<zarvox>
I would guess that what he's hoping for/expecting is a Davros feature that produces HTML indexes like nginx or apache might for all the folders
<zarvox>
er, s/he/canuky/
<asheesh>
Oh, possible.
jacksingleton has quit [Quit: Leaving.]
cted has joined #sandstorm
cted has quit [Client Quit]
cted has joined #sandstorm
<cted>
Trying out the Rocket.chat demo, sent an invitation to someone to test w/ me, & we're both logged in but don't see each other. Any ideas?
<neynah>
cted did you both join a chatroom? for example #general
<cted>
Yes, we both joined #general.
<neynah>
And you both appear online? Do you see yourself joining the chatroom?
<cted>
Yes. We each see ourselves joining the room, but we don't see each other joining.
<dwrensha>
How did you send the invitation?
<cted>
Email via the "Share access" think in the top bar.
<cted>
A second person I invited just appeared in the room, though.
<zarvox>
When you type chat messages, do you see each other's chat?
<zarvox>
(and does the second person see your chat, the other person's chat, neither, or both?)
<cted>
Person 2 sent me a direct message & I sent them a regular message. ...
<cted>
OK, I think we're set. Person 1 joined the wrong way.
<dwrensha>
How did they join?
<cted>
Through a Rocket.chat iphone app they downloaded. (!)
<cted>
Once they just followed the email link they got to the right place.
<asheesh>
Fascinating.
<cted>
Haha, yeah. So all's well, just some confusion. Thanks!
jadewang has quit [Remote host closed the connection]
<mnutt>
I'm finding it really hard to test core sandstorm changes. I'm building sandstorm on a linux box and it's working ok but in order to test I need an spk, and all existing spks are compiled with their own sandstorm-http-bridge.c++
<mnutt>
so I try to get an app compiled on this machine, but the app is set up for vagrant-spk and I can't figure out how to get it working with just spk
<kentonv>
mnutt: yeah you'll probably want to use raw spk, not vagrant-spk
<kentonv>
then you'll be able to kill and re-run "spk dev" each time you iterate
<mnutt>
right, it just so happens that my app makes a bunch of vagrant-spk specific assumptions
<kentonv>
might need to write a little test app I guess
<asheesh>
There's probably some sweet spot combination of these; fwiw, you can use "plain spk" with a "vagrant-spk" app, probably, too.
<mnutt>
(sandstorm doesn't seem to like a symlinked/hardlinked /opt/app)
<asheesh>
Try bind mount instead of symlink
<asheesh>
I love bind mounts as the world's weirdest symlink.
<mnutt>
progress!
<asheesh>
'vagrant spk' "just" ssh's in and does 'spk dev'
<asheesh>
You can steal the logic right out of the Python script.
<mnutt>
(I didn't get to work on my webdav ci test suite yet, but it seemed high-value enough to the loomio folks to go ahead and submit)
<asheesh>
!! ++ !!
<asheesh>
BTW mnutt are you 'make continuous'-ing, and 'make update'-ing?
<asheesh>
Curious what workflow you've settled on. A lot of what we use for workflow is hearsay from Kenton and you can't hear that as loudly as I can.
<mnutt>
yeah. I got into a weird state where `make update` wasn't working, I think it may have been caused by running the test suite?
<mnutt>
`make update` would say it was restarting sandstorm, but sandstorm's {{build}} wouldn't change on the about page
<kentonv>
just one small comment on the PR
<kentonv>
oh I guess zarvox's comment seems legit, too
<mnutt>
just saw the two comments, updating now
<mnutt>
asheesh: I just finished building a linux box with an i7 4970k, it sure beats building inside virtualbox on an old MBP
<asheesh>
Oh wow, that is quite a CPU.
<mnutt>
and trying to build apps...I had one using npm that took literally 25 minutes to build before, and often just died
<mnutt>
so...current setup is ssh'd into the linux box from my mac, using dnsmasq
<zarvox>
my usual workflow, FWIW, is `make continuous` on the Linux host, a vagrant-spk vm with the app under development with an additional shared folder for /sandstorm-git, running /sandstorm-git/shell/run-dev.sh /opt/sandstorm
<zarvox>
if I do something with spk, I `make bundle` and run spk out of bundle/bin/
<asheesh>
Whoa can I do that with Sandstorm too!??
<asheesh>
bundle/bin/sandstorm --config-dir /opt/sandstorm or similar?
<zarvox>
Not so much. Sandstorm assumes some things about paths relative to the executable.
<asheesh>
But I guess I could 'sudo cp bundle/bin/sandstorm /opt/sandstorm/latest/sandstorm
<asheesh>
'
<zarvox>
I tried symlinking /opt/sandstorm/latest to bundle once; it didn't go very well
<asheesh>
Yeah, as mentioned before, I don't really believe in symlinks.
<zarvox>
but copying probably works
<asheesh>
Don't tell all my vagrant-spk friends about that though!
<zarvox>
symlinks are great I don't know what you're talking about ;)
<zarvox>
I do kinda wish there were a vagrant-spk backend that ran on the host directly, debootstrap'd a chroot instead of a VM, and used filesystem namespaces and bind mounts to set up /opt/app and /opt/sandstorm, and drop the virtualization layer entirely
<asheesh>
Yeah, +1 to that
<zarvox>
or at least, to have the option to do that on Linux hosts
<asheesh>
I think we're not *so* far from getting there fwiw.
<zarvox>
The canonical setup would still be VM-based, because we still love our Windows and OSX users
<zarvox>
mnutt: thanks for the PATCH patch! ^_^
<mcpherrinm>
+1 to no vm :D
mnutt__ has joined #sandstorm
<mcpherrinm>
(but vagrant-spk is amazing)
<zarvox>
<3
<asheesh>
How's whatever you were working on during the meetup I forget what it was going?
<zarvox>
oh mcpherrinm how goes the feed reader package?
<asheesh>
Right
<asheesh>
Feed eater
<asheesh>
Freed eader
<asheesh>
Anyway.
<mcpherrinm>
zarvox: I didn't end up getting any work done since I left the meetup :)
<mcpherrinm>
I was just about to try and make a hilariously janky HTTP proxy
<asheesh>
If you want an impromptu hack session sometime this weekend or next week I am +1, mcpherrinm, fwiw
<kentonv>
we have too many users. This migration is taking forever.
<asheesh>
Maybe there's a way to do migrations in a way that doesn't take down the service. A thing to investigate when it's not nearly 00:00.
mnutt has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
mnutt__ is now known as mnutt
<kentonv>
asheesh: if we didn't have to take down the service for updates anyway then I would make people start writing migrations that can happen live
<asheesh>
++
<asheesh>
FWIW given today's status, it seems like you should start doing that anyway (-:
<kentonv>
well it only took an extra 5 minutes
tobald has joined #sandstorm
<asheesh>
Oh okay.
<asheesh>
Still, I guess 5 min will add up and remove some nines!
<kentonv>
anyway, mailing list bonus storage is live!
<asheesh>
Yay!
<kentonv>
you can now marvel at how your quota live-updates when you subscribe/unsubscribe to the mailing list outside of Sandstorm
<asheesh>
Sweet I have 1 GB extra now
<asheesh>
Bah I can't figure out a way to cheat it by changing my primary email address and getting more bonus storage.
<asheesh>
Too bad.
<kentonv>
yeah it gives you the bonus if any of your addresses are subscribed
<kentonv>
but only once
<kentonv>
um I guess you can cheat it by adding a throw-away address to your account and subscribing that!
<asheesh>
Yeah, that's clear. I wanted multiplicative cheating!
<dwrensha>
synchrone: are you interested (1) changing in non-UI-viewable settings, or (2) app-wide settings that would affect all of your grains of a particular app?
<synchrone>
non-ui-viewable, per grain settings
<dwrensha>
synchrone: is there a particular app that you think this would be useful for?
<synchrone>
It was about setting the default timezone per calendar in my radicale bundle
<synchrone>
i've found a way to detect that, but it's not guaranteed to work in all browsers and the app itself does not preserve it inside
<dwrensha>
is this a setting that's configurable on non-sandstorm Radicale?
<synchrone>
nope
jeddi has joined #sandstorm
<dwrensha>
So it sounds like you could implement this setting by adding a form in the UI somewhere and persisting the value to a file such as /var/default_timezone . Would there be any problem with that approach?
<dwrensha>
simonv3: ouch, re the subscribe notification. What browser are you using?
<simonv3>
dwrensha: firefox
<simonv3>
I’m now wondering whether it’s being blocked by an adblocker or ghostery or privacy badger, i’ll toggle those off
<dwrensha>
simonv3: have other notifcations worked for you? like when you get an app update?
<simonv3>
Yeah
<simonv3>
(doesn’t look like there’s any blocking happening)
<simonv3>
looks like it works fine in chrome
<simonv3>
which has removed my ability to debug this, sorry :/
<dwrensha>
heh
<dwrensha>
what specific version of firefox, on what operating system?
mnutt has joined #sandstorm
<simonv3>
dwrensha: FF 44.0 (the most recent) and Mac OS X
<simonv3>
Yosemite
dwrensha has quit [Remote host closed the connection]
dwrensha has joined #sandstorm
wolcen has joined #sandstorm
<dwrensha>
wow, looks like Oasis signup is broken right now
<dwrensha>
I tried creating a new account, and I get a 500 when I try to click on the free plan.
frigginglorious has quit [Quit: frigginglorious]
<dwrensha>
simonv3: I am able to reproduce that "event is not defined" error on our testing server
frigginglorious has joined #sandstorm
<simonv3>
dwrensha: cool, let me know if I can be of some help otherwise
<synchrone>
dwerensha: I have an issue. I started to package the app without looking at the sourcecode, but after taking a closer look at the source code. I did not expect to fiddle around with the source much. After realizing I'd need to fork it and patch some things ... in a word, I would like to abstain from touching it's UI at all, if possible.
<dwrensha>
I wonder whether this particular issue could be handled by Sandstorm tracking the timezone of its users and then somehow forwarding that data to apps.
<synchrone>
that'd be too much features to the platform
<synchrone>
so far this works: var globalTimeZone=Intl.DateTimeFormat().resolvedOptions().timeZone
<mrdomino>
so guys, the whole cookies-needed-for-urls thing. is it always gonna be like this? is there a way around it? like i wanna share individual wiki pages with people who maybe aren't logged in or are named wget
|jemc| has quit [Quit: WeeChat 1.3]
|jemc| has joined #sandstorm
<mrdomino>
maybe this conversation has already happened a ton, in which case don't mind me
<dwrensha>
mrdomino: if you use the "share access" button you can generate a URL that will work for anyone
<dwrensha>
you'll need to manually append the path onto the generated URL
<synchrone>
but isn't it a better fit to use static publishing for wiki pages?
<mrdomino>
oh i didn't know about the append-the-path thing
canuky has joined #sandstorm
<canuky>
Question: Someone attempts to signin to our Sandstorm server via their email ... I see their 'handle' in Admin:Users ... is there a sandstorm log where I might find their full email in order to contact them?
<mrdomino>
so yeah this reduces to "it'd be nice to have readable/typable URLs" and "apps should be able to expose capability URLs" by my reading at least
<dwrensha>
mrdomino: oh, actually the /shared/ linked generated from the "Share Access" tab still require cookies
<dwrensha>
canuky: right now, I think your best approach is to use `sandstorm mongo` to manually query the database :\
<synchrone>
i've just created static html files, set nginx to serve them as well and ... the request does not seem to get through nodejs proxy
<dwrensha>
synchrone: what is that pastebin showing?
<synchrone>
a non-random but still weird binary output when i'm trying to access my app's static resource
neynah has joined #sandstorm
<dwrensha>
canuky: db.users.find({"profile.handle": <handle>}), look in the "services" field
neynah has quit [Client Quit]
<canuky>
Perfect - thanks dwrensha
<synchrone>
I don't understand what's wrong with my app, so that some resources work ok, but other requests are not let through, and sandstorm's log is empty...
<synchrone>
ok, not empty, but there's nothing about this 404, and I made sure the file is in the vagrant-spk VM
<technician>
For DNS I'm using the *.sandcats.io dynamic IP thingy
<asheesh>
technician: Hmm! That should work. I maintain the sandcats.io dynamic IP thingy.
<asheesh>
Interesting, technician
<asheesh>
Would you be willing to tell me your server name and/or provide a sharing link to a grain here?
<technician>
IMed you
<zarvox>
also I agree it's sad that nginx can't access_log anywhere useful - nginx only gives you file paths and syslog, and we disable /proc and /dev because they are chock full of kernel attack surface
<asheesh>
"obviously" we should have a mode where nginx can access_log to /dev/stderr optionally.
<zarvox>
Or we should implement a syslog shim that reads messages that would be written to syslog and prints them to stderr, and then all apps get "syslog" support by default.
<asheesh>
Right, yeah. +1
<asheesh>
Or we can have it send it to another grain called sysblog
<simonv3>
Hey, I have now twice gotten a cloudflare-esque page for sandstorm saying that i’m viewing a cached version of the site and that I should try again for the live site
<simonv3>
Dunno if that’s something you all should be concerned about or not, or whether you know!
<asheesh>
CloudFlare and Sandstorm updates probably get along poorly.
<asheesh>
kentonv: ^
chance672 has quit [Quit: Leaving my desktop..... laptop possibly?]
<Tcf>
hey asheesh
<kentonv>
simonv3: yeah sorry, it happens whenever I push an emergency update. There were a couple bugs introduced last night that I had to fix.
<Tcf>
I just got a cloudflare
<Tcf>
but I got out of it
<Tcf>
:)
<Tcf>
I invited 20 new people to Sandstorm to discuss the future of decentralized hosting and music distribution ;)
<asheesh>
: D
<Tcf>
Is it possible to have a bot inside rocket chat?
<asheesh>
Not yet, I believe.
<asheesh>
There's no straightforward way to connect the Rocket.Chat to a bot. But it's in the works, as part of "powerbox."
<simonv3>
TCF ooh, tell me more. I’m kind of regularly in touch with the folks working on resonate http://resonate.is/
<simonv3>
a sandstorm bandcamp alt would be rad
<asheesh>
simonv3: BTW I've been listening to "mrdomino radio" all day.
<Tcf>
simonv3 I will check out that
<Tcf>
I can invite you to the sandstorm chat?
<simonv3>
TCF: Sure!
<simonv3>
asheesh: I’m not sure what that is! link?
<Tcf>
kentonv I got cloudflare all over, not inside an app
<Tcf>
refreshed the whole browser
<kentonv>
TCF: Did you intentionally refresh or did it somehow auto-refresh?
<Tcf>
I first got some sort of connection timeout
<Tcf>
then I refreshed
<Tcf>
and I got cloudflare
<mrdomino>
asheesh: if that happened, then i'd be unable to fix the settings
<kentonv>
TCF: By "some sort of connection timeout" do you perhaps mean that an app complained about being unable to connect? What app?
<mrdomino>
i did not anticipate "reset-oauth" meaning "removing the client id and secret for the oauth provider"
<dwrensha>
mrdomino: you would still be able to do `sandstorm admin-token` to access the admin settings
<mrdomino>
oh right
<mrdomino>
ok i probably could've recovered then
<Tcf>
it was a message on the top of sandstorm
<mrdomino>
and then yeah i would've seen the error message on login
<Tcf>
so not sure if it's the app or sandstorm. It went quick
<kentonv>
TCF: Did the Sandstorm top bar turn red?
<Tcf>
yes
<Tcf>
and it said connecting in ...
<Tcf>
then it just reset the counter
<Tcf>
and I refreshed
<kentonv>
when the topbar turns red it should say "Sandstorm has been updated, click here to refresh"
<dwrensha>
mrdomino: what did you hope that `sandstorm reset-oauth` would do?
<kentonv>
TCF: But it's supposed to not say that until the update is complete, so it's weird that cloudflare complained at that point. Hmm. Thanks for the info, though!
<mrdomino>
dwrensha: require everyone with an oauth session to re-log-in
<Tcf>
kentonv it did say that eventually
<mrdomino>
oh it was after i restored from a backup
<Tcf>
but some cloudflare inbetween
<mrdomino>
and i was sorta naively concerned about replay attacks without any concrete idea of what was going on
<dwrensha>
mrdomino: ah. we do in fact have "log out all users" buttons in /admin/settings
<mrdomino>
basically that was the night i wanted to make sure my backups worked the hard way :-p
pdurbin1 has joined #sandstorm
rollig has joined #sandstorm
gillisig has quit [*.net *.split]
ckocagil has quit [*.net *.split]
pdurbin has quit [*.net *.split]
sugarpuff has quit [*.net *.split]
donught has quit [*.net *.split]
rolig has quit [*.net *.split]
_iwc has quit [*.net *.split]
rollig is now known as rolig
gillisig has joined #sandstorm
ckocagil has joined #sandstorm
sugarpuff has joined #sandstorm
sugarpuff has quit [Changing host]
sugarpuff has joined #sandstorm
ckocagil has quit [Ping timeout: 250 seconds]
ckocagil has joined #sandstorm
_iwc has joined #sandstorm
wolcen has quit [Quit: Leaving]
donught has joined #sandstorm
neynah has joined #sandstorm
kecors has quit [Quit: Leaving]
<Tcf>
kentonv I did discover something interesting in this process though
<kentonv>
oh?
<Tcf>
as a new user, I would create a link to share. I see that I can relabel the links
<Tcf>
But I cannot set specific rights for users in the panel?
<Tcf>
It does say: Kenton has access
<Tcf>
but I cannot just change that to Kenton can edit
<dwrensha>
that's what I'm currently working on :P
<Tcf>
:)
<Tcf>
Another thing that can get confusing is if you create several links; how do you know which one is which? by labelling only?
<kentonv>
yeah that's why it's a good idea to label them
<Tcf>
:)
<Tcf>
got that
<kentonv>
the link secret actually isn't stored on the server -- only a hash is, so we can verify if someone has the right link, but the link cannot be extracted from the database.
<kentonv>
hence we can't display the link itself in the "who has access" table
<Tcf>
I see!
<Tcf>
So labelling is very important
<Tcf>
As I cannot revoke links now as they are public
<Tcf>
could you combine links?
<Tcf>
or is that impossible in relation to what you are saying above?
<Tcf>
Just some thoughts as I now dont dare to delete my links ;)
<dwrensha>
what would "combine" mean?
<kentonv>
I suppose it would be possible for the UI to group links together, such that you only see one entry even though there are multiple links under the hood.
<dwrensha>
ah
<kentonv>
we are actually planning something like that, where there would be a button to "get this link again", but it actually generates a new link that is grouped with the old one
<mnutt__>
yes, a bug report would be greatly appreciated!
<mnutt__>
the PROPFIND (directory listing) is actually returning a 403
<zarvox>
I'm seeing an error on the JS console
<mnutt__>
the /api/permissions call shows the "view" permission, so that's strange.
<zarvox>
which I will check in Chrome which might give me better stacktraces
<zarvox>
Ahhh, I see. It's the 403.
<canuky>
sorry, typo - yes, 403
<mnutt__>
oh, that's not great. I see the issue, and it's a screwup on my part. canuky sorry about that, I'll have a fixed version (0.15.3) up tomorrow
<zarvox>
that was pretty quick.
<zarvox>
:)
<canuky>
Right on! Love Davros - looking to use it with a group of educators
<mnutt__>
(the share has a "view" permission, but "view" is just for web UI and does not include webdav's "PROPFIND". We used to have a special way of getting directory listings, but have since started using webdav PROPFIND directly from the browser)
<simonv3>
kentonv: sorry for the slow response. Sadly I don't remember what I was doing exactly, but I was trying to access something via a URL I think, rather than through browsing (re cloudflare)