00:26 <simonv3> I’m followed the steps for packaging a Meteor app for Sandstorm http://localhost:3000/

00:26 <simonv3> And I’m using accounts-sandstorm package https://github.com/sandstorm-io/meteor-accounts-sandstorm

00:26 <simonv3> Testing it on Meteor works fine

00:27 <simonv3> but when I open the app in Sandstorm (local.sandstorm.io:/6080) my Meteor.user() returns:

00:27 <simonv3> Object { _id: "5k5i9FdPo7tNHTYwt", hasCompletedSignup: true, isAdmin: true, signupKey: "devAccounts", loginIdentities: Array[1], nonloginIdentities: Array[0] }

00:28 <simonv3> which is not quite what I was expecting

00:28 <simonv3> Is there something else I have to do to get that working?

00:33 <simonv3> Meteor.users.find({}).fetch() returns 2 objects, the first is the one listed above - the second is a user profile for Alice Dev Admin, but under services there’s only {dev: {name: “Alice Dev Admin”}}. Don’t see permissions or anything like that :/

00:42 <simonv3> I’ll open an issue in that repo as well

00:51 <simonv3> https://github.com/sandstorm-io/meteor-accounts-sandstorm/issues/14

01:17 <asheesh> Howdy simonv3

01:18 <asheesh> Are you using the console in the right tab, fwiw?

01:20 <simonv3> asheesh: how do you mean?

01:21 <asheesh> brb

01:23 <simonv3> (this is from the javascript console in the browser, which I wasn’t sure would work, but all the standard Meteor things are available, and the test logs show as well)

01:23 <asheesh> Back

01:24 <asheesh> FWIW you should be aware that since the app is running inside an iframe, you need to specify *which* frame to see the JS console for.

01:24 <asheesh> https://stackoverflow.com/questions/3275816/debugging-iframes-with-chrome-developer-tools

01:24 <asheesh> You're going to need to choose the "app-iframe" from the dropdown shown in the first answer there.

01:26 <asheesh> simonv3: Let me know if the above makes sense. If not I should explain more. If it does make sense I can wander off!

01:29 <simonv3> It makes sense, though I’m not sure whether Firefox has the same thing, going to test in Chrome now

01:30 <asheesh> https://developer.mozilla.org/en-US/docs/Tools/Working_with_iframes for firefox

01:31 <simonv3> Yep, in Chrome the right user is shown

01:31 <simonv3> Still doesn’t do the right thing, but I guess the bug isn’t what I thought it was

01:31 <asheesh> (-:

01:31 <asheesh> "yay, a different error!"

01:34 <simonv3> yay for debugging

01:34 <simonv3> done for tonight though

01:38 <asheesh> simonv3: https://github.com/sandstorm-io/sandstorm/pull/1354 & thanks for asking

02:19 <mnutt> asheesh: do you know how I’d switch?

02:19 <mnutt> (blowing away all built dependencies, upgrading to clang-3.5, and forcing ekam to build with clang didn’t help)

02:34 <simonv3> asheesh: do you know if there’s a way to see the logs from the packaged app?

02:34 <simonv3> probably they’re locatable somewhere in the vagrant instance?

10:41 <asheesh> simonv3 has wandered off, but: they're usually in the "grain log", if we're talking about things that the app prints to stdout.

10:41 <asheesh> aka server-side console.log()

17:02 <dwrensha> I'm trying to think of a cute name for a Rust crate that will be the home of my port of kj::Exception

17:06 <dwrensha> "err" and "error" are not available, but "er" is.

17:07 <dwrensha> but I probably want something more googleable anyway

18:15 <XgF> dwrensha: Hmm, what does kj::Exception do that std::Error doesn't?

18:15 <dwrensha> std::io::Error is not Clone

18:15 <dwrensha> kj::Exception is way simpler

18:15 <dwrensha> and has a copy constructor

18:16 <dwrensha> errors being copyable is important for Promise::fork()

18:17 <XgF> Could you not just pass around std::error::Errors?

18:17 <dwrensha> that's a trait

18:17 <dwrensha> you could use Box<::std::error::Error>

18:20 <dwrensha> it seems like the usual thing that Rust library do is to define their own custom Error enum

18:21 <dwrensha> and impl ::std::error::Error for it

18:22 <dwrensha> they end up with hierarchies of errors wrapping other errors

18:24 <dwrensha> and as far as I can tell, it's rare to actually care about all dealing those cases, once you've caught an error

18:24 <dwrensha> *dealing with all of those cases

19:33 <mnutt> anyone have an idea about how I’d go about debugging a segfault when running `make` to try to build sandstorm from source? I’m running the sandstorm/debian-jessie64 VM and following the instructions from the installation page, getting a segfault that looks like it’s related to capnp

19:33 <mnutt> here’s the code: https://github.com/mnutt/davros/compare/sandstorm-from-source

19:34 <mnutt> (the idea being that you run `FROM_SOURCE=true vagrant-spk up` and get a VM where you can hack on sandstorm itself at the same time as your app)

22:39 <mnutt> here’s the backtrace of the segfault: https://gist.github.com/mnutt/9fdebec76bcc71304f19 my best guess is that capnp doesn’t like being built on a vboxsf filesystem

22:56 <simonv3> Is there a way to maintain session for non-logged in users by storing in local storage? Right now I’m noticing that my app is storing things in the <longhash>.local.sandstorm.io, but that changes with a page refresh

23:28 <kentonv> simonv3: currently there isn't any way.

23:29 <kentonv> arguably there shouldn't be any way for "incognito" users.

23:49 <mnutt> is there a reason that ekam is in deps, but you need to separately compile it in order to get network dashboard support? (for `make continuous`)