ELLIOTTCABLE changed the topic of #elliottcable to: a _better_ cult ˙ ͜ʟ˙ embrace, extend, extinguish.
<glowcoil>
french press are hard
sarcasticsimba is now known as vigs
<ELLIOTTCABLE>
11:22 AM <+devyn> so essentially I think Executions now just become graph containers (statement groups)
<ELLIOTTCABLE>
devyn: that's what #18 is about, exactly.
<ELLIOTTCABLE>
that phrase basically refers to what I see as one of the possibilities: discarding executions. specifically, the move-forward semantics.
<ELLIOTTCABLE>
but I've got issues with that, huge ones.
<ELLIOTTCABLE>
so trying to come up with alternatives; although I still don't like many of them very much. /=
<ELLIOTTCABLE>
devyn: did you understand what I wrote on the issue?
<ELLIOTTCABLE>
agreed re: the fact that unstaging was a terrible idea from the get-go for managing ‘arguments.’
<ELLIOTTCABLE>
it's a hack that was trying too hard to avoid adding a new semantic to the language.
<ELLIOTTCABLE>
but I never did find a good replacement that I liked. maybe one will fall naturally out of the new semantics we haveh ere.
<ELLIOTTCABLE>
devyn: `foo [a b ; c d]` doesn't impose *any* ordering constraints *between* the first and second sub-chains of nodes
<ELLIOTTCABLE>
it's identical to `foo &[a b ; c d]`, except the first node of the last sub-chain within the body becomes data-dependant on the outside, instead of purely-dependant like the rest.
<ELLIOTTCABLE>
if you *don't* want `a b` dependant, there's no reason to put it in the brackets at all.
<ELLIOTTCABLE>
so, basically, given an existing node X, if you want to add a *new* node that is purely dependant on that node, you either …
<ELLIOTTCABLE>
(call the new node Y)
<ELLIOTTCABLE>
1. if there's an existing directly-data-dependant node (i.e. `X a …`); you insert a new &[] clause afterwards that contains Y and its' descendants:
<ELLIOTTCABLE>
X &[Y …] a …
<ELLIOTTCABLE>
2. if there's an existing indirected-data-dependant node (i.e. `X [a …]`); you prepend Y-etc to the data-dependency chains
<ELLIOTTCABLE>
X [Y… ; a …]
<ELLIOTTCABLE>
3. if there's no existing indirected-data-dependant node, but there are already purely-dependant nodes, you prepend onto those.
<ELLIOTTCABLE>
(i.e. starting with `X &[a …]`)
<ELLIOTTCABLE>
X &[Y …; a …]
<ELLIOTTCABLE>
I'm using your syntax, by the way; but I'm not sold on &[] vs () yet.
gq has joined #elliottcable
gq has quit []
devyn has quit [Read error: Connection reset by peer]
devyn has joined #elliottcable
devyn has quit [Read error: Connection reset by peer]
devyn has joined #elliottcable
<devyn>
what the fuck? I upgraded my kernel and my `dmesg` is colored now
<devyn>
ELLIOTTCABLE: no, obviously, [a b; c d] is done in parallel. was just thinking whether `foo [a b; c d] bar` should wait upon both `a b` AND `c d` to complete before `bar`, or just `c d` and then let `a b` spin off.
<devyn>
joelteon: yeah
<joelteon>
try nixos
<joelteon>
you still get edge stuff
<joelteon>
but atomic rollbacks
<devyn>
that's not a feature I have ever needed and I enjoy Arch's structure
<joelteon>
you've never needed rollbacks?
<devyn>
nope
<joelteon>
man i wish i was you
<joelteon>
ok well anyway, nixos is cool, but so is arch
<devyn>
IIRC, nix provides a way to easily configure a system though, right? from like a config file for the whole thing?
<joelteon>
yeah, one config file manages every service
<joelteon>
though you can of course modularize your config and do imports
<joelteon>
e.g. joelt.io is defined by two nix expressions
<devyn>
okay, I might look into using that for a server/backend setup I'm doing soon
<joelteon>
one for the basic system config, one for the linode connection data
<joelteon>
i use the system config one and a separate config to spin up a copy in virtualbox
<devyn>
I can't really have nix as the main OS but I can run it in a container... on the virtualized Debian, I guess, haha
<joelteon>
it's like mixins for operating systems
<joelteon>
it is nice. on the plus side, no mucking around with dhcpcd
<devyn>
I've never had to muck around with dhcpcd, either lol
<purr>
lol
<devyn>
always use wicd or connman or something
<joelteon>
ok, well
<joelteon>
email server
<joelteon>
anyway, clearly you've never used online.net
<joelteon>
don't use online.net
<devyn>
haha, ew
<joelteon>
it's cool in theory
<devyn>
nah, we actually have to use a VPS hosted in Canada for this project
<joelteon>
20 euro for an 8 core dedicated server
<joelteon>
and ipv4 works just fine
<joelteon>
but the way they do ipv6 is retarded
<joelteon>
also, i don't like sshing to france
<devyn>
which up until recently sucked hard but now there's an OpenStack-based provider in New Brunswick & Nova Scotia
<devyn>
so basically everything Rackspace provides
<joelteon>
yeah don't go with online.net
<joelteon>
well unless you speak french
<devyn>
heh
<joelteon>
canadians do that right
<devyn>
only on the east coast, and only if you're not in a hardcore anglophone area
<joelteon>
well and all the guides for how to set up your system on online.net are written in french
<devyn>
on the west coast we're basically the same as the rest of the west coast
<joelteon>
what about the middle
<joelteon>
are there are people in the middle
<joelteon>
*any people
<devyn>
the middle ranges from rednecks to just really lame praries that aren't that interesting
<joelteon>
so it's like america
<devyn>
yes, anything seriously of note is concentrated more toward the coasts
<devyn>
which makes sense, 'cuz logistics
<joelteon>
but it's pretty in the middle right
<devyn>
I guess that depends on what you define as pretty
<devyn>
if you're interested in mountains, no, not really
<joelteon>
yellowstone, the mountains in wyoming, salt lake city...
<devyn>
the middle's pretty flat
<joelteon>
damn
<devyn>
most of the mountains are on the west coast
<joelteon>
never going to the middle of canada then
<devyn>
they have fun accents though
<devyn>
like in Manitoba
<devyn>
kinda like Mississippi I guess
<joelteon>
yeah that sounds like a place that would be in canada
<joelteon>
for sure
<devyn>
provinces of Canada: British Columbia, Alberta, Saskatchewan, Manitoba, Ontario, Quebec (the very French province), Newfoundland & Labrador, New Brunswick, Nova Scotia, Prince Edward Island
<devyn>
frenchies are concentrated toward the east
<joelteon>
but anyway, nixos
<joelteon>
nixops is not like any of the other deployment systems
<devyn>
I'm about to flash my router with DD-WRT so I probably won't be around and may miss some messages so
<devyn>
put that on hold
<devyn>
:p
<joelteon>
ok
<joelteon>
i was about to go on a tirade
<joelteon>
good call
devyn has quit [Ping timeout: 264 seconds]
devyn has joined #elliottcable
devyn has quit [Ping timeout: 255 seconds]
devyn has joined #elliottcable
devyn has quit [Ping timeout: 260 seconds]
devyn has joined #elliottcable
sharkbot has quit [Ping timeout: 246 seconds]
devyn has quit [Ping timeout: 260 seconds]
devyn has joined #elliottcable
eligrey has quit [Quit: Leaving]
sharkbot has joined #elliottcable
gq has joined #elliottcable
gq has quit []
<devyn>
ELLIOTTCABLE: anyway I think we're heading in the right direction
<devyn>
ELLIOTTCABLE: I believe if we go with this the way I'm envisioning it, it actually won't really be a huge change from Paws as far as usability/abstractability and even the general interface go… but it will be much more sound and logical
<devyn>
ELLIOTTCABLE: and perhaps my strongest argument for a prefix is that its meaning is pretty straightforward: the only difference between `item` and `&item` is that the latter doesn't affect the data stream at all. but both have order dep.
<devyn>
ELLIOTTCABLE: so it's effectively a discard operation, or rather, don't-wait-for-data.
<devyn>
`foo [a b; c d] bar` and `foo &[a b; c d] bar` are almost identical, except that the latter explicitly says "we don't care about the data, but still wait for order as usual here."
<devyn>
and while I'm aware that that works nicely with [] vs. () as well, it doesn't work so nicely with named references.
<devyn>
I think `foo #baz bar` vs. `foo &#baz bar` is much better than `foo #baz bar` vs. `foo (#baz) bar`, because the latter to me seems to totally violate expectations
<devyn>
I would expect `(#bar)` to have behavior like `&[#bar]`, not `&#bar`
<devyn>
the only way I can think of is that would be a special case, and definitely a weird one IMO
Sorella has joined #elliottcable
nuck has quit [Ping timeout: 264 seconds]
nuck has joined #elliottcable
Sgeo has quit [Read error: Connection reset by peer]
prophile has joined #elliottcable
alexgordon has joined #elliottcable
prophile has quit [Quit: The Game]
gq has joined #elliottcable
alexgordon has quit [Ping timeout: 264 seconds]
eligrey has joined #elliottcable
eligrey has quit [Read error: Connection reset by peer]
eligrey has joined #elliottcable
<joelteon>
people are fucking stupid
<joelteon>
so i just went into ##mac and asked "all my HTTP connections are timing out, i went to activity monitor and i'm getting 900,000 packets/s in, maybe that has something to do with it, what's going on?"
<joelteon>
and they're like "ask IT, something must have changed"
<joelteon>
they say "can you ping google.com?"
<joelteon>
and i say "no, timeout"
<joelteon>
"oh, well the networking guys must have just added port security"
<joelteon>
"oh is that why i'm getting packet flooded?"
<joelteon>
"what? no"
prophile has joined #elliottcable
<joelteon>
so apparently, IT has changed something in the network between last night and today so that my computer, and only my computer, can't access anything outside the network
<joelteon>
and coincidentally, i'm also geting DoSed
<joelteon>
but that has NOTHING to do with it
katlogic has quit [Ping timeout: 250 seconds]
katlogic has joined #elliottcable
Sorella has quit [Ping timeout: 260 seconds]
Sorella has joined #elliottcable
Sorella has quit [Changing host]
Sorella has joined #elliottcable
vil has quit [Quit: :qa!]
vil has joined #elliottcable
gq has quit [Quit: Connection closed for inactivity]
prophile has quit [Remote host closed the connection]