►
From YouTube: 2021-06-21 meeting
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).
A
A
B
So
I
quickly
looked
at
what
issues
we
have
and
what
new
pull
requests
there
isn't
much.
The
there
was
an
issue
reported
against
api
only
build,
and
I
think
it
was
caused
by
one
of
my
changes
related
to
and
longman
jason.
I
already
fixed
it
in
the
pr
attachment
for
it.
B
I
refreshed
I
refreshed
this
so
that
if
we
build
with
api
without
sdk,
so
then
then
I
do
not
require
analogman
json,
because
we
don't
include
it
anywhere
in
the
api
projects
that,
if
built
with
api
specified
api,
only
is
specified,
we
don't
include
any
of
the
sdk
pieces
so
that
that
should
address
the
the
issue
that
the
person
recorded.
I
think
that
was
the
latest
comment
from
lalit.
B
He
asked
me
if,
if
we
should
be
fixing,
I
fixed
it
in
the
latest
update
since
he
refreshed
I
added
another
condition
around
it.
So
if
api,
only
if,
if
not
api
only
then
require
json
else,
don't
require
it.
B
We
we
don't
include
the
json
library,
yes,
because
none
of
the
public
headers
require
it.
We
should
be
safe.
B
And
I
also
look
at
the
initial
three
days
ago
about
resource
support,
initiate
w
exporter.
I
don't
have
that
done
yet
I'll
send
apr.
This
was
related
to
one
of
our
customers
asking
for
a
common
set
of
four
attributes
so
that
they
don't
have
to
pass
it
every
single
time
in
the
attributes.
Collection
I'll,
add
that
proper
resource
support
so
that
they
can
reset
resource
and
aggregate
things.
B
That's
right:
do
you
progress
right
pr?
It's
working
progress.
I
looked
an
issue
it
to
myself.
I
see
that's
pretty
much.
It
there's
one
issue
from
five
days
ago
reported
by
chris.
I
don't
know
what
project
he's
working
on,
but
neither
me
nor
lead
have
this
reproduced
and
we
run
like
different
setups.
I
think
little
late
runs
a
physical
box,
I'm
running
on
wsl
same
linux
version,
which.
B
B
B
A
B
But
I
don't
know,
I
think,
the
person's
not
blocked
right
now,
because
I
have
seen
another
references
like
to
them
doing
some
engine
stuff
on
their
box,
so
they're
not
blocked
per
se,
but
it's
an
interesting
one.
I
I
don't
really
have
a
clue
like
why
it's
happening
on
that
machine.
That
way,
mine
is
a
running
point.
B
B
For
the
fourth
one
I
reopened
it,
I
the
the
pr
that
I
have
should
fix
that
and
I
also
included
the
fix
for
that.
The
sim
make
issue
for
the
first
one
and
the
list,
so
they
both
will
be
closed
and
one
can
be.
B
The
funny
thing
is
that
in
the
hd
double
exporter,
when
some
refactoring
was
done
for
the
target,
I
think
wendell
did
the
proper
way
of
discovering
in
lokman
jason,
because
the
tw
exporter
uses
it
in
one
of
his
head
and
it's
like
in
a
public
head.
So
then
cmake
doesn't
let
you
build
if
you
don't
also
publish
and
log
on
as
a
result
of
your
bill.
So
it's
like
you
cannot
keep
a
private
target.
B
B
It
detects
that
if
it's
not
a
in
the
list
of
exported
targets
as
a
package,
it
won't
let
you
build
so
even
if
they
had
that
is
in
the
past
at
the
time
of
the
bill,
it
still
wants
it
to
be
actually
like
installed
as
a
as
an
installable
package,
because
in
the
target
exporter
it
says
interface
and
longman
json
and
the
list
of
exported
libraries,
which
is
true
because
it's
used
in
the
public
headers.
So
I
I
fixed
that-
and
I
hope
that
one
commit
is
gonna
close.
B
The
issue
number
one
on
the
list.
869.
and
the
issue
number
four
on
the
list:
843
properly
fixed
this
time.
Okay,
so
I
I
I
when
I
was
testing
it,
I
tested
the
linux.
Linux
did
not
have
a
tw
target,
that's
why,
for
linux
it
was
all
all
fine
but
yeah
and
on
windows
I
use
dc
package
usually
and
the
vc
package
already
installs
the
the
package
in
the
cmake.
B
So
we
didn't
go
through
that
else,
if
not
found
to
use
local
and
that's
why
all
my
bill
passed
in
our
ci,
we
also
install
it
externally.
Still,
yes,
that's.
A
B
A
B
That
case,
it's
gonna
use
the
one
that
comes
from
the
sub
module.
Yes,
I
think
that's
probably
yeah
yeah
would
be
helpful.
I
don't
have
much
to
discuss
for
today.
I'm
still
trying
to
finish
some
of
the
fluent
export
of
stuff
right
now
and
I'll
have
more.
You
know
to
review.
B
A
B
B
B
Let
me
see
if
I
had
this.
Actually
this
is
the
one.
So
this
is
the
one
that
I
opened
13
days
ago,
buffer
test
simulation,
so
we
had
that
reproducing
once
in
the
let's
see
I
run
four
days
ago,
which
means
that
it
does
happen.
It
doesn't
happen
all
the
time
because
most
of
our
ci's
look
clean.
A
B
B
I
remember
in
my
previous
career
I
had
very
tricky
way
of
like
there's
that
linux
facility
groups,
which
allows
to
reserve
execution
to
only
given
set
of
resources.
B
I
don't
know,
I
think.
B
Tell
because
it
may
depend
how
it
gets
scheduled
on
what
machine
it
may
get
scheduled.
I
think
we
may
need
to
try
both
like
forcing
it
on
as
if
it's
a
single
core
machine
and
maybe
multiple
reruns,
just
that
this
see
if
we
can
hit
it,
because
I
think
I
don't
see
it
because
in
my
case
I
always
run
multi-core.
B
A
B
So
I
think
we
went
through
the
issues
before
the
prs.
B
Yes,
I
think
so
so
for
the
first
one
I
addressed
the
last
comment
from
wally
I'll,
let
lilly
to
take
a
look
if
it's
good
when
he
comes
online.
I
guess
I.
B
A
B
Okay
cool,
so
I
actually
have
a
confirmation
from
david
lee,
the
guy
who
reported
that
other
issue
was
the
api
that
built
that
the
way
I'm
fixing
it
is
fixing.
Thank
you
yeah,
so
he.
B
Be
nice
if
we
could
still
build
the
rest
of
the
labyrinth
without
this
dependency,
yeah
yeah,
so
my
comment
to
that
would
be
in
most
configurations
in
most
actually
like
for
z
pages
for
hcw,
for
zip
cam,
for
which
is
the
required
exporter
by
spec
and
then
in
a
few.
Are
their
sports
I'd
say
even
for
otl
p,
http
exporter?
So
it's
like
80
of
standard
modules
require
this
component.
B
Yes,
I
can
do
the
build
tweaks
to
make
it
so
that
if
it's
not
there
try
to
skip
the
build
of
those
eighty
percent.
But
my
point
is,
since
we
also
kinda,
always
include
it
in
the
sub
module
and
it's
technically
always
available
this
or
the
other
way,
I'm
doing
check
if
it's
already
installed
somewhere
else
through
cmac,
like
external
modules
with
analog
manager,
and
I
use
that,
if
not,
I
fall
back
to
the
one
that
is
in
our
repo,
the
sub
module
and.
A
B
Somebody
cloned
without
submodule
I'll,
give
an
error
saying:
hey
you
either
install
external
work
alone
with
submodule,
so
I
give
a
descriptive
verb
either
way.
There
is
a
way
out
of
that
state,
because
user
gets
a
descriptive
explanation
of
what
is
going
on
wrong
and
the
my
logical
reasoning
is
that
since
48
of
scenarios
for
80
percent
of
sdk,
we
actually
require
this
I'd
rather
not
make
it
more
complicated.
I
think
the
only
case
when
we
don't
require
is
if
we
only
build
a
otl,
jrpc
exporter,
okay
and
without
the
pages.
B
A
But
I
think
the
otp
exposure
will
will
be
the
major
component
correctly
or
tlp.
Http
requires
it.
Okay,
so.
B
If
we
say
if
we
build
our
top
http
yeah,
we
still
need
an
aluminum
json.
It
still
requires
it.
So
my
logical
reasoning
is
that
in
most
cases
not
in
all,
but
in
majority
of
cases
it
is
required.
So
I'd
rather
keep
that
the
descriptive
information-
and
it's
not
a
big
overhead.
Let's
say
if
you
build
just
a
otl,
pg
or
pc
as
long
as
we
recursively
clone
the
ripple
and
we
build
with
our
grpc
and
without
the
rest,
the
cmake
is
gonna
check.
B
But
the
funny
thing
is,
I
tried
to
look
at
how
we
consume
it,
and
this
is
a
bit
inconsistent
in
some
places.
We
just
assume
that
it
is
in
the
path,
so
we
just
do
include
no
command
json,
but
in
the
other
cases
we
do
target
link
library
and
then
we
do
either
interface
or
private
and.
A
B
Add
it
to
the
path
it
is
header,
only
library,
but
the
way
we
consume
it
across
different
places.
Sometimes
we
just
assume
somehow
it's
magical
in
the
path
and
in
other
places,
places
we
explicitly
do
target
link
library,
so
I
think
best
would
be
is
always
do
target
link
library.
I
think
that's
a
new
way.
B
A
B
B
The
one
that
this
keeps
the
install
yes,
it
is
actually
there
like
that
variable
if
you
set
a
tool,
I'm
just
thinking
that
I
did
add
the
check
that
it's
like.
If
there's
no
json
installed
for
back
to
sub
module,
then
I
added
the
check
is
submodule
there
like
cloned
recursively
yeah,
so
that's
here
and
if
I
didn't
find
it
I'd,
give
an
error
and
right
now
it's
showing
as
fatal
here
I
can
change
it
to
warning
or
like
see.
B
The
guy
david
actually
confirms
that
he
understands
that
it
gets
complicated
if
we
are
to
handle
this
case,
like
200
percent
of
I'd,
rather
not
right
now,
because
I'm
making
it
easier
for
the
most
and
not
making
it
harder
for
the
rest.
B
B
We
could
have
made
it
a
bit
smarter,
but
I
need
to
invest
a
bit
more
time
and
I'd
rather
treat
it
as
a
lure
priority
issue,
because
it's
more
of
an
optimization
how
to
set
up
a
build
pipeline
without
it
completely,
and
I
already
fixed
it
for
the
api
only
bill.
So
david
scares
has
been
addressed
and
if
david
builds
api,
only
I
don't
require
and
lockman
json
anymore.
A
A
B
The
majority
of
cases
it
is
required
the
way
how
I've
done
it.
I'd
rather
keep
it
than
not
improving
here
and
I'll.
Think
a
bit
more
about
how
to
make
it
better
I'll
ask
him
if
he
wants
to
make
it
really
better,
we
can
trace
it
track
it
in
the
separation.
B
Right,
that's
about
it.
I
I
think
what
else
so
for
the
semantic
convention
attribute
pr
my
feedback
right
now
is.
We
could
have
done
it
a
bit
better
in
terms
of
our
compile
time.
Optimization
is
right.
Now
we
generate
these
getters
and
we
we
wrap
the
string
values
into
string
view
class
every
time,
which
means
that
every
time
we
populate
a
dictionary
we
actually
need
to
run.
B
I
don't
think
compiler
is
optimizing
it
efficiently
that
that
way,
I
my
feedback
is
if
we
could
use
a
constant
expression
and
just
make
sure
that
this
field
name
gets
substituted
at
compile
time,
that'd
be
awesome
and
I
kind
of
agree.
B
A
B
As
all
of
these
standard
name
standard
attribute
names
are
described
in
some
yaml
file
or
elsewhere.
We
can
just
generate
it
from
there.
A
B
So
he
actually
said
that
he's
going
to
try
that
I
I
have
a
blind
spot
in
there.
I
know
that,
for
the
string
pulling
for
the
string
pulling
msvc
has
a
compiler
option
that
allows
to
aggregate
all
strings
together.
So
even
if
we
have
these
as
constants
for
constant
charstar,
a
compiler
is
actually
going
to
use
and
aggregate
them
in
the
same
spot
so
that
it's
not
like
we
duplicate
them
all
over
the
map.
B
B
I
don't
think
that
language
guarantees
that
what's
the
option
you
have
to
please
shut
up
yeah
option
for
the
msvc
just
a
moment.
I'll
mention
that
in
the
comment-
okay,
I
I
mentioned
it
in
a
comment.
It's
shrink
pool
in
just
a
moment:
element
duplicate,
string,
gf.
A
B
For
the
msvc-
and
I
think
by
default
I
I
can
double
check
on
that.
I
don't
have
a
confirmation,
but
my
understanding
is
that
by
default,
gcc
and
clank
would
do
it
for
you
without
that,
like
without
any,
but.
A
This
this
feature
means
to
merge
different,
different
currents
of
the
same
constant
right
or
string
constant,
but
in
the
current
table,
since
there's
no
duplication
right
or
what.
A
B
B
Okay,
and
I
think
if
I,
if
I
use
like
50
of
them,
I
will
be
bloating
my
code
unnecessarily
as
well,
and
I
think
that
option
allows
to
reuse
that,
because
they
are
postal
and
they
have
the
same
value,
they
are
not
modifiable,
they
can
be
grouped
all
together.
B
So
it's
like,
I
think,
it's
linker
time
they
would
all
be.
B
B
Let
me
do
just
that,
because
I
think
that,
in
terms
of
perf
anyways
grabbing
an
address
with
this
with
this
non-modifiable
or
string,
it's
going
to
be
a
bit
more
efficient
than
calling
a
method
that
grabs
an
address.
Perhaps
in
another
object.
Yes,.
A
B
B
So
it's
like
gonna
be
like
if
we
populate
the
map,
unless
we
also
recommend
that
for
any
map
like
this,
we
initialize
the
map
once
and
reuse
it
many
times
in
that
case.
That
won't
be
that
noticeable
at
all,
like
a
static
map
with
values
in
it
once
reused
many
times
and
then
the
individual
course
to
which
member
initialization
is
not
going
to
matter
much
right.
B
Okay,
yes,
so
it
depends
again,
but
it's
also
depending
on
how
users
use
it,
because
we
cannot
again
force
everybody
to
use
that
practice
where
they
initialize
the
map
once
they
may
have
different
modules,
and
then
it
matters
like
maybe
they
should
profile.
B
You
know,
let
me
take
a
look
because
I
I
I
kind
of
have
some
ideas
about
how
to
make
it
a
bit
better
and
maybe
comparing
that
is
this
significant
that
we
wrapped
it
into
the
getters,
how
much
overhead
we
get
because
of
that
or
if
it's
not,
that
significant
get
another
indirection
pretty
much.
B
We
can
also
use
extern,
I
think,
if
we
use
extern
and
if
in
open,
chalametric
column
common,
we
use
those
constants
in
a
compilation
unit,
then
in
all
four
sports.
It's
gonna
resolve
to
the
same
address
like
in
one
like
manually,
pulling
them
all
into
one
single
compilation
unit.
B
B
Yeah
anyways,
I
think
lalit
is
going
to
take
a
look.
Yes,
I
think
so.
B
B
Else,
if
you
have
time,
take
a
look
at
my
build
farm
here,
it's
mostly.
I
had
that
client
support
in
there
and
I
renamed
a
few
things
in
the
batch
file
because
it
was
all
referring
to
a
visual
studio,
but
the
clients.
B
Not
from
mimics,
okay,
yes,
build
form
for
windows,
it's
just
a
built
script,
so
I
made
it
so
that
we
can
actually
run
this
thing
with
any
compiler,
15,
17
or
19
or
clang
clang,
like
a
standalone
plank.
I
don't
even
have
to
have
ins
and
visual
studio
install
which.
B
Client
village,
I
tried
with
10
and
11.
and
one
not
one
thing.
I
noticed
when
you
install
font.
It's
just
suggests
to
install
it
all
with
the
same
directory
program
files
below
vm.
I
think,
and
from
that
you
cannot
really
tell
like
you.
You
don't
have
standard
paths
for
side-by-side
installation.
B
It's
like
visual
studio,
always
have
a
standard
path
for
15
for
17
and
490,
and
you
can
kind
of
sort
out
and
have
all
of
them
installed
at
once
and
for
clang.
It
always
dumps
into
same
folder
by
default.
Okay,
you
can
still
force
it.
You
can
say
install
to
adjust
the
default
installation
path,
say
clank
dash
nine.
Then
you
install
it
again.
It's
gonna
say:
oh,
I
I
already
see
that
you
have
clang
lavien
installed,
say
no.
B
I
I
still
want
to
install
again
and
you
specify
a
different
path
and
it
star
installs
again
into
a
different
path.
That's
way
that
way,
you
can
have
two
side-by-side
plugs
installed
on
one
machine,
but
it's
not
it's
not
the
the
default
way
of
doing
it.
That's
why,
for
clang
I'd
say
any
clan,
but
I
only
tested
10
and
11..
So.
B
No,
no,
no,
it's
just
the
usual
plan
and
the
way
how
it
works
is
I'm
building
now
with
this
ninja
build
tool,
2,
and
it
actually
discovers
that
as
long
as
clang
compiler
is
in
the
path
it
detects,
the
supply
plus
compiler
so
and
it
doesn't
require
a
clank
cl.
B
A
B
Bunch
of
errors
actually,
like
warnings,
we
have
quite
a
bit
of
warnings,
but
it
all
compiles
and
works
so,
which
means
that
we
don't
necessarily
have
to
force
the
users
to
use
visual
studio
on
windows.
If
some
projects
like
chromium,
for
example,
they
use
their
own
clank
compiler
so
far,
so
good,
it
works
and.
B
If
I
use
llvm
like
yeah,.
B
A
B
Why
I
renamed
a
few
things?
They
were
called
like
some
variables.
We
were
called
vs
underscore
tools
version,
because
I
I
assumed
before
that
it's
only
visual
studio,
but
now.
A
B
Added
any
compiler
included
non-microsoft,
I
just
renamed
it
to
build
tools
version
without
using
vs
printers.
Yes,
okay,
that's
about
it!
Anyways.