►
From YouTube: NixOS Office Hours, 2019-11-29
Description
Today we review some PRs, deploy ofborg and the aarch64 build box, and talk a bit about how you can get involved with the Nix community.
A
A
B
B
C
A
B
A
A
B
B
B
Cool
sounds
good,
there's
some
other
poor
request.
We
could
take
a
look
at
in
a
few
minutes,
but
and
if
anybody
out
there
who,
who
is
a
longtime
member
of
the
community,
wants
to
help
merge,
affordable
requests,
let
you
know
and
I'll
grant
people
access
very
happily
to
them
all
right.
Would
you
like
to
take
a
look
at
a
different
poor
quest
and
I'll?
Try
and
get
the
patches
for
the
word
infrastructure
pushed
sure.
A
A
So
I
do
actually
have
a
pending
review
here
that
went
through,
and
the
thing
is
that
the
peph
print
I
happen
to
there's
an
update,
but
it
happened
to
murrs
APR
that
added
a
fork
of
liver
print
to
next
package,
isn't
sort
of
embedded
it
into
the
same
expression
and
now
do
I
think
about
that
I.
Don't
really
think
I
should
have
merged
that,
because
having
Forks
like
that
and
NYX
packages
isn't
really
the
best
thing.
It's
not
really
the
best
situation,
in
my
opinion
like,
and
it
even
adds
on
some
code.
A
So
it's
like
basically
an
override
that
happens
in
NYX
packages
that
someone
could
just
do
in
an
overlay
now
that
it's
actually
maintained
with
someone
I'm
not
going
to
like
go
out
of
my
way
to
remove
it.
So
I
didn't
mention
that
here
because
it
is
going
to
be
maintained
by
them
so
I
just
but
I
just
as
leave
it
that
way,
but
I
think
it
might
actually
be
ready
to
merge.
I
mean
look
over
one
more
time
so
yeah.
So,
let's
see
here
it
just
lifts
it
off
into
that
on
variant
package.
A
And
this
is
just
a
simple
update,
which
won't
I
will
note
that
actually
didn't
look
at
the
changelog.
So
one
thing
when
you're
reviewing
an
update
PR
is
to
try
your
very
best
to
find
if
there's
a
change
log
somewhere
anywhere
that
may
have
one,
you
should
probably
find
it
and
read
it
because
you
don't
know,
what's
actually
gonna
happen
or
the
intended
side-effects,
all
the
changes
or
dependencies
changed
and
stuff
like
I
can
see.
They
either
go
object.
Introspection
here,
which
means
there's
maybe
I,
could
share
a
binding
now
or
there
wasn't
before.
A
A
It's
just
a
beta
release,
which
is
like
I.
Actually,
the
pull
request
to
merge
like
a
an
unstable
version
of
fall,
config,
so
I
can
get
tested
before
release,
because
that's
a
really
difficult
thing
to
get.
Organized
and
honestly
I
actually
probably
merged
data
release
to
this
before
so
I'm,
probably
not
going
to
spend
it.
A
A
And
also
Graham
I
think
you
noticed
this
you
to
me
last
time,
I
reviewed
EPR
is
that
I,
probably
don't
pay
attention
to
all
for
as
much
as
I
should
sometimes
I
may
forget
to
check
if
all
the
checks
past,
sometimes
that's,
probably
because
there's
no
button,
that's
like
delay,
merge
until
checks
have
passed.
If
that
was
there,
it
actually
probably
helped
me,
but
that's
more
of
a
github
thing
than
anything.
Yeah
I
think
there's
actually
a
like
a
github
browser
plugin
that
lets.
You
gives
you
a
button
like
that.
A
B
B
A
B
A
B
A
Update
PR,
is
it
like?
The
messages
will
not
be
Corrections
like
they
did
say
they
read
the
contributor
markdown
so
doesn't
say
it's
checked,
but
I
kind
of
feel,
like
it's
a
little
bit
difficult
to
understand
like
if
you
look
at
the
submitting
changes
here.
This
looks
a
little
bit
kind
of
like
it
could
really
to
any
tongue
kind
of
interpretation,
and
my
first
thought
was
like.
Maybe
it's
not
so
maybe
like
language
friendly,
possibly
like,
maybe
for
people
that
English
is
not
their
native
language.
A
Actually
I,
like
that
pretty
much
like
Graham
has,
you
know,
crit
turned
himself
into
you
know
about
partially
he's
taken
part
of
his.
You
know
existence
until
about
he's
not
off
fork,
so
if
he
can
assimilate
every
single
PR,
so
I
think
for
world
of
peace.
We're
gonna
need
like
a
world
of
peace
pot,
eventually
just
review
everything
automatically
so
I
want
you
think
about
that
idea.
A
A
They
can
they
should,
but
then
again
it
is
just
editing
a
commit
message.
So
I
don't
really
sweat
it
well,
it
seems
they
also
put
their.
They
use
the
master
brunch,
which
I
always
feel
kind
of
like
a
bit
like
weird,
like
I've
pushed
to
that
might
want
to
stop
screen
sharing
here
for
me
to
type
in
my
PGP
password.
B
I
ended
up
pushing
to
change
the
work
infrastructure,
repository
and
so
I
mentioned
before
that
it's
really
safe
to
do
this
dry
acted,
and
indeed
so,
not
that
one,
how
about
this
one?
So
it
did
that
build
and
deploy,
but
it
was
just
the
dry
activation.
So
even
though
it
was
out
of
date,
it
didn't
do
anything
so
that
was
fine.
I
didn't
need
to
cancel
it.
Our
stress
on
this
new
one
I
did
the
exact
same
thing.
I
did
direct
the
dry
build.
We
can
see.
B
A
That
they
don't
have
like
they
allow
maintainer
to
push
to
branch
enabled
so
I
will
mention
if
you're
like
thinking
of
contributing
to
mix
of
us.
So
that
really
does
help
maintain,
is
to
be
able
to
just
fix
up
your
own
brunch,
really
quick
and
merge
it,
because
this
way
I
actually
had
to
push
directly
to
master.
A
B
B
A
A
A
A
So
now
I'm
thinking
about
that,
perhaps
I
should
wait
for
someone
to
do
that.
But
the
thing
is
someone
I've
actually
reviewed.
This
placard
several
times
is
that
it
has
not
been
tested
by
many
people
because
I
don't
think
a
lot
of
a-kor
contributing
body
house
like
a
fingerprint
reader
or
something
so
it
has
gone
untested,
not
sure
what
to
do
about
that.
I
think
I
might
just
merge
it.
Does
a
Lib
F
print
have
a
registered
maintainer,
yes,
Abra
dark!
Well,
if
that's
just
him,
and
he.
B
A
B
A
B
Box,
this
machine
has
like
I,
think
it
has
50
or
100
cores
or
something
they're
kind
of
slow
course,
but
it's
a
lot
of
cores,
so
it
makes
up
for
it
and
it's
pretty
easy
to
get
access
you
just
have
to
one.
We
I
definitely
want
to
recognize
who
you
are,
but
you
have
to
understand
and
read
the
security
problems
with
using
this
builder
in
that
a
trusted,
but
malicious
acts
are
could
hack
your
system.
B
If
you
actually
trusted
anything
that
came
out
of
this,
let's
see
Daniel
Fullmer,
so
I
don't
recognize
their
name,
but
Samuel
Dr
suggested
that
he
asked
anyway,
so
I'm
just
gonna.
Take
a
look.
I
see
that
they
added
that's
nice
to
see
that
for
use
and
you
baqia,
probably
alright.
This
looks
fine
to
me
so
I'll
click
merge,
I've
contracted
with
them
before
cool.
B
B
Build
box
so
I'm
a
pretty
big
fan
of
built
kite,
so
we'll
jump
over
to
the
build
kite
build.
We
see
that
it's
actually
already
building
this
isn't
ideal
because
it
kind
of
breaks
things,
so
we
will
cancel
both
of
these
and
start
a
new
build
I'm
seeing
scheduled,
builds,
have
failed
for
some
time.
So
maybe
this
one
won't
work
either.
B
The
first,
the
first
step
it
does-
is
it
instantiate
s'
the
Nix
OS
config
in
two
derivations.
It
then
copies
it
over
to
a
remote
builder,
which
is
art
64.
It
does
the
build
on
the
remote
art
64
builder
and
then
uses
some
trickery
to
copy
the
netboot
I'm
that
Budaj
pixie
ramdisk
and
the
kernel
over
to
a
net
boot
server,
and
then
it
just
hopefully
I'm
not
showing
an
API
key,
probably
not.
It
uses
the
packet
net,
API
pack,
economy,
API
to
issue
a
hardware
reboot,
and
that's
supposed
to
happen
once
week.
B
B
B
B
C
Because
I
know
it
was
waiting
for
like
several
weeks
multiple
times
and
the
author
has
always
like
quickly
responded
to
reviews.
So
now
it's
again
been
waiting
for
two
days
without
any
comments,
so
maybe
I
thought
it
would
be
nice
to
ya,
get
to
get
an
answer
again
and
have
you
have
you
tested
this
poor
request?
C
B
B
A
They're
not
quite
trivial,
sort
of
Laurel
moments,
they're
kind
of
really
trivial
and,
like
your
descriptions
not
proper,
or
this
is
not
a
proper
way
to
wrap
or
something
like
that-
that's
not
really
useful
information
mm-hmm.
So
it's
kind
of
like
trippy
on
the
pics,
maybe
I'll
message
Freddy
in
the
background.
A
B
B
So
I
can
share
my
screen
again
about
that
build,
so
this
build
seems
like
maybe
it
might
end
up
succeeding.
It's
so
right.
So
the
the
netboot
process
creates
a
root
file
system
using
squash
FS,
which
is
part
of
the
RAM
disk,
and
then
that
gets
mounted
with
a
read,
write,
Union
FS
layer
on
top
there's
some
additional
trickiness
that
gets
rid
of
the
Union
FS
later
on,
but
so
this
squash
FS
contains
the
everything
that's
needed
for
the
system
to
boot
and
then
once
the
once,
this
is
complete.
B
The
reboot
downloads,
the
squash
FS
and
the
kernel
off
this
web
server
and
then
boots
off
of
it.
So
every
single
boot
is
completely
pristine,
except
for
just
a
few
secret
credentials
which
are
stored
on
a
certain
partition
on
the
disk.
I
was
hoping
it
would
finish
by
now,
so
it
would
be
interesting
to
look
at,
but
it
doesn't
look
like
it
will
we're
on
a
piece.
Let
me
know
when
you're
ready
to
share
something
yeah
I
have
a
PR.
B
A
That
we've
cut
pango
at
like
sort
of
an
older
version
and
actually
I
think
at
some
point
was
an
insecure
version
unstable
because
it
breaks
a
super
deprecated
on
binding
P
by
GT
k,
which
is
it's
completely
archival
and
maintained
by
mind
you-
and
there
was
quite
a
few
packages
in
Nick's
packages
that
use
that
and
I
did
actually,
because
this
package
needs
to
actually
be
removed.
I'm
pretty
sure
Debian
has
already
eliminated
it
or
will
be
eliminating
it.
This
January.
A
So
I
have
been
doing
with
that,
and
so
the
biggest
issue
was
a
lot
of
these
applications
are
either
like
some
of
them.
They
may
continue
to
work,
but
there's
nothing
underlying.
Are
anything
that's
going
to
allow
them
to
continue
to
work
because
pango
as
it
makes
changes
it's
just
going
to
continue
to
break
p
by
g
TK
and
it's
python
to
mind.
Use
of
all
so
that's
already
has
to
go,
so
I
create
a
pull
request
to
remove
a
lot
of
those
packages.
A
I've
been
working
on
that
a
bit,
but
it's
kind
of
difficult,
because
some
of
the
packages
have
Forks
that
are
now
poured
the
Python
3
and
go
object,
introductions
bindings,
but
some
dough.
So
it's
a
bit
difficult,
tricky
situation
when
people's
packages
are
going
to
just
get
removed
and
I
don't
want
it
to
come
off
that
way,
like
I'm,
just
going
to
abandon
someone's
package
because
I
think
it's
not
maintained
it's
because
py
g
TK,
the
as
a
library
is
it
maintained
and
in
its
state,
is
going
to
become
defunct
very
soon.
A
It's
because
of
its
existence,
so
it
seems
it's
progress,
is
probably
better
to
be
MERIS
precisely
mechanic
message
before
I
give
it
another.
Look
so
Jordan
are
said,
my
collaborator.
It
looks
good
to
me,
except
the
Contessa
jizz,
the
Pengo
one
should
be
that
and
the
py
JDK
commit
should
be
squashed,
and
the
commit
party
should
mention
that
is
for
contact
with
tango,
144
I.
A
A
A
A
A
That
is
that
reason,
because
that's
quite
a
strong
severity
for
a
user,
especially
your
package,
this
all
the
Sun
is
gone,
so
I
would
only
go
start
from
lares
the
highest
severity
of
what
can
be
removed
and
then
what
things
are
broken
and
what
things
are
maintained
should
be
allowed
to
work
until
next
release.
So
that's
what
I'm
going
to
do
with
this
migration.
A
B
A
Sending
the
maintainer
seeker
role-
yes,
we
do
certainly
need
some
sort
of
needs
to
maintain
our
package,
because
I
know
if
these
packages
were
in
a
different
district
I
would
have
been
marked
as
dead
and
gone
right.
A
lot
of
them.
One
thing
I
learned
last
week
is:
there
are
some
companies
who
can't
who
have
very
strict
rules
about
talking
about
what
they
used
publicly.
B
And
actually,
some
important
packages,
some
packages
to
them
that
were
really
important
to
them,
were
removed
from
NYX
packages
because
they
didn't
have
a
maintainer
and
they
wish
they
had
had
the
opportunity
to
go
and
and
take
up
maintain
ership,
but
they
didn't
know
that
was
happening
until
packages
that
had
actually
been
deleted
from
NYX
packages,
so
I'm
when
thinking
a
little
bit
about.
How
could
we
get
people
like
them
to
speak
up
in
times
like
this?
That
is
actually
a
tricky
situation.
You
did
say
they
couldn't
actually
tell
what
they
were
using.
B
A
Would
be
a
part
of
the
removal
process
if
I'm
not
certain
is
that
it
would
be
a
process
to
find
someone
to
maintain
this
package
until
the
deprecation
path
is
just
to
completely
remove
it.
So
if
we
had
some
sort
of
a
process
for
that,
they
could
atop
themselves
on
one
of
those
steps,
so
they
could
be
a
part
of
it
because
currently,
it's
just
sort
of
like
up
to
the
decision
of
a
certain
kinetic
okay,
this
is
going
to
be
gone
and
that's
not
really.
That's
not
a
very
open
sort
of
attitude.
B
A
A
So
Simon
leader
said.
Maybe
we
need
a
process
that
includes
the
recurring
announcement,
I
thought
of
something
similar,
because
I
think
there's
like
a
RFC
for
like
I,
think
it's
like
for
moving
maintainer
as
to
like
sort
of
an
Emirates
team,
and
there
was
a
discussion
of
notifications
about
that,
and
so
we're
keeping
up
with
I
also
thought
about.
Maybe
in
a
further
change
we
could
keep
up
with
maintainer
that
are
missing
or
not
present
or
not
responsive
to
requests
update
the
package
so
having
some
outs.
A
A
B
C
A
When
it
comes
that
when
I
removed
the
packages
in
this
blow
request,
I
actually
added
a
throw
aliases
that
it's
been
removed
instead
of
just
suddenly
being
gone,
so
that's
a
little
bit
better.
I
do
wish
that
was
sort
of
a
documented
process,
because
I
have
been
sort
of
instructing
people
to
do
this
regularly.
So
no
documentation
idea,
that's
a
cool
idea,
so
it
turns
out
we're
at
45
minutes
yep.
A
B
B
A
A
B
B
Always
interested
in
feedback
on
what
you
all
think
about
the
format
and
how
all
of
this
works
there's
since
this
is
started.
Zimba
tim
has
begun
a
weekly,
Nick's
Friday
event.
You
can
learn
about
that
as
in
CIMB,
ATM,
comm,
slash,
Nick's,
Friday
I
think
that's
really
cool.
You
get
one-on-one
teaching
from
him.
At
the
same
time,
petty
has
started
doing
hackage
and
Haskell
update
streams.
Every
Friday
Friday
has
become
a
big
day
for
the
fornix
in
the
video.
What
trendsetters
really
yeah?
It's
super
cool,
but
so
yeah.