the
example
was
basically,
if
dictionary
contains,
if
dictionary
after
I
get
value
and
the
value
equals
zero
I,
which.