►
From YouTube: Elsa Workflows Community Meeting 2 ( 2022-03-29 )
Description
**Apologies for another video with no audio from the audience -_- **
I assumed that as long as OBS Studio is recording my headphone audio, it would include audio from the audience. Well, my assumption made an ass of u & me ;-) Needless to say, but next time I'm gonna test with an actual participant to figure this out.
- Steering Committee Elections / Candidacy
- 7 candidates
- Next week: close candidacy and start voting process
- Everyone gets one vote?
- Someone to collect votes anonymously.
- Status
- Elsa 3
- Ability to display coded workflows in designer
- Ability to store position separately from C# workflows.
- ForEach: find ways to make the Body implementation visually make more sense.
- RunAt: consider replacing this with Cron activity?
- Demos
- Elsa 3 DSL
A
A
We
can
talk
about
elsa,
3
progress
and
about
the
steering
committee
elections
candidacy.
Let's
start
with
that,
christina
you
joined
chris
welcome.
So,
let's
see
if
this
works,
but
thank
you
for
joining.
So,
let's
see,
if
I
can
get
rid
of
this,
don't
know
how,
but
that's
okay.
So
as
I
was
saying
yeah,
I
wanted
to
look
at
the
steering
committee
candidates.
We
have
one
two
jay
mordali
mohammed.
I
think
yeah
mohammed
matthias
gurkhan,
stina
and
tommy,
also
known
as
toby.
So
that's
one
two,
three,
five:
six:
seven
candidates:
it's
pretty
awesome!
A
So
I
was
thinking
to
give
this
two
weeks,
so
we
started
the
election
the
candidacy
last
week.
So
this
is
then
now
you're
one
week
ahead
ahead.
So
next
meeting
we
will
see
how
many
candidates
we
have
and
then
maybe
that's
the
point
where
we
decide
what
the
election
procedure
would
be.
So
I
was
thinking.
A
Maybe
everyone
who
wants
to
vote
gets
one
vote
and
they
can
cast
their
votes
on
one
of
their
favorite
favorite
candidates,
and
then
this
selection
procedure
would
then
take
maybe
a
week
or
maybe
two
weeks
we
we
can
decide
on
that.
But
what
we
need
is
somebody
who
isn't
part
of
the
voting
process.
So
none
of
the
candidates,
but
somebody
from
the
community
that
is
objective
who
receives
the
votes,
because
I
think
we
will
want
people
to
be
able
to
vote
anonymously.
A
So
maybe
we
set
up
some
yeah
yeah
exactly
so
maybe
we
set
up
some
google
form
and
then
we
have
a
secretary
who
handles
that
that
and
then
and
then
we
count
the
votes
so
that
will
be
phase
one
of
the
election
and
there
will
be
a
second
phase
and
that's
between
the
five
last
therian
committee
members,
who
will
receive
the
most
votes
between
them.
We
should
then
elect
benevolent
dictator
and
this
again
that
this
will
be
up
to
the
community
members
who
want
to
vote.
They
can
choose
who
should
be
been
a
benevolent
dictator.
A
So
let's
say
next
week
close
candidacy.
Let's
start
voting
process.
I
just
received
a
message
from
gurkhan.
He
wants
to
join
us
while,
but
he's
still
on
the
phone,
all
right
so
status.
You
will
be
happy
to
see
that
I
have
a
couple
of
issues
created
for
lsat
3,
specifically
there's
a
new
projects
beta,
it's
been
out
for
a
while,
but
I'm
still
experimenting
with
it.
So
this
board
contains
some
sort
of
backlog
for
also
three
specific
items,
stuff
that
I
know
we
will
want
to
do
for
sure.
So
some
of
these
are
designer
specific.
A
Actually
most
of
them
are,
but
some
of
them
are
things
like
the
http
endpoint
activity.
We
have
it
in
lsat3,
but
it's
very
bare
bones.
For
example,
if
we
look
at
the
current
implementation
it
supports,
it
allows
the
user
to
select
the
supported
methods
to
check
that
yeah.
So
this
is
the
default
value,
but
it
can
support
multiple
and
I
think
the
middleware
for
the
method
right,
yeah
and
we
should
maybe
extend
this-
can
be
extended,
of
course,
and
we
could
even
consider
making
this
configurable
from
the
application
points
of
view.
A
Some
people
may
want
to
use
more
verbs.
Others
may
not
want
to
use
anything
at
all
or
allow
the
user
to
change
this
programmatically
like
we
can
do
in
lsat
too,
where
you
can
implement
a
handler
that
overwrites
the
property
descriptor
with
completely
custom
values
could
be
custom.
Ui
hands
could
be
different
options.
None
of
that
exists
in
lsr3,
but
that
would
be
a
feature
to
to
add
as
well
so
so
it's
one
example
that
needs
to
be
extended.
Another
thing
is
actually
reading
the
content.
A
I
don't
think
we
are
doing
it
right
now,
and
not
only
that
there's
also
the
so
it's
it's
the
reading
of
the
content.
That's
been
posted,
but
then
also
being
able
to
parse
it
into
json
objects
or
strongly
type.net
objects.
That
sort
of
thing.
So
that's,
basically
what
this
ticket
is
about
the
wait.
It
was
just
there.
A
A
So
right
now
you
can
manually,
drag
and
drop
the
activities,
but
maybe
you
want
to
have
a
if
you
make
a
mess,
then,
if
you
want
to
automatically
lay
it
out,
which
is
a
feature
we
need
in
any
case,
for
workflows
that
have
been
programmed
using
c
sharp
for
which
we
don't
have
positional
information
right.
We
don't
have
x
and
y
coordinates
when
you
create
something
in
c
sharp
like
this.
Take
one
example
right
so
here
we
want
the
workflow
to
be
rendered
in
the
designer,
so
it
should
have
a
tool
right
line
activities.
A
We
don't
have
any
positional
information,
so
we
should
render
these
two
nodes
on
the
on
the
display
and
then
some
right
now
it
doesn't
so
this
is
not
supported.
So
this
cannot
currently
be
loaded
onto
the
designer
and
the
way
this
works
in
lsat
2
is.
We
have
different
sources
of
workflow
blueprints
and
one
such
source
is
a
workflow
definition,
which
is
a
database
record
created
by
the
designer.
A
So
the
designer
invokes
an
api
endpoint
that
endpoint
creates
a
workflow
definition
based
on
the
posted
json
created
by
this
designer
and
then
in
order
to
make
it
available
to
the
workflow
engine.
These
workflow
definitions
are
transformed
into
a
thing
called
workflow
blueprints,
and
the
same
goes
for
these
sort
of
workflows
that
are
created
using
c
sharp
codes.
They
are
are
a
different
source
that
are
also
turned
into
a
workflow
blueprint.
So
you
have
two
kinds
of
source
models:
basically,
they
are
mapped
to
a
workflow
blueprint
and
that
gets
executed
by
the
engine.
A
This
mapping
from
code
to
a
workflow
blueprint
that
works
in
ls2
and
lc3.
It
works
a
little
bit
differently
here
we
are
the
engine
works
with
I
activities
and,
as
you
can
see,
this
bit
of
c
sharp
code
creates
this
structure
directly,
so
there's
no
additional
mapping,
but
still
in
order
to
make
it
available
to
the
designer.
We
need
to
expose
some
endpoint
that
can
transform
such
a
workflow
into
a
bit
of
json.
That
is
then
understood
by
the
designer.
That's
another
thing
that
needs
to
be
added.
A
Then
it
retains
that
information
that
positional
information
that's
great
yeah.
I
think
we
should
do
that.
So
this
elsa
three
thing:
it
has
a
service
that
can
turn
the
workflow
that
we
have
here.
The
structure
into
an
actual
hierarchical
tree
view
shape
and
there's
then
a
walker
that
walks
this
tree
and
then
assigns
unique
identifiers
to
each
of
them.
So
as
long
as
you
don't
change
the
structure
of
the
workflow,
then
the
naming
should
be
the
same.
A
I'm
doubting
myself
now
because
I'm
not
sure
if
I
recall
generating
unique
identifiers
because
then
that's
useless,
but
if
it
uses
a
pre-deterministic
algorithm
by
using,
for
example,
4-1
for
this
one,
because
it's
the
first
four
and
then
sequence,
one
etc.
Then
we
could
use
that
name
yeah.
We
can
figure
something
out
and
as
long
as
the
workflow
isn't
changed,
then
the
name
will
be
deterministic
and
there
and
that
could
then
be
bound
to
the
positional
information,
yeah
and
and
worst
case,
the
user
does
change
the
it's.
It's
no
biggie.
A
It
will
just
out
of
layout
that
thing
and
you
can
always
correct
it
yourself
in
the
designer
yeah.
So
I
will.
I
will
turn
this
into
issues
later,
just
making
quick
notes
yeah.
So
this
is
for
one
example
of
toolbar,
so
it's
there's
a
lot
of
little
things
to
do
automatic
layout,
sub
actions
need
to
be
implemented.
So
what
are
sub
actions?
That's
this
little
button
here.
These
are
inactive,
publish
words,
but
the
rest
aren't
implemented
it's
little
nice
details.
A
I
think
that
can
be
easily
picked
up
by
anyone
who
is
interested,
there's
also
activities
themselves.
For
example,
the
brake
activity
is
lacking
in
elsa
3
still,
and
the
brake
activity
is
important
to
be
able
to
break
out
of
a
loop
which
could
be
a
while
loop
could
be
four.
Each
could
be
four,
maybe
other
grouping
constructs.
A
This
enables
the
user
to
break
out
of
loop
once
a
certain
condition
is
met
yeah,
so
both
in
lsat,
two
and
also
three.
You
will
be
able
to
loop
back
using
by
connecting
these
nodes.
So
then
you
have
a
let's
say,
a
natural
loop,
but
in
elsa
3
you
will
have
the
same
structure
as
in
lsat
2,
and
let
me
just
show
you,
for
example:
here
we
have
a4
hits
and
it's
nice.
I
think,
when
using
the
c-sharp
language
to
define
your
workflow.
So
here
we
start
with
a
right
line.
A
And
then,
of
course
you
can
add
it
to
your
sequence
here
and
then
at
some
point
in
the
sequence
you
want
to
well,
you
want
to
do
some
stuff.
So
let's
say
for
body
is
some
sequence
and
now
let's
say
you
have
a
whole
bunch
of
activities
here
and
one
of
these
activities.
Now
you
want
to
basically
connect
back
to
your
voice
from
within
your
loop.
It
would
work,
but
from
c
sharp's
perspective.
A
Yes,
but
I
understand
what
you're
saying
from
the
designer
point
of
view,
it's
a
different
story,
because
now,
if
you
imagine
what
this
would
look
like
from
the
designer,
it
kind
of
depends
how
you
visualize
it,
I
think
if
you
visualize
it
like
windows,
workflow
foundation,
I
don't
know
version
3
or
so,
where
you
don't
have
the
flow
chart.
But
you
have
these
blocks,
so
you
and
then
they
are
little
cells.
A
So
then
you
can
put
this
right
line,
for
example
in
a
nested
cell
or
if
it
would
be
nested
within
the
forehead,
so
then
visually
it
would
make
more
sense.
But
from
a
when
you
look
at
a
float
chart
model,
I
agree
it
can
be
confusing.
Probably
we
can
come
up
with
something
that
the
designer
understands
this
for
the
for
its
activity,
where
you
would
still
have
an
outcome
of
well,
so
it
would
have
an
iterate.
I
think
it's
called
this
for
its
activity
has
a
body
it's
an
outbound.
A
I
actually
don't
know
if
I
already
implemented
this.
My
thinking
was
this
would
show
up
as
a
port
on
the
activity
in
the
designer
wait.
Let
me
see
if
I
can
just
show
all
right,
so
here
we
have
a
48
and
then
see
if
it
shows
it
doesn't
currently
show
the
label
of
the
output,
but
this
would
then
be
the
the
body.
A
Basically,
this
needs
to
be
enhanced,
because
I'm
not
sure
if
this
is
the
body
of
or
if
it's
if
it
received
an
automatic
done
outcome
for
because
this
is
the
thing
that
we're
designing
right
now
is
actually
a
flowchart
activity,
and
it
does
some
special
thing
to
allow
you
to
connect
arbitrary
nodes
to
one
another,
but
also
mixing
with,
for
example
this.
If
this,
if
it's
a
then
and
an
else-
and
this
is
a
similar
construct.
So
if
we
go
to
this,
if
activity
has
a
then
property
and
an
else
property.
A
So
this
is
the
true
brands
and
the
false
brands.
But,
as
you
can
see
from
a
code
point
of
view,
you
just
pop
in
some
activity
to
execute
which
again
could
be
a
flowchart
could
be
another.
If
could
be
a
sequence
and
based
on
these
names,
it
renders
these
outputs
and
the
flowchart
designer
allows
you
to
assign
basically
this
activity
to
this
property
here.
So
it's
it's
a
little
process
called
transposition
or
something
but
yeah.
A
My
main
point
or
the
point
I'm
trying
to
make
is:
we
can
probably
come
up
with
something
nice,
so
that
would
have
four
heats
if
we
think
that
we
want
to
do
it
like
that,
we
can
just
have
it
hook
back
or
maybe
add
an
an
inbound
port,
so
they
give
a
clear,
visual
hint
to
close
the
loop,
basically
yeah.
We
can
think
about
this
for
a
nice
way
and
by
the
way.
A
Currently,
all
of
these
activities
look
the
same,
but
we
can
customize
the
look
for
each
activity
individually,
which
I
think
I
want
to
do
for
things
like
if,
but
also
the
flow
chart.
This,
of
course,
isn't
a
good
flowchart.
Maybe
it
should
use
some
icon,
the
more
reflects
a
flowchart,
and
then
you
should
be
able
to
double
click.
A
new
tab
would
be
opened
within
the
designer
that
allows
you
to
basically
drill
into
your
workflow
bottom
line.
Lots
of
work
to
do
so.
A
Let
me
see
so:
let's
imagine
that
you
don't
have
this
one,
but
we
have
it
for
it's.
Let's
imagine
this
is
the
body,
and
then
we
have
a
bunch
of
activities
so
from
white
line
to
this
one.
To
this
one,
and
now
let's
say
here-
we
want
to
add
a
break.
So
imagine
this
is
here
is
another
outcome,
so
this
would
be
the
iterate
or
the
body
port,
and
there
will
be
a
done,
so
that
would
mean
indeed,
as
soon
as
it
encounters
a
break.
A
It
will
not
con
continue
the
remainder
of
the
body
of
the
foreheads,
but
it
would
then
execute
whatever
is
connected
to
the
done
port
that
we
don't
see
right
now.
I
think
that
will
be
more
natural,
more
flexible,
because
you
can
have
multiple
conditions
at
certain
points
in
your
process,
at
which
point
you
may
want
to
break
the
loop
versus.
If
you
have
a
separate,
socket
or
an
outboard
for
break
where
you
can
add
a
condition,
it
could
work,
but
then
it
may
be
a
little
bit
too
limited
in
terms
of
under
with
circumstances.
A
A
We
could
also
come
up
with
a
visualization
where
these
activities
become
nested
within
your
freight,
so
maybe
we
require
the
user
to
expand
the
for
it,
and
then
this
thing
becomes
a
big
box
and
then,
within
that
box
you
have
like
a
mini.
Workflow
could
be
a
sequence,
could
be
a
single
activity
or
it
could
be
a
floater
that
you
drop
into
it,
and
then
you
have
a
little
flowchart
designer
within
your
fruit.
That's
entirely
possible
as
well.
Do
you
ever
happen
to
have
an
example,
yeah
yeah?
A
That
would
be
good
for
inspiration
if
nothing
else
yeah,
so
the
break
activity
run
at
is
a
timer
activity.
It's
similar
to.
Well,
it's
not
similar
to
delay.
Let
me
rephrase
that
it
is
similar
to
delay,
but
here
you
specify
an
explicit
time
in
the
future
at
which
this
activity
should
trigger
the
workflow
in
elsa
2.
It's
called
start
at
here.
It's
called
run
at
and
it's
an
arbitrary
name
change.
Probably
we
should
use
the
same
unless
enough
people
think
run.
A
It
is
a
nicer
name
than
star
that
probably
we
should
keep
it
as
start
at,
but
that
needs
to
be
implemented.
Same
goes
for
fitness,
and
then
this
entire
lane
will
will
grow
over
the
next
few
days
and
weeks
with
little
tasks
are.
Some
are
little.
Some
are
bigger,
it
will
give
a
hopefully
a
good
overview
of
remains
to
be
done
for
ls3
preview.
One.
A
But
just
once,
and
a
cron
in
my
mind
at
least
represents
something
that
can
be
recurring
unless
it's
generally
a
valid
use
for
crown
activities
to
also
just
be
a
one-off
event,
then
maybe
you're
right,
but
would
it
also
allow
you
to
specify
a
specific
timestamp
in
the
future
with
with
chrome,
so
a
connectivity
would
allow
you
to
specify
a
date
and
the
time
up
to
the
second
and
maybe
even
millisecond,
although
that's
probably
not
us,
maybe
we
can.
We
can
think
of
through
some
of
the
use
cases
for
cron
and
run
at
see.
A
If,
because,
if
they
all
match
up,
then
we
can
get
rid
of
the
run
out,
maybe
and
make
it
just
use
the
the
chronic,
because
it
would
also
can
execute
once
in
this
activity,
allows
the
workflow
to
stop
immediately.
That's
one
that
needs
to
be
done,
notifications,
ui.
So
that
would
maybe
look
something
or
similar
to
what
the
azure
portal
has
so
some
some
icon
here
that
you
can
always
invoke
to
see
what
the
what
were
the
recent
notifications.
A
I
thought-
maybe
that's
nice,
because
right
now,
there's
there's
nothing
as
with
any
issues.
Any
suggestions
are
welcome,
but
this
was
my
initial
id.
I
noticed
we
are
very
close
to
the
end
time.
I
I
did
want
to
give
a
little
demo
of
something
geeky.
So
it's
a
dsl
for
lsat3
and
pull
it
up
here
and
this
dsl.
A
What
allows
it
that
the
user
to
do
is
is
to
have
an
alternative
to
visual
workflows
and
see
sharp
code
to
model
a
workflow,
but
using
a
specific
domain,
specific
language
that
hopefully
makes
it
a
little
bit
easier
to
define
workflows
for
some
users.
It
will
be
useless
because
they
need
visual
diagrams.
I
just
prefer
to
just
use
c-sharp,
which
you
know
gives
you
intelligence
and
it
gives
you
syntax
coloring,
but
another
set
of
users.
A
A
Potentially
the
first
sample
here
is
a
workflow
that
has
just
one
activity:
it's
the
right
line
activity
and
it
uses
a
a
syntax,
that's
similar
to
json,
where
you
can
configure
each
property
with
a
value,
and
the
value
could
be
a
literal
as
you
see
here,
but
it
could
also
be
an
expression,
so
it
could
be
a
javascript
expression
so
that
you
can
assign
values
dynamically.
So
let's
just
run
this
first,
just
to
prove
that
it
works,
and
this
works
of
course,
very
simple.
A
So
that
means
this
argument
gets
mapped
to
the
text
property
of
the
rightline
activity.
Here
I
have
another
one
read
which
we'll
see
in
a
second
as
well.
This
works
too
just
to
run
it
quickly.
So
here
again
prints
hello
world
here
is
a
more
complicated
workflow
because,
usually
you
will
have
constructs
like
for
eats
and
if
and
in
this
case,
a
sequence
and
and
that's
supported
too
so
you
can
specif.
You
can
add
multiple
activities
in
the
form
of
an
array
like
this,
and
then
it
will
just
go
from
well.
A
It
will
execute
sequence
which
which
in
turn
will
schedule
the
right
line
activity
the
recline
activity
assigned
to
a
variable
here,
so
you
can
reference
it
later
for,
for
example,
here
imagine
the
user
is
prompted
to
enter
their
name.
Then
the
recline
activity
allows
the
user
to
input
their
name
through
the
console,
and
then
you
want
to
echo
back
that
name
using
a
javascript
expression,
and
you
could
do
it
like
so,
but
with
the
caveat
that
this
does
not
currently
work,
I
am.
This
is
not
yet
implemented.
A
So
here's
a
demo
that
actually
does
allow
you
to
do
this.
Actually,
this
is
not
a
good
example.
This
would
be
better
but
we'll
get
to
that
in
a
second.
This
is
just
relying
on
the
on
the
functions
and
this
syntax
here
being
able
to
declare
a
workflow
variable
is
what's
missing
in
demo
3a
here,
there's
no
way
to
tell
the
the
interpreter
that
I
want
to
define
a
workflow
variable
or
even
use
this
output
of
this
read
line
because
it
has
a
return
value
that
could
then
be
used
here.
There's
no
nothing
for
that.
A
I
I
was
very
close
to
finishing
this,
but
I
just
didn't
make
it
in
time,
but
let
me
show
you
the
demo
three,
as
you
saw
as
for
my
name,
I
entered
well,
not
my
name,
but
it
still
works.
Hello
jack
was
echoed
back
and
small
detail
here
there.
This
was
evaluated
at
runtime
when
it
executed
the
writeline
using
the
javascript
interpreter.
So
this
is
important.
This
allows
you
to
specify
the
syntax,
so
this
is
javascript
could
be
any
future
language.
A
Wouldn't
it
be
cool
if
you
can
use
something
like
this,
where
you
actually
have
keywords
where
you
can
use
a
programming
language
that
looks
like
javascript
looks
like
c
sharp,
but
it's
actually
a
workflow,
and
then
you
can
have
things
like
well.
These
are,
of
course,
short
lived
functions,
but
let's
say
now:
I
want
to
wait
for
user
approval.
This
could
be
some,
you
wouldn't
even
have
to
use
a
weight,
just
user
approval.
This
will
be
an
activity
that
you
configure
with
anything
you
you
need
could
be
a
user
id
could
be
many
more
properties.
A
You
need,
and
in
this
way
have
a
more
well
two
developers
would
be
more
natural,
not
to
end
users
for
end
users.
This
would
not
be
suitable.
I
imagine,
but
this
thing
works.
So
if
we
go
from
top
to
bottom,
we
again
print
out
a
prompt.
We
store
the
readline
result
into
a
workflow
variable.
Then
we
enter
some
if
statement
and
then,
if
this
condition
is
met,
it
performs
these
two
statements.
A
If
not,
it
performs
this
statement
and
then
finally
an
end
statement
just
to
show
that
this
one
works
as
well.
So
let's
enter
12
and
when
you're
12
you're
too
young
to
drive
a
car,
so
you
need
to
use
a
bicycle
and
of
course
we
can
try
another
value,
and
now
the
other
branch
is
executed.
But
again,
I
think,
for
some
users
it's
useful
for
many
uses.
Maybe
not
a
couple
of
big
disadvantages.
Of
course,
is
you
don't
have
color
syntax
highlighting
it
could
be
added?
A
You
probably
also
need
some
proper
debugging
support
to
be
added
in
the
future,
but
seems
complicated
to
me
to
do
it
in
the
short
run,
but
it's
a
bit
of
a
side
project
that
I've
been
working
on
just
for
the
final
kick
is,
is
to
show
that
this
language
is
actually
turned
into
a
workflow
definition.
So
here
we
have
this
script
that
I
loaded
in
which
we
just
looked
at.
That's
demo4
this
this
fonts
and
then
here
we
have
the
workflow
definition.
A
It
took
the
script,
pushed
it
through
the
dsl
engine,
and
then
we
have
a
root
sequence,
its
activities,
write
line,
redline,
etc.
The
if
and
the
else
brands
and
the
then
friends
itself
is
a
sequence
of
two
activities.
But
that
concludes
this
little
demo.
Thank
you
all
right,
then.
I
think
it's
really
time
to
close
off
alright
take
care.
Everyone,
bye-bye.