►
Description
Ticket: https://gitlab.com/gitlab-org/incubation-engineering/ai-assist/meta/-/issues/12
A
Hello,
welcome
back
to
another
update
of
AI
assist
and
today
we're
going
to
talk
about
the
updates
and
what's
coming
back
up
next,
so
I
had
a
lot
of
progress
that
went
very
quickly,
so
I've
added
support
for
formatting
Docker
files
and
support
to
write
Docker
files
to
doctor,
which
again
I
still
need
to
find
a
better
name
for
it.
So
if
you
have
any
suggestions,
leave
a
comment
down
below.
A
First
up
is
the
formatting.
So
if
we
take
a
look
at
the
faulty
Docker
file,
we
see
a
bunch
of
rules
being
violated
and
now
I
can
add,
minus
s
to
show
it
and
it
will
create
the
new
Docker
file
and
it
will
just
print
it
out
nicely
nicely
formatted,
more
readable
and,
more
importantly,
it
will
auto
correct
mistakes.
A
Not
all
of
the
rules
will
be
auto
corrected
yet
some
are
quite
complex
and
require
more
coding,
but
others
like,
for
example,
this
we
got
a
few
yeah.
We
need
to
chain
multiple
run
instructions
together.
So
that's
what
this
automatically
been
done
here,
as
you
can
see,
I
have
to
get
and
then
another
again
Optical
we're
not
parsing
the
actual
command
itself
or
just
chaining
them
together,
making
sure
that
they
are
at
least
in
one
layer
now.
I
can
also
do
I
know
that
both
minus
s
m
minus
W
for
right.
A
So
if
I
write
it
and
then
look
at
my
commit
difference,
we
can
see
all
the
changes
that
have
been
made.
This
is
particularly
useful
for
developers
to
know
what
has
changed
and
why
it
is
optimized
how
it
is
optimized,
I
should
say.
Of
course,
there
are
still
preferences
like
the
space
here,
but
yeah,
maybe
we'll
change
it
later.
A
This
also
opens
up
the
possibility
to
integrate
it
in
a
merge
request
to
create
a
merge
request
with
these
changes
on
how
to
optimize
it
I
also
improved
a
bunch
of
bug
fixes
because
I
started,
collecting
Docker
files
and
I
figured
I
would
get
like
between
5
and
6
000
Docker
files,
but
it
turned
out
to
be
only
1600.
Ish
and
I
ran
doctor
against
all
of
those,
and
that
gave
a
lot
of
valuable
information.
It
was
able
to
parse
all
of
them,
except
for
one
one
had
a
merge
conflict
in
it.
A
Yeah,
that's
just
a
complete
faulty
photographer,
so
I
scored
him
against
doctor
I
also
didn't
try
to
build
them,
but
most
of
them
could
not
be
built
actually
partially
because
of
thoughts
not
being
present
or
invalid
instructions
in
the
docker
file
or
having
a
very
strange
setup
like
it
wasn't
as
straightforward
as
the
repository
is
complete.
This
is
the
command
how
to
build
it.
A
Probably
you
have
to
build
it
from
a
certain
subdirectory
or
from
the
root,
and
I
only
tried
building
it
from
the
root
and
without
and
then
I
also
ran
into
a
lot
of
issues
with
my
machine
with
using
Colima
or
an
out
of
memory
that
took
a
long
time,
but
I
was
able
to
build
around
300
Docker
images
for
now
and
I
scored
them
against
gripe,
which
is
a
container
scanning
tool
now
I'm
still
in
the
process
of
writing.
A
Wrapping
up
the
conclusions
and
notebooks
so
I
will
share
that
next
week,
but
spoiler
alert
I
need
way
more
data
in
order
to
make
anything
meaningful
with
machine
learning
other
than
finishing
up
the
notebook
next
week,
I'm
going
to
update
my
handbook
page
to
explain
why
I'm
working
on
Docker
and
how
it
fits
in
with
the
vision
of
AI
assist.