►
Description
IOHK Team Grothendieck Meeting, March 14, 2017
A
A
So
this
time
last
week
we
talked
about
as
I
said,
completing
that
phase
and
we
spent
the
whole
week
working
on
testing
I'm
trying
to
get
trying
to
get
through
trying
to
get
everybody
on
the
team
through
a
set
of
tasks
that
exercise
the
client
from
end
to
end,
with
respect
to
hosting
a
blockchain,
download
and
downloading
a
blockchain.
With
the
caveat
that
we
don't
execute
any
transactions,
I
might
add,
just
block
sink
just
involves,
as
I
say,
moving
the
blocks
around.
A
So
it
was
a
very
successful
testing
week
in
that
we
found
and
fixed
quite
a
few
things,
and
we
basically
learned
that
there
were
some
gaps
in
what
we
had
in
what
we
had
put
together.
So
this
is
a
the
ideal
time
to
find
these
things
and
get
them
sorted
and
I
think
we're.
Probably
we
have
about
one
or
two
POS
which
are
coming
in
this
afternoon
and
after
that
we
should
be.
We
should
be.
A
Everybody
should
be
able
to
basically
walk
through
our
our
set
of
tests,
our
suite
of
tests
and
get
the
entire
block
chain
student
downloaded
meant
end,
but
that's
not
to
say
that
we
haven't
managed
to
successfully
run
through
the
tests.
In
fact,
yeah
and
you're
going
to
tell
us
a
little
bit
about
some
of
the
some
of
the
tests
results
that
we
have.
A
B
Hello,
hello,
everyone,
so
let
me
start
with
some
very
basic
explanation:
was
the
difference
between
the
regular
sink
and
fasting
regular
send
is
the
procedure
of
replaying
every
transaction
on
on
your
note
and
say
we
are
saying,
if
you'd
like
to
start
the
Blanco
transactions
starting
from
Genesis
blog,
it
would
take
I'm
not
sure
how
long,
but
it
probably
more
more
than
one
day
more
than
24
hours.
First
thing,
on
the
contrary,
is
much
faster,
as
the
name
says.
B
It's
basically
downloading
whole
state
of
the
network
for
a
given,
given
very
recent
blog
and
fast
thinking
usually
takes
around
around
a
couple
of
hours,
so
we
spent
one
week
so
far
over
actually
over
11
week,
polishing
fasting
and
regular
sink,
and
they
are
basically
working
apart
from
a
few
bugs
which
I
I'll
get
I'll
get
get
back
to
it
later.
But
I
can
already
show
you
some
numbers:
how
long
does
it
take
on?
It
is
Clyde.
How
long
does
it?
How
long
does
hosting
take
on
other
clients,
I'd
like
to
show
my
smile
screen?
B
I'm
taking
it
once
again
application.
Oh
there
we
go
so
today
we
got
first,
let's
say
real
numbers:
how
long
fasting
types
on
TC
client
you
can
see,
but
it's
we
clocked
around
three
hours
and
45
minutes
compared
to
two
hours
and
20
minutes
and
poverty
is
slightly
different.
Because
party
has
its
own
sinking
sinking
algorithm
called.
As
you
can
see
web
saying
warp
sink
well
I.
B
I
I'm
not
going
to
talk
about
it,
but
one
sink
is
implemented
only
by
parity
and
polity
note,
in
order
to
do
a
word
up
sink
has
to
connect
only
two
other
party
notes.
So
it's
not
it's
not
universal
everything.
It's
not
supported
by
all
nodes
on
the
network.
So
for
now
we
are
not
going
to
work
on
it.
B
Just
a
small
note.
Regarding
these
three
hours,
45
minutes,
you
manage
to
decrease
this
time
from
like
10
hours,
the
beginning,
I
think
Paris
veg
products
were
showing
yes
around
10
10
hours
of
passing,
and
I'm
pretty
sure
that
we
still
have
a
lot
of
people
improvement
to.
I
think
we
can
go
down
as
low
as
gap
so
size
of
the
database.
B
B
B
A
So
we're
not
we're
not
at
the
stage
where
we're
going
to
look
at
optimizations
as
of
yet
really
what's
important
is
that
we
get
it
that
we
get
it
down
and
that
we
figure
out
that
all
that
it's
M,
as
I
say,
functionally
functionally
correct,
I.
Suppose
one
of
the
interesting
things
that
that
we
learned
during
the
week
just
for
for
people's
general
interest
is
we
had.
We
had
an
issue
with
the
leveldb
java
version.
So
when
we
were
downloading,
we
ended
up
with
with
exceptions
being
generated
on
the
compaction
thread.
A
Now,
as
it
turned
out,
we
may
that
actually
may
have
been
something
to
do
with
with
the
load
that
we
were
putting
on
it.
So
that's
something
that
we're
going
to
to
to
carry
forward
basically
and
and
keep
an
eye
on
over
this
week.
So
that's
something
that
we'll
be
looking
at
closely
I
guess
when
we're.
When
we're
all
testing
see
if
we
still
generate
errors
in
them
in
level
DB
Java.
Now
we
did,
we
did
just
yeah
and
you
can
you
can
stop
sharing
that
screen
now,
if
you
like,
just.
A
Again,
hi
Sam,
so
just
again
for
people's
interest.
We
did
also
try
to
replace
that
database
with
the
with
the
native
version
just
to
see
if,
if
that
run,
that
that
run,
okay,
Donnellan,
yes,
it
ran
ok,
a
we
ran
some
tests
and
we
didn't
have
the
same
error
or
saturday
pure
java
version
yeah.
But
equally
since,
since
our
latest
change,
we
we
don't
have
we
don't
have
those
we
haven't,
seen
those
errors
either.
So
that's
something
that
we're
going
to
keep
an
eye
on.
So
the
status
is
basically
we've.
A
We've
we're
almost
code
complete
with
that
and
we
all
expect
to
start
testing
at
this
evening
or
tomorrow.
But
again,
as
I
said
last
week
in
this
game,
things
are
either
done
or
not
done,
so
it
still
has
to
go
into
the
category
of
not
done
now
when
we
look
at
the
schedule,
the
schedule,
luckily
enough,
we're
still
well
well
within
our
reserve,
so
we're
still
tracking
to
the
to
the
schedule
as
published.
But
it's
it's
just
it's
it's.
It
is
what
it
is.
A
We've
learned
a
lot
over
the
over
the
last
week
we
fixed
a
lot
of
stuff.
We've
done
a
lot
of
good
work
and
just
today
has
come
probably
a
day
or
two
too
early,
so
we'll
continue
on
we'll
continue
on
blocks
ink
for
the
next
few
days
and
in
our
planning,
assuming
we
have
any
kind
of
bandwidth,
we'll
talk
again
about
resuming
work
on
the
transaction
execution
phase.