►
From YouTube: 2023-03-07 Frontend Create:Source Code Weekly
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
In
real
life
welcome
everybody,
this
is
the
first
edition
of
the
front
end
teams
weekly
for
create
source
code
or
the
front-end
create
Source
called
weekly.
Whichever
way
you
want
to
say
that
it's
a
new
call-
and
the
goal
here
is
to
discuss
topics
that
are
currently
in
our
minds.
Any
blockers,
crazy
ideas
concerns.
A
It
also
allows
us
time
this
time
to
use
this
use
it
for
retro
review,
kickoffs
and
with
time.
We
also
want
to
encourage
and
promote
demoing
in-flight
work
that
you
have
so
demoing
Mrs
things
that
you
might
want
to
show
to
the
team
so
definitely
worthwhile
bringing
up
here
in
this
call.
A
B
Yeah,
so
the
first
one
is,
rather
than
like
a
small
fan
exercise.
So
in
this
issue,
which
is
linked
in
the
agenda,
Sean
raised
an
interesting
point.
So
whenever
you,
whenever
user,
tries
to
clone
the
project,
they
go
to
this
clone
button
and
then
they
can
pick,
for
example,
to
clone
into
the
IntelliJ
IntelliJ
IDEA
or
to
xcode,
and
the
problem
is
that
we
do
not
detect
whether
any
of
this
are
installed
on
the
user
on
the
client's
machine.
B
So
what
this
leads
to
is
that
JavaScript
silently
fails
and
throws
and
console
without
giving
any
any
notification
to
the
user.
The
tricky
part
here
is
that,
like
speaking
front
and
language,
we
could
get
to
the
conclusion
that
okay,
it
throws
so
we
have
to
use
the
try
and
catch.
No.
It
doesn't
work
with
this
type
of
things,
because
this
is
not
called
by
try
catch.
B
Unfortunately,
so
it
just
throws
I
created
the
non-working
draft
in
terms
it
does
solve
the
problem
of
the
of
the
non-installed
applications,
but
it
messes
up
for
the
scenario
where
the
application
is
actually
installed.
So
it's
like
fixing
One
Thing
by
breaking
another
one.
So
General
front-end
approach,
it
would
be
great
to
to
figure
out
and
I
was
I
was
actually
stuck
with
this.
B
With
this
Tusk,
it
was
really
interesting
to
figure
out
how
actually
we
can
do
this
on
the
front
end
whether
we
can
detect
things
like
this
somehow,
because
in
this
particular
case
it
is
the
front
and
task
notify
the
user
that
the
application
is
not
installed
to
either
show
the
the
alert.
B
So
what
I
did
in
my
non-working
draft
is
I
show
the
alert
with
the
link
to
download
the
application
if
user
wants
to
to
use
that
application
for
their
needs,
but
this
sort
of
interferes
with
with
the
case
when
the
application
is
actually
installed.
So
Jacques,
you
have.
You
have
the
next
sort
of
the
first
discussion
Point
here.
C
Sorry
yeah
I
was
wondering
if
it's
possible
to
Ping
these
links
or
endpoints
like
in
a
normal
HTTP
protocol.
It
returns
like
status
codes,
I,
don't
know
if,
if
it's
the
same
with
these.
B
Paying
like
like
like
like
like
what,
like
just
ping,
the
problem
is
that
just
this
is
not.
This
is
not
the
HTTP.
This
is
not
an
HTTP
link
right.
This
is
this
is
completely
different.
Protocol.
A
Okay
yeah.
This
reminds
me
of
some
I
remember
a
couple
of
times
a
couple
years
ago.
There
was
this
huge
I
mean
flaw
in
browsers
where,
through
the
visited
state
of
the
link,
we
could
violate
the
privacy
of
users
and
we
can
guess
whether
they
had
been
to
certain
websites.
A
So
I
think
browsers
make
it
kind
of
hard
to
kind
of
like
Get
Smart
about
this.
But
did
your
working
non-working
draft
tennis
get
anywhere
closer
to
that
to
understanding
whether
it's
installed
or
not?
So.
B
The
let
me
just
share
my
screen
for
a
second:
do
you
see
my
screen?
Now?
Yes,
okay!
So
let's,
let's
get
to
this
to
this
draft.
So
there
are
only
two
changes
surprisingly,
but
they
are
pretty
pretty
interesting,
so
the
one
of
those
the
constants.
Let's,
let's
increase
the
font
size
again,
I
guess
it's
gonna
work
out,
so
the
I
just
have
the
constant
for
the
for
the
environments.
B
So
what
is
coming
into
the
constants
is
the
just
the
sort
of
the
ID
the
protocol,
the
protocol
for
that
or
another
thing,
so
vs
code
jetbrains
and
the
xcode
the
title,
which
is
shown
in
the
drop
that
clone
drop
down
and
the
URL
to
download
the
latest
version
of
this
application.
If
user
clicks
the
application
is
not
found,
we
show
the
alert
suggesting
user
to
go
and
download
this
application.
If
that's
what
they
want.
B
So
at
least
they
have
some
some
sort
of
feedback
and
it
is
implemented
like
like
this,
so
we
try
so
we.
What
do
we
do?
Is
we
just
get
the
the
URL,
the
user,
from
the
link
user
clicked?
B
We
update
the
window
location
and
here
comes
the
tricky
part
we
set
the
time
out
for
the
seconds.
I
know
this
is
lime,
it's
just
the
proof
of
concept,
so
we
in
a
second
we
try
to
check
whether
we
actually
have
sort
of
we.
We
show
the
alert
with
with
the
download
link
to
the
to
that
application,
but
the
problem
is
that
in
this,
in
this
particular
case,
we
are
showing
the
alerts
still
even
if
the
application
is
opened.
B
So
this
alert
is
staying
on
the
screen,
even
if
the
application
is
is
available
and
the
users
are
already
in
it
and
I
was
trying
to
figure
out
to
to
figure
out
how
to
detect
whether
the
user
has
been
getting
to
the
to
the
to
another
application
or
something
so
I
tried
to
catch
the
things
on
before
unload,
because
technically,
when
the
that
pop-up
or
alert
is
shown
whether
you
want
to
open
the
the
xcode
application,
for
example,
the
on
before
and
load
event
is
fired.
B
That
thing,
but
again
it
is
fired
before
that
alert
is
shown.
So
we
have
no
idea
whether
alert
is
shown
or
not,
and
that
event
is
fired,
no
matter
whether
user
has
the
application
installed
or
not
because
otherwise,
either
user
goes
to
to
the
application
or
to
the
or
or
they
don't
go
anywhere.
D
B
Anyway,
this
this
proof
of
concept
shows
the
alert
provides
user
with
ability
to
download
the
application,
but
it
still
shows
the
alert,
even
if
the
user
has
the
application
installed,
because
I
didn't
find
the
way
of
saying
okay,
user
has
moved
to
the
application
or
user
has
installed
the
application
installed.
So
it's
it's.
We
don't
need
to
solve
it
now.
B
It's
just
something
interesting
to
to
play
with
and
think
about
how
to
how
to
detect
a
really
under
things
undetectable
on
the
front
end,
because
front-end
by
design
has
no
access
to
the
to
the
system
right
to
know
what
applications
are
installed
or
not.
A
The
census
for
it
yeah
so
I,
put
in
the
agenda
link
that
I
found
on
Google,
but
there's
there's
a
apparently
a
library,
you're
still
sharing
the
screen
by
the
way
Dennis
yeah
they're
right,
yeah
that
he
they
say
that
it
can
detect
the
presence
of
the
custom
protocol.
Handler
I'm,
not
sure
if
this
will
cause
an
issue
like
because
some
of
the
times
what
they
do
is
they
try
to
open
it,
and
this
would
eventually
result
in
the
browser
trying
to
open
the
xcode.
So
this.
D
B
When
I'm
very
kind
of
frightened,
because
if
in
order
to
detect
whether
an
application
is
installed
or
not,
we
have
to
pull
in
another
package.
This
is
where,
like.
B
A
This
works,
we
know
it's
possible
and
then
we
can
reverse
engineer
that
a
little
bit
yeah
trying
to
see
exactly
how
they're
doing
that
and
write
our
own
code,
but
but
yes,
I,
wouldn't
want
to
integrate.
Yet
another
package,
yeah
yeah
yeah.
B
A
A
B
D
A
B
Big
problem,
if
we
would,
if
it
would
somehow
notify
user
by
by
any
you
know
any
other
mean
other
than
throwing
into
the
console,
but
that's
the
problem.
A
Yeah
I
think
at
the
time
what
we
decided
was
any
so
from
the
perspective
of
like
stepping
out
of
the
of
the
UI
thing,
but
more
into
the
user.
Experience
of
things
is
you're
clicking
to
opening
the
xcode
a
project
that
is
Apple
based.
You
yourself,
the
user
is
very
likely
somebody
who's
used
to
this
sort
of
stuff.
B
Yes,
yes,
the
next
one,
so
this
one
is
much
more
in
line
with
our
work.
Really
I've
created
this
issue
about
the
the
view
applications
we
have
on
the
repository
View.
B
At
the
moment
we
have
tan
unnamed
view
applications
together
with
three
or
like
two
three,
depending
on
what
feature
Flags
you
have
enabled.
B
So
so
we
in
total,
it's
about
up
to
13
14,
feature
view
applications
on
the
repost
review.
So
surprisingly
enough,
so
if
you
take
a
look
at
the
at
the
issue
at
the
link,
some
of
the
applications
are
not
even
used
on
the
Repose
review.
They
are
meant
to
be
to
be
enabled
only
on
the
blob
View.
B
So
some
of
those
are
instantly
like
I
I
have
I,
have
included
the
screenshot,
where
all
the
active
applications
are
highlighted
and
clearly
the
the
rough
selector
and
the
the
project
selector.
With
new
new
new
file
new
product,
a
new
folder.
They
they
reside
together,
so
they
live
next
to
each
other,
so
they
can
be
merged
into
one
View
application.
B
It's
a
bit
more
complex
with
other
other
things,
but
at
least
we
can
start
with
cleaning
up
and
not
bootstrapping
the
applications
which
are
not
needed
for
the
Repose
review,
because,
again,
those
are
not
not
named
and
for
some
of
the
applications
it
took
me
took
me
a
while
to
figure
out
what
those
applications
are,
because
it's
just
app
and
there
is
no
child
in
that
application.
B
So
it's
kind
of
hard
to
understand
what
what
application
that
is,
but
I
think
I
have
identified
all
of
those
and
it
would
be
great
to
clean
up
a
bit.
A
Thanks
my
feedback
here,
yeah
epic,
sounds
great
to
track
this
effort.
I
guess
one
of
the
things
that
will
be
interesting
to
understand
is
exactly
like
what
would
be
the
benefits
of
performance,
and
one
of
the
things
that
might
not
be
very
apparent
here
is
the
the
page
that
you're
in
is
the
same
view
app
that
is
going
to
render
the
blog
right.
So
it's
still
the
same.
A
So
could
it
be
relevant
to
reload
the
app
that
is
going
to
be
probably
eventually
needed
once
we
get
to
the
blog,
having
that
user
Journey
being
very
likely,
so
there's
a
trade-off
there
between
you
know
reloading
the
app
so
that
it's
ready
right
away
when
the
glove
is
rendered
so
to
make
that
faster,
so
that'll
be
a
nice
conversation
to
having
the
Epic
exactly
the
benefits.
What
are
the
most
beneficial
apps
to
integrate
into
that
same
app,
but
we
can
have
that
discussion
of
synchronously.
B
The
the
one
obvious
obvious
reason
to
to
remove
the
applications
that
are
not
used
is
that
it
will
make
working
with
this
video
a
bit
easier
because,
like
when
you
look
at
this
list
of
app
no
named
apps
in
the.
If
you,
when
you
go
to
the
view
tab
of
the
dev
tools,
it's
kind
of
hard
to
realize.
What
is
this,
especially
when
you
hover
over
the
supplication
and
no
element
is
highlighted.
So
it
makes
things
a
bit
harder
to
understand
and
to
figure
out
what
element
to
to
get
to.
B
But
there
are
some
things
that
we
can
just
convert
to:
The
View
application
in
in
the
same
effort
or
in
additional
effort
and
yeah
I
will
convert
this
into
the
Epic
and
split
into
like
brainstorm.
The
ideas
of
how
we
can
combine
these
things
and
combine
and
split.
A
Yep,
please
bring
the
issue
we
have
for
the
find
file
app
so
for
bringing
the
find
file
app
into
the
repository
app.
Do
you
have
that
issue?
I
can
find
it
for
you,
so
this
was
a
Remnant
from
The
Blob
refactor
of
like
making
a
blob,
refactor
The
Blob
view
part
of
the
repository
view
app.
A
One
of
the
next
things
we
would
like
to
absorb
was
the
find
file
right
because
it
just
makes
that
switching
between
find
file
and
the
projectory
and
the
blog
view,
all
of
that
much
faster
between
like
contact
switching
and
it's
a
very
common
thing.
Like
you
find
a
file,
you
go
to
the
blob
view
right,
you're,
watching
the
file
repository
you're,
finding
something
you
can't
find
it.
You
go
to
the
find
file.
So
that's
probably
got
a
very
obvious
one.
A
So
I'm
going
to
put
the
link
to
issue
here
later,
but
please
bring
that
over
to
that
epic
when
we're
the
ones.
Okay,
great.
Thank
you
Natalie.
You
have
the
next
one
go
ahead.
D
Yeah
all
right,
so
this
is
something
I've
been
recently
working
on
and
I
guess:
Igor
has
the
most
context
of
it.
So
when
I
do
call
the
pork
steam
commutation,
sometimes
like
once,
this
was
taken
really
a
lot
of
time
for
this
process
happening
on
back
end.
So,
like
one
hour
or
something
and
I
don't
know,
I
actually
I
can
demo
what
I
have
now.
If
you
guys
are
interested
but
yeah
I
was
just
the
question
I
have.
Maybe
I
should
stop.
Do
HTTP
polling
at
some
point
if
it
takes
too
long.
E
Yeah
one
hour:
it's
it's
really
long
time.
I
guess
maybe
maybe
I.
E
E
It
was
just
yeah,
no
error,
keep
loading
yeah,
it's
just
like
an
internal
error
that
happens
in
the
background.
It's
just
a
guess
and
regarding
stopping
the
pollen,
it's
actually
makes
sense
to
me,
but
I
guess
currently
on
front
end.
We
have
Pauline
implemented,
which
we
just
which
is
just
pulling
every
second
or
every
two
seconds,
yeah
like
every
period,
but
maybe
it's
possible
to
have
like
a
gradual
incremental
period.
E
E
Like
this
yeah,
but
it
definitely
makes
sense
to
stop
bowling
yeah.
B
I'm
wondering
what
we
have
you
here:
can
it
be
that
actually
the
polling
makes
things
makes
special
things
worse,
because
we
by
constantly
pulling
we
are
overloading
the
the
back
end
with
this.
E
Yeah,
it's
it's
not
like
the
pre-test
solution,
yeah,
but
with
current
implementation.
What
do
we
have?
We
have
a
background
job
and
which
and
on
front
end.
We
can't
get
the
status
of
whether
this
background
job
finished
or
not.
Until
we
explicitly
ask
so
okay,
one
of
the
solutions
is
a
websocket,
for
example,
here
to
have
a
websocket
connection
and
receive
a
signal.
E
Another
solution
is
to
constantly
pull
the
endpoint
and
that
particular
endpoint
that
we
are
working
with
Natalia
is,
is
being
cached
and
it's
quite
like
Fast
doesn't
like
I
think
so
yeah
it
doesn't
like
produce
a
huge
load
but
yeah.
It's
it's
a
good
question
and
this
on
this
situation.
We
have
it
like
this.
A
So
Igor,
one
of
the
contexts
I
want
to
bring
up
here,
is
in
the
code
review
group.
We've
been
dabbling
with
moving
things
from
spooling
and
smart,
pulling
on
the
front
end
into
graphql
subscription
and
there
the
problem
is
trickier
because
we
have
to
migrate
things
to
graphql
and
then
graph
class
subscriptions.
A
Maybe
we
can
create
an
issue
as
a
follow-up
to
move
this
check
to
a
graphql
subscription,
because
that
seems
like
the
best
of
both
worlds,
like
the
browser
doesn't
have
to
do
any
work
in
the
front,
the
front
that
is
just
notified
from
the
back
and
once
that
thing
is
completed
and
it
will
be
immediate
and
that
way
it's
just
the
better
experience
of
Ross.
So
does
that
make
sense
to
open
an
issue
for
that.
E
Yeah,
it
makes
sense
to
me
currently
we
we
implemented
it
using
graphql,
endpoints
action
points.
Action
item
for
me
is
to
read
about
graphql
subscriptions
because
I
have
to
admit
amazing
indicated
on
this
topic
here.
A
So
it's
amazing
there's
a
I
can
put
a
link
here
to
the
efforts
we're
doing
on
on
there,
because
it's
usually
actually,
since
we
haven't
how
much
time
do
we
have
oh
three
minutes:
okay,
so
I'm
going
to
link
to
an
epic
where
we
have
both
the
backend
and
the
front-end
work
for
graphql
subscription.
A
So
essentially
the
back
end
adds
like
a
trigger
to
where
the
work
is
done
or
where
you
don't
want
to
notify
it
and
the
front
end
just
needs
to
instead
of
doing
a
query
or
invitation
is
to
do
a
subscription
to
that.
So
take
a
look
at
that
those
examples,
but
please
who
wants
to
open
an
issue?
Please
claim
it
as
a
volunteer,
Igor
or
Natalia.
Please
one
of
you
tackle
that.
A
I
will
do
thanks
igori
two
minutes
to
go
through
that.
E
Yeah
I
I
will
be
quick,
so
I
can
share
my
screen.
So
let
me
so
we
have
like
I
created
this
issue,
yeah,
it's
related
to
code,
intelligence,
yeah,
and
currently
it's
sometimes
it
doesn't
work
correctly,
for
example,
for
example,
for
cases
like
this,
we
have
gitlab
shell
project.
We
have
a
file
and
like,
for
example,
this
line
of
code
yeah.
E
It
should
like
display
a
pop-up
for
every
like
every
word
here
for
every
variable,
but
and
it
worked
before
highlight.js,
because
we
wrapped
every
like
lexam
in
a
Spam
yeah,
and
that's
why
it
was
possible
to
implement
a
logic
like
like
this.
That,
for
example,
starchar
is
full,
and
it's
actually
meant
like
the
fourth
Span
in
the
in
the
div,
but
with
highlight.js.
Sometimes
it
happens
that
it
it
wraps,
like
the
whole,
the
whole
world
in
a
single
span,
yeah
and
yeah.
E
It
actually
makes
sense
from
performance
perspective,
but
it's
actually
now
doesn't
work
correctly,
sometimes
so
yeah.
My
my
question
was
whether
we
can
have
a
look
at
this
issue
and
yeah
particularly.
Is
it
possible
to
wrap
every
word
and
spend
or
do
we
have
concerns
from
performance
perspective.
C
But
we
can
definitely
take
a
look.
I'm,
not
sure
I
haven't
actually
had
a
look
at
that
bug.
Yet,
like
I
think
we
don't
really
have
control
over
how
how
highlighters
wraps
these
words.
The
only
thing
we
can
do
is
like
create
our
own
plugin
to
individually
about
to
wrap
the
words
but,
like
I
said,
it'll
have
a
big
performance
impact
Zach.
B
What
I,
I
I
I
know
what
you're
talking
about
with
the
with
the
with
the
performance
here,
but
what?
If
we
would
do
this
asynchronously
after
the
blob
is
rendered?
B
We
do
go
through
the
through
the
rendered
part
of
The
Blob
and
just
drop
that
afterwards,
so
it
won't
won't
affect
the
loading
performance.
It
might
affect
technically
the
the
time
like
the
time
to
interactive.
B
C
Yeah
I
think
that's.
We
can
definitely
take
a
look
at
that.
So,
after
everything
is
loaded,
we
go
through
the
page
and,
like
just
read
all
the
words
in
an
individual
span
like
certainly
something
we
can
look
at.
The
only
thing
is
like
doing
that
in
a
performant
way
like
let's
say
we
viewing
a
massive
file
like
how
do
we
do
that?
Performantly
I
guess,
there's
ways
around
it
like
only
wrapping,
what's
currently
visible
to
the
user.
A
Yeah
and
we're
working
to
to
improve
the
way
that
we're
doing
the
highlighting
my
question,
my
idea
is
probably
a
little
different,
so
this
is
for
the
code
intelligence
that
we
implemented
right,
there's
also
Source
graphs
code
intelligence.
A
If
we,
if
we
end
up
starting
to
mangle
the
market
too
much,
then
we
might
start
causing
problems
for
the
other
people,
the
other
tool
that
my
piggybacking
on
the
market.
What
I'm
wondering,
though,
you're
saying
that
the
back
end
Returns
the
following
points
for
code
units,
like
start
Char,
start
character
in
multiple
different
ways:
the
code-
that's
handling
that
so
the
code,
that's
handling
that
the
start
character
is
that
the
character
in
the
code-
or
you
said
it's
a
fourth
span-
is
the
character
in
the
code.
It's.
E
It's
a
character
in
the
code
yeah
and
like
a
tab
is,
is
counted
as
a
single
character,
yeah
character
and
yeah.
As
a
as
I
mentioned
in
the
description.
Maybe
to
implement
this
logic,
we
also
need
an
an
end.
End
character,
yeah,
like
Stark,
sure,
start
interval,
yeah
and
it
can
be
implemented.
Yeah
feel
free
to
mention
it
right.
A
So
what
I'm
doing
what
I'm
thinking
is?
We
do
have
control
on
the
code.
That
applies
that
to
the
front
end
right.
It's
our
implementation,
so
I
wonder
if
we
could,
when
needed,
when
it's
saying
to
insert
this
in
a
place
where
it
doesn't
have
a
span
around
it,
the
code
could
eject,
could
inject
in
insert
it
could
inject
it.
A
We
can
have
a
channel
in
the
issue
thanks
for
raising
visibility
and
that
we
already
have
a
time
but
I'll
I'll,
add
a
comment
there
and
see
what
we
can
do
there
and
then
we'll
see
when
it
comes
up
in
the
prioritization,
but
thanks
for
thanks
for
bringing
it
up,
Igor
we're
at
time
so
I'll
take
this
I'll.
Take
this
question
less
then,
to
the
slack
channel.
So
we
can
wrap
up
here
and
then
we
can
carry
on
just.
A
This
is
very
just
a
formal
thing
so
and
the
last
of
last
questions
just
was
this
call
worth
it.
Please
say
yes
or
no.
If
we
all
agree
that
it
wasn't
worth
it
we'll
stop
doing
it
so
for
now
just
answer
that
question
please
and
see
you
next
week.