►
Description
Rust Cologne User Group - Meetup June 2016
More on http://rust.cologne/2016/06/06/rust-anniversary-part-2.html
Help us caption & translate this video!
http://amara.org/v/2Fi7/
A
Okay,
hello
and
welcome
to
my
talk.
I
threw
together
yesterday
with
the
beautiful
title:
why
parts
10,000
lines
of
C++
code
to
rust
and
what
I
learned
about
this
I'm
visiting
hacker
from
boham
I'm
working
in
the
industry,
I'm,
currently
doing
only
open
source
and
I'm
starting
to
2020
computer
security,
and
my
interests
are
mostly
cryptography
and
reverse
engineering
and
I
will
talk
a
bit
about
the
wrestling
topic
here.
I
don't
have
much
about
the
actual,
wasn't
very
project
I'm
doing,
because
I
wasn't
anticipated
that
anybody
here
cares
about
that.
A
So
it's
more
about
rust
and
bit
critique
of
modern
C++,
but
now
that
some
people
are
interested
in
the
stuff
and
maybe
I
talk
a
bit
more
about
attacks,
reverse
engineering,
ok,
the
reverse
engineering.
This
is
what
presently
looks
like
in
reality
in
the
industry.
That's
either
pro!
That's
that
standard,
2
or
4
was
engineering,
especially
in
the
security
industry.
A
It
looks
like
Windows
1595.
It
feels
like
it
and
it's
mostly
out
from
that
era.
It's
definitely
one
of
the
best
tools
we
have.
There
is
so
there
that's,
there's
a
reason
why
it's
used
in
industry,
but
it
costs
800,
bucks
upwards,
it's
more
or
less
ignores
the
results
of
the
last
10
years.
Program.
Analysis,
research,
so
around
5
years
ago,
when
I
wanted
to
resently
something
on
Linux.
This
is
at
that
time
was
the
windows
only
I
thought
I
can
do
that
better.
A
A
But
the
problem
is,
you
have
libraries
for
support,
vector
machines
and
SAT
solvers,
but
you
don't
have
libraries
for
uisce
display
an
image
to
display
button.
You
can
press
why
there
are
bindings
for
QT
they're,
either
out-of-date
or
not
very
good,
portraits
and
the
fact
that
Tasker
is
extreme
different
from
all
other
libraries
and
tools.
Ef
it's
hard
to
port
libraries
and,
in
the
end,
I
had
to
learn
that
I'm,
probably
not
smart
enough
for
ESCO,
so
I
used
the
Hema
I
know
which
was
C++
I
start
with
C++
around
ten
years
ago.
A
At
that
point,
C++
started
to
started
to
move
forward
again
a
bit.
We
had
this
super
fast,
o
X,
initially
called
it,
and
then
it
became
2010
in
depth
because
I
came
to
eleven
and
finally,
it
released
and
we
had
C
versus
eleven,
which
feels
like
a
whole
new
language.
So
I
will
write
a
bit
of
C++
and
the
simple
community
and
how
they
solve
problems.
But
it's
definitely
it's
great
language
and
in
case
you
can't
use
rust
or
Haskell.
A
A
Well
the
project
at
that
time
used
what
people
call
modern
C++,
which
is
mostly
template,
metaprogramming
and
have
use
of
templates.
That
became
the
driving
factor
behind
modern
C++
because
it
helps
you
to
build
abstractions
that
are
typesafe,
that
kind
of
integrate
internal
language
at
the
expense
of
usability.
But
it
helps
you
to
actually
build
new
language
constructs
and
well
you
as
a
superstores
programmer.
You
learn
how
to
push
that
language
constructs
beyond
its
limits.
A
So
what's
the
problem
of
C++
that's
code
for
velocity
was
first
version
of
protocol.
I
would
try
to
figure
out
all
the
errors
here,
I'm,
not
sure
if
I've
managed
to
well.
Let's
start
with
the
first.
In
the
last
line
that
ensure
that's
what
other
languages
call
assort,
so
if
that
the
expression
in
the
braces
and
parents
aren't
true,
isn't
rules
that
program,
is
this
blows
up?
A
There's
a
certain
C++,
but
that
calls
just
exit
to
ensure
I,
build
this
bit
smarter
on,
but
more
or
less
than
that
first
probability
plus
plus
and
see
I'm
sure
as
a
macro.
So
it
doesn't
behave
like
a
function
and
the
initial
versions
of
ensure
were
compiled
out
in
release
mode,
and
this
means
that
everything
you
did
there
just
disappear
to
release
much.
A
So
if,
like
in
the
last
line,
you
do
something
that's
actually
needed,
even
really
small,
like
inserting
your
value
into
this
lambda
container,
your
application
stops
working
every
small,
so
you
have
something
that
looks
like
an
functional
but
doesn't
behave
like
one
alphabet
in
rust.
You
have
this
thing
at
the
end,
but
would
help
you
to
figure
out
that
maybe
that's
another
function.
Okay,
know
where
I
see
you
do
it
Oh
an
uppercase
but
nobody's
forcing
you
to
do
this?
A
Okay,
next
up
that
the
remainder
of
this
function
is
just
one
function
call
inside
another
function.
Call
well,
this
accumulate
fraction
call
is
just
what
Rosco's
fold
and
it's
get
fed
to
it
to
iterators
in
Sivas
plus.
For
most
of
you
probably
know
you
don't
have
this
one
iterator,
that's
called
next
next
next
until
it
ends,
but
you
have
to
iterate
as
the
point
at
the
beginning
and
past
the
end
and
you're
supposed
to
move
one
of
the
situated
towards
the
end.
A
You
can
only
compare
them
by
equality,
so
if
you
move
one
of
these
iterator
past
the
end,
you
maybe
blow
up
your
function
so
there
by.
While
you
had
a
generalization
of
pointers,
they
are
strictly
worse
from
the
safety
perspective,
so
instead
of
looking
if
they're,
smaller
or
equal,
if
you
can
just
check
for
equality
in
Russells,
what
we
have
pretty
hard
to
do
this,
because
your
iterator
will
one
day
or
one
situation
will
start
to
just
return.
None
the
rest
of
this
function
computes
some
value
in
some
graph
algorithm,
it's
not
so
important.
A
But
what
Caleta
see
is
that,
were
there
C++
at
least
you
are
told
to,
but
they
are
more
interpretations
of
C++,
one
for
GCC,
one
for
clang
and
one
for
Microsoft
and
all
of
those
interpret
that
standard
differently.
So
you
have
to
anticipate
all
of
them
actually
I
implemented
most
of
the
stuff
and
GCC
and
part
of
it
to
Microsoft,
compile
and
took
me
weeks
until
it
finally
compiled
just
because
sometimes
Microsoft,
as
you
can
see
here,
beliefs
that
you
don't
need
that
type
name.
A
A
Well,
the
rest
of
that
is
a
llama
function.
That
looks
absolutely
you
started
with
dead
parents
and
the
ampersand,
and
after
that
you
have
to
write
out
all
the
arguments
again.
There's
some
limited
type
of
trails,
but
not
in
the
rust
free,
so
again,
extremely
verbose.
It
doesn't
have
anybody
if
it
had
type
inference
here.
At
least
I
would
have
skipped
that
macro
for
the
Metro
compile
right
and
at
the
end
of
course,
we
compute
the
maximum.
A
We
catheter
an
int
and
return
it
into
an
unsigned
int,
and
this
is
depending
on
your
compiler
flags,
in
a
warning
or
nothing
just
works
as
long
as
really
not
scented
and
not
below
zero.
So
this
is
most
of
the
problems.
I
had
tried
to
find
and
represent
of
line
and
perhaps
icon,
where
I
could
show
you
that
C++
is
terrible,
but
superclasses
not
expressive
enough
to
have
all
the
terrible
into
one
line.
So
I
have
this.
A
Yeah,
why
did
I
stop
for
us?
So
I
found
all
about
rust
around
that
ol
nine
I
think
where
I
was
there
I'm
confusing
amount
of
tillers
and
at
the
science
or
look
like
pearl,
but
what
I
heard
about
first
and
what?
What
what
made
me
fell
off,
love
initially,
was
that
roughness,
sometimes
or
called
check
enums
the
possibility
that
you
have
something
like
a
union
and
see,
but
instance,
the
value
of
the
union
know
which
part
they
are,
which
you
can
measure
on
them.
A
Of
course,
every
language
constructs
you
can
come
up
with
this
already
implemented
in
c
and
c++
using
boost
larry's
using
template.
Metaprogramming,
they
tell
you
everything
is
fine.
It
works
the
same
as
in
all
other
languages.
If
you
make
a
mistake,
it
doesn't
compile
what
they
don't
tell
you
is
that
doesn't
compile
means
you
get
error
messages,
libel!
You
have
no
idea
what's
happening.
A
Great
a
few
good,
the
the
thing
where
I
stopped
and
thought.
Maybe
this
template
metaprogramming
thing,
isn't
that
greatest
idea
is
where
I've
found
out
about
the
boost
way
of
making
these
error
messages
possible
for
a
human
user
is
that
they
build
upon
that
idea
of
concepts
that
will
never
appear
in
C++
and
have
narrow
messages.
A
Essentially,
you
have
an
identifier
for
each
of
the
possible
errors,
and
you
put
a
bunch
of
stars
forward
before
the
identifier
and
after
and
so
that's
it
will
still
a
valid
identifier
and
it
shows
up
into
your
1
gigabytes
or
so
error
messages.
So
you
know
at
least
where
to
look
for
that's
the
point
where
I
stopped
using
template,
metaprogramming
and
okay
back
to
a
rough
yeah.
This
is
a
shortened
version
of
what
happens
in
particle.
Often
is
when
I
match
against
these
well
tech
unit
limitation
done
in
booths.
It's
called
boost
variant.
A
The
ideas
you've
been
functional,
object
that
overloads
the
call
operator
for
all
the
possible
at
scientists
in
the
Union
and
well,
you
put
an
aspect:
function,
apply,
visitor
and
depending
on
which
option
this
instant
is.
It
will
call
that
and
function
great
and,
of
course,
if
you
miss
one
of
those
options
that
doesn't
comply,
you
get
five
pyramids
error
message.
A
First,
this
is
extremely
verbose
because
you
have
to
write
this
function
object
again
and
again,
you
can
deconstruct
away
here.
So
if
you,
if
like
here
I,
want
to
use
that
at
arguments,
I
have
to
painstakingly
take
their
values
out
and
match
on
top
of
them.
So
let's
say:
if
I
have
this
app
and
want
a
special
version
of
the
at
where
the
first
argument
is
in
constant
value.
I
can
just
mention
the
rust
and
it
C++
I
have
to
write
another
function,
objective
measures
on
the
first
argument,
depending
on
that
implement
all
the
functions.
A
A
No
threat
doesn't
start
modifying
it
and
I
build
some
vilely
how
to
understand
smart
pointer.
That
did
all
that
or
these
are
try
to.
As
you
can
see,
I
committed
out
the
right
function
because
I
never
managed
to
actually
that
correctly.
I,
never
understood,
sir.
Why?
But,
when
I
came
to
rust,
I
realized
what
I
did
there
was
more
or
less
boudoir
checker.
That
was
llanto
checked,
and
this
was
another
point
for
me.
Worse
or
maybe
rust
is
the
right
wave
selfs.
It
was
person
doing
everything
yourself,
okay.
A
So
when
rust
came
Wes
rust,
one
came
out
around
a
year
ago.
I
started
trying
with
that
idea.
The
problem
is
the
project
I
started
now.
Five
years
ago,
no
head
or
e4
reverts
and
I
was
programmed
with
all
it
wasn't
one
my
project
all
alone,
past
muscle,
most
of
the
time
wasn't
even
Akita,
or
so
it
was
always
tink.
Tink
aid
never
actually
surpassed
that
10k
limit,
because
I
always
started
to
rewrite
it.
A
A
Trees,
so
I
could
translate
it
more
or
less
directly
to
rust,
and
it
took
me
around
four
months
it's
hard
to
tell
because,
of
course,
the
51
I
learned
a
bit
more
about
the
other
four,
probably
a
bit
differently
and,
of
course,
was
long
experience
or
I
start
with
the
basic
part.
The
utility
libraries
and
this
foundational
part
of
protocol
is
actually
diverse,
designed
because
I
had
no
idea
about
rust
and,
just
probably
like
a
thought,
was
a
good
idea.
So
parts
of
that
changed
again
but
after
four
month
mark
I
was
around
that
point.
A
That
I
could
do
the
same
with
the
rest
version.
As
I
could
do
this
C++
one
yeah,
it
was
in
trial
experience.
Definitely
the
compiler
messages
was
really
when
I
program
has
care.
I
always
had
the
same
situation.
I
had
no
idea
what
a
language
and
read
the
book
and
thought
was:
oh
I
understand
this
and
I
started
grabbing
and
I
realize
I,
don't
know
anything
and
the
kabbalah
gives
me
helpful
messages,
and
why
don't
you
try
it
here?
This
is
wrong
that
doesn't
compile,
or
here
type
inference,
doesn't
work.
A
A
I
found
bucks
and
four
bucks
by
straightforward
translation
of
code
from
C++
to
rust
and
there's
the
situation.
They
had.
This
graph
type
template
that
more
or
less
was
a
container
for
graphs
and
I
started
to
translate
it
like
I
did
all
the
other
card
and
there
was
some
position
where
there
wasn't
code.
We
are
deleted
vertices
and
first
I
deleted
the
address
that
connects
to
the
auditors
and
then
delete
the
vs.
itself
and
I
couldn't
for
life
of
my
get.
This
code
compiles
all
right
took
me
hours
and
after
a
day,
I
realized.
A
There
was
interrater
and
validation
back
in
there.
The
poro
checker
thought
that
the
C++
version
had
this,
but
for
years
and
never
came
up,
but
they
said
forward
saturation
from
syphilis
to
rust
for
that
program.
So
that's
why
I'm
cross
pen
and
cargo
makes
development
easier,
especially
windows,
getting
all
the
dependencies
in
with
the
marks
of
compiler
that's
pushing
8
gigabytes
interval.
Vm
is
work
off
for
a
week
that
definitely
and
don't
know
if
cargo
is
the
best
route
for
everything,
especially
packaging,
but
I
can't
read
later
yeah.
A
What's
not
so
great
or
what's
the
problems
I
had
or
have?
There
are
no
libraries
for
anything
so
really
again
in
Haskell,
but
there
are
galleries
and
I
use
qml,
there's
some
kind
of
JavaScript
Prolog
lovechild
that
helps
you
to
specify
a
goose
and
do
most
of
the
programming
in
JavaScript,
and
this
is
really
nice.
But
the
bindings
for
the
skew
mat
part
in
rust
are
pretty
rough
and
when
I
started
with
the
guidance
program
that
was
or
working
for
month
or
so,
and
he
started
to
comment
out
parts
where
he
didn't
know.
A
Whenever
I
was
safe,
we
worked
with
the
threading
and
what
I
did
was
for
project.
Remove
the
comments
so
I
could
use
threading
again
without
thinking
about
safety
and
put
some
shitty
code
on
top
of
that
and
I'm
using
this
for
now.
So
after
this
private
Fork,
where
I
sometimes
put
back
changes
from
the
original
project-
and
this
is
it
kind
of
works,
but
it's
sometimes
have
crashes
sundresses
of
classical
pylon
specific
machine.
So
that's
real
bad.
A
qml
RS
is
the
project.
There's
guys
need
help
it's
moving
with
moving
forward,
but
very
slowly.
A
They
have
not
not
documentation,
mattress
mattress,
a
bit
the
template
metaprogramming
for
rusts
and
I'm
part
of
enjoying
that
because
I'm
from
the
C++
world,
how
to
get
how
to
push
that
construct
beyond
its
limits.
On
the
other
hand,
I
don't
think
Russ
needs
to
be
as
complicated
as
C++
and
needs
to
be
as
expert
friendly
as
C++
still
are,
using
some
pretty
complicated,
macros
and
currently
I
have
to
use
a
shell
script
to
generate
the
macros
which
again,
which
I
was
code.
A
A
Oh
yeah,
yes,
the
cargo
I
actually
was
the
person
who
actually
reads
that
Debian
package
king
guide
and
the
pretty
anal
about
some
parts
for
except
we
can't
have
network
access.
Why
building
applications,
that's
pretty
difficult
to
do
with
cargo,
and
so
as
or
both
cargo
you
have
some
dependency.
Dependencies
is
armed
obvious.
So,
for
example,
with
the
human
arse,
you
need
an
C++
glue
code
between
the
C++
word
and
on
the
rust
world
and
that's
compared
with
CMake.
So
you
have
this
dependency
ensimag
that
doesn't
show
up
in
your
cargo
file.
A
That's
kind
of
bad
I.
Don't
really
have
a
solution
for
that.
Currently
I
just
pump
I
owe
everything
for
a
little
package
and
people
can
unpack
it
and
start
it.
It's
everything
compiled
statically,
so
it
works
pretty
good,
but
I'm
not
sure
whenever
that's
the
best
purchase
but
I
think
that's
a
problem.
The
distributions
have
to
solve,
because
if
you
look
at
Ruby
and
nodejs
and
pison,
they
all
have
their
own
picture
measure
and
if
I
have
to
update
my
system,
I
have
to
emerge,
I
have
to
gem,
update
and
and
p.m.
A
update
and
compile,
update
and
pip
update,
and
maybe
cargo
update
is
another
one.
That's
sorry
everything
for
me.
If
you
want,
you
can
check
out
my
disassembly
project.
It's
on
github
and
it
already
was
on
I
can
use
I
think
three
weeks
ago.
That
project
blew
up
enough
nine
hundred
worthless
internet
points
and
get
up
now.
A
B
The
one
guy
from
Mozilla
is
PC
Walton
right.
Have
you
looked
into
his
rusts
binding
for
system,
agnostic
interfaces,
lip-lip
UI
or
something
it's
called?
No,
it's
it's.
Basically,
a
C
is
C
system,
agnostic,
C
binding,
for
he
wrote
a
rust
wrapper
or
whatever,
which
allows
you
to
basically
write
your
interface
and
rust
and
have
it
then,
just
by
kindness,
kind
of
the
same
on
on
Mac,
Linux
and
without
plus
so
and.
A
Without
C++
is
always
a
good
thing,
the
provenance
that
is
the
Q
Mac
code
is
a
lot
of
work
and
part
of
the
UI
code
and
layout
into
graphs
is
done
in
JavaScript.
This
was
a
lot
of
work
and
I'm,
not
seeing
the
upsides
of
really
throwing
everything
away
and
do
it
in
another
UI
library,
especially
because
in
ql
you
can
do
many
things
that
are
extremely
complicated
in
c++
and
probably
in
rust
too,
like
animations
and
transparency
and
stuff
like
that
very
easily,
and
a
really
a
fan
of
QL
and
I.
C
D
A
Well,
the
idea
of
prop
lacan
is
not
to
emulate
or
re-implement
our
approach,
because
I
approach,
a
very
local
maximum
and
it's
a
great
tool,
but
the
last
10
years
of
program,
analysis,
research,
I'm
sure
that
you
can
do
definitely
more
especially
the
semantics
based
endlessness.
Most
of
the
tool
TF
now
and
aya
is
more
of
the
best
is
only
syntax
based.
So
you
have.
A
Human
reverse
engineer,
definitely
to
speed
things
up
or
automate
things
and
make
sense
of
code
faster,
and
this
is
the
point
where
I
want
to
go
with
protocol
and
there
are
a
lot
of
research
prototypes
that
do
part
of
parts
of
that.
But
there
are
reasons
research
prototypes.
They
are
mostly
done
an
objective,
Pesce,
objective
camera
or
high
scale
or
there's
an
Java
code
that
compiles
years
ago
on
that
obutu
10
version
of
that
PhD
student
and
that's
it-
and
this
I
cannot
stand
this
because
these
guys
are
from
university.
A
They
are
not
paid
to
but
usable
software,
but
to
build
graduate
research
prototypes.
But
what
I
want
to
do
with
practicum
is
just
take
this
last
few
years
of
research
and
implemented
into
a
tool,
that's
usable
for
people
who
totally
knowledge,
not
PhD
students
and
Twitter
that
can
be
deployed
easily.
It
just
works
like
a
problem,
so
there's
no
original
thought
in
panopticon.
It's
all
reading
papers
and
implementing
this
stuff.
And
if
you
had
look
at
issue
tracker,
there's
a
long
list
of
things.
I
want
to
do
and
implement
that
abstract
domain
and
implemented
decompiler.
A
E
E
It's
pretty
good,
but
I
suggest
reading
it
at
least
three
times,
because
it's
information
density
is
so
high
and
there
is
one
example
of
a
macro:
that's
like
200
lines
of
just
generic
macro
code,
and
it
shows
basically
everything
you
can
do
with
macros,
including
custom
conventions
too,
like
the
pool
based
macros
that
call
themselves
with
other
arguments,
it's
pretty
fantastic.
You
should
give
it
a
try.
Ok,.
B
Two
versions
of
that
book
of
different
different
timeliness.
The
older
one
will
have
a
notice
on
the
at
the
at
the
top,
linking
you
to
the
new
version.
Oh
well
know
what
one
of
those
also
added
like
I,
think
the
little
one
is
the
older
one,
and
he
also
said
it's
not
appropriate
because
the
book
is
rather
large.
Now
so.
A
Upwards
to
10,000
I,
don't
know,
I
didn't
check
when
I
was
done
with
the
conversion
actually
miss
it
was
around
8,000
or
so,
and
I
think
it's
probably
more
than
10,000.
Now,
mostly
because
there
are
stuff
like
in
x86
disassembly
in
there.
It
takes
a
lot
of
code
alright,
but
not
a
lot
of
thought.
It's
just
copying
the
instructions
from
the
menu
into
the
code.
A
I've
brought
some
macros
that
make
the
code
that
matches
certain
demonics
more
or
less
look
like
what
it
looks
in
the
manual,
so
this
kind
of
pushes
the
amount
of
corn
upwards.
Maybe
if
I
come
around
to
put
some
documentation
in
the
project,
the
mod
of
Lights,
at
least
in
that
code
files
will
exceed
15
K.
A
Definitely
around
one
to
two
thousand
lines
of
code,
but
it's
hard
to
tell
if
this
really
expressiveness
of
rust
or
just
me
throwing
cold
way
where
I
realized.
This
can
be
done
easier
or
it's
just
not
worth
it.
So
where,
as
I
said,
is
not
this
really
one-to-one
translation,
because
I,
of
course,
did
some
things
better.
The
fifth
time
ever
did
therefore,
before
for
my
gut
feeling,
I
would
say:
I
maybe
had
less
line
with
1000
nice
lines,
at
least
from
that
variant
thing
there.
A
B
So
if
one
of
us
wanted
to
wanted
to
like
help
you
out
here
them
there,
but
has
no
idea
what
he
or
she
is
doing
in
regards
to
to
your
actual
domain,
who
you're
dealing
with
what
is
the
the
the
unit
test
coverage
and
and
and
the
V
how?
How
accessible
is
the
code
to
someone
who
has
no
idea
about
the
actual
recent
engineering
stuff?
Is
there?
Is
it
easy
to
just
like
say,
I,
say
someone
who
says:
okay
I'm
for
me,
it
was
rust.
B
A
Definitely
a
code
coverage
F
around
3/4
of
the
code
covers
of
the
unit
tests.
There
is
documentation
on
the
C++
version
still
inside
the
tree.
The
eye,
detector
didn't
change,
and
most
of
the
stuff
still
applies,
even
if
the
names
of
the
functions
kind
of
changed,
that's
definitely
happening.
If
you
just
want
to
understand,
what's
happening
aside
from
that
really
code,
just
starting
it
compiling
it
realizing
it
doesn't
compile,
because
I
are
terrible
programmer
or
your
platform
has
something
different
and
sending
a
patch
to
me
is
already
helping
a
lot
aside
from
that.
A
Try
it
out.
I
have
some
basic
instructions:
how
to
use
the
program,
and
aside
from
that,
I
try
to
be
the
try
to
writes
code
as
easy
as
possible
as
real
as
possible
for
people
who
not
familiar
main,
but
that's
bit
hard,
because
on
one
end
you
have
the
graph
now
layouting
part,
that's
pretty
complicated,
I'm,
sorry,
stuff,
the
dot
or
graph
is
dust
and
sandwich
between
the
part.
It
does
the
disassembly
and
semantic
analysis
and
the
graph
algorithms.
So
it's
probably
pretty
hard,
if
you
don't
have
give
it
a
science
education.
A
But
if
you
don't
know
about
resolution,
I
think
don't
think
that's
that's
hard,
because
most
of
the
wrestling
stuff
is
more
or
less
computer
science,
normal
stuff.
You
do
as
clear
side.
They
are
not
many
complicated
towards
their
the
algorithms
they
actually
implemented.
Two
fights
also
the
other
parts,
are
just
decompiling
that
this
is
semi
a
stuff,
forgetting
the
data
structures
civilizing
them.
That's
a
big
problem
with
me,
because
I
use
this
rust,
serialize
Makarov
arrived
thing
and
the
former
changes
every
time.
I
change
something
with
the
structures.
A
A
A
So
myself
of
this,
but
now
especially
the
exposure
I
had
was
I,
can
use
more
and
more
people
come
to
me
and
ask
me
your
code
doesn't
compile
or
how
can
I
have
or
why
does
it
work
why
it
doesn't
work
in
my
a
platform
and
a
definite
trying
to
have
interests
where
you
can
at
least
we
can
at
least
help
you
to
actually
compile
code.
I
can
get
user
code
and
fix
easy
stuff.
A
We're
a
serialization
library,
Stephanie
problem,
the
Cerritos
CDs
30
thing
looks
a
bit
better.
I
did
not
have
time
to
try
it
out.
The
problem
I've
had
I
develop,
always
in
LA,
stable
and
now
I'm
fixed
to
one
seven,
because
Ubuntu
16
that
actually
so
when
people
want
to
use
the
project,
do
they
don't
have
to
install
that
Ross
compiler
as
binary
again
it?
So
they
have.
Can
it's
the
everything
with
app
get?
That's
definitely
nice
things
I
want
really
to
its
that
work
on
standard
distributions
and
well
with
Ross
table
you
can
use
clipping.
A
You
can't
use
most
of
the
cool
parts
of
the
new,
more
easy
usable,
theorization
libraries.
That's
the
very
program
I
have
aside
from
that
I
started.
I
tried
to
use
specialization
in
my
very
Russ
code
and
realize
there's
no
specialization,
arrest,
I!
Think
it's
for
a
better
looking
back,
because
that
means
the
project
already
in
this
template.
Metaprogramming
way,
I
would
don't
want
to
go
again,
but
sometimes
definitely
it
would
help
I
wrapped.
It
is
now
part
of
frost,
probably
definitely
worth
checking
out
and
aside
from
that,
I
don't
have
much
problems.