<maurer>
Well, it took me long enough, but I finally found the free time to add canonicalization (https://github.com/sandstorm-io/capnproto/pull/293) I look forward to being told how terrible my out of date C++ knowledge is :)
frigginglorious has joined #sandstorm
<asheesh>
: D maurer
<frigginglorious>
someone wanna play a quick match of a browser game where we shoot arrows at eachother?
jadewang has quit [Remote host closed the connection]
notevil has quit [Quit: Leaving]
wolcen_ has quit [Ping timeout: 276 seconds]
frigginglorious has quit [Quit: frigginglorious]
frigginglorious has joined #sandstorm
luckre_ has quit [Quit: No Ping reply in 180 seconds.]
luckre has joined #sandstorm
mnutt has joined #sandstorm
msoucy has quit [*.net *.split]
xobs has quit [*.net *.split]
dcb has quit [*.net *.split]
fkautz has quit [*.net *.split]
msoucy has joined #sandstorm
xobs has joined #sandstorm
dcb has joined #sandstorm
fkautz has joined #sandstorm
posi has joined #sandstorm
<posi>
We would pay big bucks for managed sandstorm in a gov cloud environment . We'd be happy if it was aws gov cloud that's what we use now
<posi>
I bet it's TAM of 500 million yearly
<kentonv>
posi: I would love to talk to you about this! :)
<kentonv>
posi: Can you email me at kenton@sandstorm.io?
jadewang has joined #sandstorm
<mrdomino>
wow, just saw cryptag. nice! anyone using it as a password manager yet?
<mrdomino>
oh, no real UI to speak of yet. still, neat.
mnutt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<frigginglorious>
mrdomino: i really need to start using a password manager. its one of those things i know i wont do until after its become an issue
<mrdomino>
fwiw github.com/habnabit/passacre + chrome password store as a fallback cache has been working really well for me
<digitalcircuit>
mrdomino: I've got cryptag installed.. but haven't actually started using it yet. I'll try to soon, for it'd be better than my current setup of a Bash script and GPG on a text file :)
<mrdomino>
passacre + cryptag would be a reasonable sub for passacre + chrome password store
coyotebush has quit [Ping timeout: 268 seconds]
OOCoder has joined #sandstorm
<OOCoder>
Hi, just curious, Is there a reason y sandstorm us kiwiirc vs gitter.im?
coyotebush has joined #sandstorm
XgF has quit [Ping timeout: 246 seconds]
jadewang has quit [Remote host closed the connection]
<BigShip>
Hmmm, would allowing the user to change the calendar name fix the problem? Maybe you could append a random number to the end of the calendar name when it's generated?
<BigShip>
Then you could keep one-doc-per-grain but grains wouldn't interfere with sync
<synchrone>
random sha-like strings don't make much sense
<synchrone>
i might get away with semantic names
<BigShip>
Anything that ensures each grain had a different name would probably do the trick
<maurer>
Does caldav not have separate user-friendly names and ids?
<synchrone>
maurer: it does, but 1) infCloud does not support setting them, and 2) #1629 is not solved yet
<maurer>
OK, so it seems like the "ideal" way would probably be fixing #1629 and getting infCloud to support that
<synchrone>
and i don't feel like implementing another infCloud hack to let the user do that, since syncing with grain name is much better
<synchrone>
i have a way to do a required PROPPATCH
<synchrone>
but i need that name from somewhere
<synchrone>
UI development with infCloud is a pain, and eventually I'd need to remove that feature when #1629 is fixed
jemc has joined #sandstorm
mnutt has joined #sandstorm
<BigShip>
synchrone: are you working on radicale then? I noticed that the couple of contact managers on sandstorm don't have import options. I was curious, are they a pain to implement?
sknebel has quit [Quit: sknebel]
sknebel has joined #sandstorm
home has joined #sandstorm
zeroish has quit [Remote host closed the connection]
home has quit [Ping timeout: 240 seconds]
<synchrone>
BigShip: I do. Radicale did not natively have this option, but it was easy enough for me to add support for, because it does not actually care about your contact format. It knows about *some* options, but generally stores whatever you give it
<BigShip>
Oh! Radicale has import @_@?
<synchrone>
since v9, I think
<synchrone>
radicale on sandstorm does, not radicale.org
<brocktice>
so the dir should have been gone entirely
<kentonv>
the weird permissions could make sense if your umask had somehow been set to 0077, but that doesn't appear to be the case
jadewang has joined #sandstorm
<brocktice>
I didn't change anything from default install
<brocktice>
Aside from installing vlan and setting up the vlan to pull a public
<brocktice>
(we keep a public IP vlan on the office network)
wolcen_ has quit [Ping timeout: 276 seconds]
<brocktice>
do all ports need to be open for a sandcats.io install to succeed?
<asheesh>
It depends what succeed means, and do you mean inbound vs. outbound?
<brocktice>
Because I realized I was missing 6080 and the 30kwhatever one partway through
<brocktice>
I meant the inbound ports, per docs I assumed I needed 80,443,6080,30025
<asheesh>
Sandstorm doesn't do any reachability test, so you can go run it in a way that isn't reachable, and fix ports after the fact.
<brocktice>
ah ok that's not it then
<brocktice>
fresh server is ready
<brocktice>
maybe I should snapshot this first...
<asheesh>
: )
canuky has joined #sandstorm
<asheesh>
BTW brocktice ... were you at cs.jhu.edu like 10 years ago?
<asheesh>
'Cause I was and your name looks super familiar.
<asheesh>
If not, that's OK, but I figure I might as well ask.
<BigShip>
asheesh, secretly trying to figure out where we all live =P
<asheesh>
So that I can send you all stickers!!!
<BigShip>
Ooo, giving up personal information for stickers is tempting...
<brocktice>
hahaha yes
<brocktice>
well sort of
<brocktice>
Institute for Computational Medicine
<asheesh>
I think I remember your username being the only other person who was actively pegging the CPUs on the ugrad.cs.jhu.edu machines, or something like that.
<brocktice>
Natalia Trayanova's lab
<asheesh>
"other" meaning I was too.
<brocktice>
oh... yeah I'd do that probably
<brocktice>
I was in an MPI class
<maurer>
asheesh: Wait, you're affiliated with jhu? Did you know Rich Ercolani?
<brocktice>
that's probably when
<asheesh>
maurer: Boy do I!
<maurer>
....small world
<brocktice>
I only used the cs machines for my MPI class I believe
<kentonv>
maurer: To be fair, Asheesh knows everyone.
<brocktice>
otherwise we had our own supercomputers :)
<asheesh>
Well that's hilarous. Nice to see you again, brocktice.
<brocktice>
likewise!
<asheesh>
And hi canuky !
<brocktice>
btw I ended up trying sandstorm because Meredith Patterson likes it
<brocktice>
Good news and bad news
<brocktice>
It worked on the new machine
<canuky>
hey @asheesh
<asheesh>
The bad news is, we don't really know how to repro? I guess I can try a double-install now and see if does the same weird thing, brb.
<brocktice>
so unfortunately I was unable to duplicate my problem yeah
<asheesh>
BTW you're not the only one to have mongo mysteriously fail to start.
<asheesh>
Someone dropped by a week ago, or so. So I'm glad that you showed up for long enough for us to get *some* information out of you.
<brocktice>
hah, well, I may have just taken you on a wild goose chase somehow
<brocktice>
really no idea what would have been different though
<asheesh>
If there's anything in your shell history that is useful, I hope you'll stare at it for a while and let us know.
<brocktice>
yeah let me look
<asheesh>
Glad that Meredith Patterson is sharing her enthusiasm!
<brocktice>
aha
<brocktice>
I did re-run it before completely wiping
<brocktice>
at least once
<brocktice>
not sure why
<brocktice>
however I would think the complete wipe would fix that...
<maurer>
kentonv: Can I take the fact that the description of far pointers says "located in some other segment" to indicate that far pointers to the same segment are illegal?
NOTevil has quit [Quit: +++ ATZ]
<kentonv>
maurer: I honestly have never thought about that.
<kentonv>
I believe the code as written will accept them.
<kentonv>
And I can't think of a good reason why it shouldn't...
<maurer>
kentonv: So, the thing I am worried about here is that if you are in a segment, and you follow the path to its arena, and it resolves to the same segment
<maurer>
the segment may be in an inconsistent state when you call the next method
<maurer>
In particular, I'm working on dwrensha's code (he wanted more of it to be safe) and trying to retrofit a raw pointer into a Rc<RefCell<Foo>>
<maurer>
the important distinction here is that only one piece of code can be mutating a refcell at a time, to prevent exactly this kind of thing
jemc has joined #sandstorm
<maurer>
(it doesn't want two &mut pointers to the same thing)
<maurer>
in the C++ code, I think it is probably in the "probably works, but it'd be easy to introduce a bug here" kind of situation
<maurer>
e.g. if segments had a lockable data structure like arenas do, it might be possible to produce a deadlock, etc.
<maurer>
As it stands, I can't see anything in the C++ code that indicates it'd have a bug
<maurer>
Actually, nvm, I'm wasting your time, I just came up with a workaround - I can just check to see if the id is the segment's own ID, and return itself if so rather than consulting the arena
<kentonv>
yeah, this seems like an implementation issue, not something that ought to be solved by an encoding spec change
<maurer>
Sure, I was mostly wondering if the spec forbid that case, because if it did I could just return an error and punt on it
<maurer>
which seemed easier, and plausible given the wording of the encoding spec