jadewang has quit [Remote host closed the connection]
neynah has joined #sandstorm
joshbuddy has quit [Ping timeout: 265 seconds]
jadewang has joined #sandstorm
<paulproteus>
kentonv, dwrensha, jparyani, zarvox -- talking with Ben of Known suggests to me that we should rename X-Sandstorm-Base-Path => X-Sandstorm-Session-Base-URL
<paulproteus>
+1 -1 ?
<paulproteus>
I realize this is bikeshedding-y but he said out loud to me, "Well, a path is very different from a URL. Is it a path or a URL?"
<paulproteus>
We could perfectly happy have sandstorm-http-bridge send out both headers for the time being, so the docs can tell people to do the latter and that is the behavior of a released version of Sandstorm.
<paulproteus>
In other news, Ben of Known is pretty happy with Sandstorm, and doesn't make faces or squint at our weirdness, which is always nice.
<paulproteus>
We also might get an IndieWeb Protocols driver out of him.
<zarvox>
paulproteus: really, it's protocol + host + port, which constitute an origin, so maybe it should be X-Sandstorm-Session-Origin
amyers has quit [Ping timeout: 244 seconds]
<paulproteus>
Origin sounds to me, "as a non-security-minded web developer", to be the source of something, so presumably it should be
<paulproteus>
X-Sandstorm-Session-Origin: Yes
<paulproteus>
As in yes, this originates from Sandstorm.
<paulproteus>
Admittedly I *am* a security-minded web developer, but I meet many people who are not.
<zarvox>
IDK, the terminology for Origin as part of URL is pretty well-established
<paulproteus>
If you want I can commission a usability study of PHP developers : P
<zarvox>
proto://host:port/path#fragment
<paulproteus>
brb let me do that
<paulproteus>
I guess I have to figure out how to phrase the question.
<paulproteus>
Maybe that's more trouble than I want get into by now.
<zarvox>
if the issue is "path already has a meaning in the context of URIs and we don't use that meaning" then I don't see how you could object to using the correct noun in the context of URIs ;)
<zarvox>
oops I meant to write scheme://host:port
<zarvox>
we could call it X-Sandstorm-Session-Scheme-Host-Port
<paulproteus>
I could live with that!
<paulproteus>
Also semi-ridiculous but I can totally live with it.
<paulproteus>
I would still prefer Url-Without-Trailing-Slash but OK.
<zarvox>
it's not a URL; it's missing the path and query string and fragment
<zarvox>
or rather, it's not *the* URL for the page being requested, so that would be confusing
natea has joined #sandstorm
<paulproteus>
(the absence of the path component is what "without-trailing-slash" indicates)
<paulproteus>
(an empty query string and fragment is I believe OK)
<paulproteus>
(I like to language-lawyer people who refer to bare domain names as URLs, too, by complaining that their URLs have no path component and are therefore wrong)
<paulproteus>
(I'm hiding in parentheses for some reason)
<zarvox>
so I don't find the addition disambiguating
natea has quit [Ping timeout: 252 seconds]
<paulproteus>
"Clearly you don't write enough PHP software"
<paulproteus>
I can see how if you're thinking in terms of "parsed URL data structure" then yeah, without-trailing-slash isn't clear.
<paulproteus>
If you are writing PHP and are "just" asking the question, "Do I need to put a / before index.php?" then "without-trailing-slash" lets you know "yes, I do".
<zarvox>
I think the best thing we could do here is give an example of what things would look like in our docs
<zarvox>
because the docs are the only place people are going to learn about these headers anyway
<paulproteus>
I like your "X-Sandstorm-Session-Scheme-Host-Port" quite well now, given that I think "trailing slash" is a PHP-world concept but not e.g. a Django-world concept.
<paulproteus>
And yes, I agree that examples++++++++++++
<zarvox>
and I actually had to look up what X-Sandstorm-Base-Path sends in the source to know myself <_<
<paulproteus>
I should go back to docs-ing
<paulproteus>
i,i read the ducks
<zarvox>
write the ducks
gopar has joined #sandstorm
<paulproteus>
Hi gopar !
<gopar>
paulproteus, hey! I saw your email. I'm interested in working with the proj
<paulproteus>
: D
<paulproteus>
I replied with some ideas of docs to read, and I can provide you links to any other things related to Web-CAT or Sandstorm that will help you on your way.
<paulproteus>
As you might have been able to tell from the email, I remain super excited about Web-CAT.
<paulproteus>
I think I might be more excited than upstream. (Upstream being a term for the main authors.)
<paulproteus>
I actualy think upstream got bored of Web-CAT around 2008, shortly after giving talks about how great Web-CAT is.
<paulproteus>
I don't know, since I haven't chatted with them.
<gopar>
ha yeah, I talked to katie and she said that there really wasent any on goign dev on web-cat for a while
<gopar>
ill go over the resources you put in the email
<paulproteus>
Awesome. It'll be my honor to answer any questions you have.
<paulproteus>
I can also sort of pave the way by testing things out before you do, gopar
<paulproteus>
Since the "Sandstorm platform" is very much a work in progress.
<paulproteus>
To make a good package of Web-CAT, it'll definitely some Java coding.
<paulproteus>
win 24
<paulproteus>
Yargh.
<paulproteus>
BTW gopar it's great to e-see you here. Do you have classes over the summer?
<gopar>
paulproteus, sweet thanks. It would definitely be awesome to have your help with this :) I don't have class but I work full time.
<gopar>
I need to set up an irc bouncer since the only time I can get on is around 8. Miracle i got home early
<paulproteus>
Great (-:
<paulproteus>
We don't have an IRC bouncer that's part of Sandstorm yet, but I'm hopeful we will soon.
<paulproteus>
At the moment, in case you're interested, the main thing msising is we need a "driver" for the IRC protocol. Sometime soon (1-2 weeks?) we'll publish documentation on how to write drivers.
<paulproteus>
Sandstorm is a pretty strange, but seriously great, system, in which apps that need IRC access to one server have a way to request precisely that from the platform, but for that to work, there has to be something that can take app IRC commands as input, and relay that to the broader Internet.
<paulproteus>
That's the role of drivers.
<paulproteus>
The nice thing is that apps that have IRC access to one server, even if they're compromised or malicious, can't speak to the rest of the Internet, and e.g. leak your personal data out willy-nilly.
<paulproteus>
(you could also decide you find this interesting enough to write an IRC driver! But you might want to wait until the docs exist, so that you know what you're getting into.)
<gopar>
paulproteus, awesome! seems like you guys are busy with sandstorm :) btw hows the meetup going?
<paulproteus>
Pretty well! We've only had one full meetup since the one you came to, though we had a mini-meetup which was at the Electronic Frontier Foundation office. We do have a tentative South Bay location for a ~6 weeks from now, so if you come to the South Bay one, it might be closer to you.
<paulproteus>
We have 105 members.
<paulproteus>
My secret personal stretch goal is for the Sandstorm meetup to have more members than the entire San Francisco Linux Users Group email list, and we only need about 200 more people to reach that.
<gopar>
paulproteus, Nice! yeah the south bay would be pretty cool. I go to the Python meetups near Mountain View, so if you could do it more down south then I would defintely try and go more often (and drag miguel with me ha)
<paulproteus>
: D
<paulproteus>
win 8
<paulproteus>
....
isd has left #sandstorm [#sandstorm]
<paulproteus>
I have pushed many changes to docs.sandstorm.io
<paulproteus>
I hope+intend to push more tomorrow/later.
<paulproteus>
BTW have I mentioned recently that docs.sandstorm.io uses GitWeb Pages by XgF . I am grateful, XgF!
joshbuddy has joined #sandstorm
joshbuddy has quit [Client Quit]
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
joshbuddy has joined #sandstorm
gopar has quit [Remote host closed the connection]
jadewang has quit [Remote host closed the connection]
<erikoeurch>
having a bit of trouble pushing a git repo to Gitlab on Oasis
<dwrensha>
tell me more. :)
<erikoeurch>
just making sure I haven't missed anything silly first :)
<dwrensha>
Are you using a URL that looks like this? https://<ANY_USERNAME>:<TOKEN>@api.oasis.sandstorm.io/gitlab/repo.git
<erikoeurch>
dwrensha: ah thanks, didn't realize I could put the token in there
<erikoeurch>
I did get it work though
<erikoeurch>
but having to enter the token/password manually
<dwrensha>
the right way to store the token is using one of git's credential helpers
<dwrensha>
but, yeah, it works to just shove it in the URL
<erikoeurch>
is there any security downside to having it in the URL? (when sending over HTTPS)
<dwrensha>
the downside is that it's easier for the token to get compromised on your end
<dwrensha>
if it's in the URL, then it's just stored in plaintext in .git/config
<dwrensha>
which might be readable by any number of people
<dwrensha>
if you use the osxkeychain credential helper, for example, then the token is protected behind your OSX keychain passphrase
<erikoeurch>
right, reading up on gitcredentials. (I'm using Ubuntu)
<dwrensha>
(which you'll typically enter only once, on login)
jadewang has joined #sandstorm
<paulproteus>
I still wish we supported SSH keys for auth to the git instances.
<paulproteus>
Sadly we don't support SSH inbound yet to app instances, so there's no straightforward way to do that yet.
<erikoeurch>
yeah, SSH key support would be handy
* dwrensha
tries to imagine how the would work
<erikoeurch>
how do I get the webkey I generated to display again? Feeling silly, but when I click the key icon and then the webkey name, it just offers me to change the name...
<paulproteus>
We could put the API token in the path, for routing, perhaps, dwrensha
<paulproteus>
I'm unlikely to do so in the very near future, but there's a random thought for you.
<dwrensha>
paulproteus: yeah, it might be that when I update gitlab I should move over to vagrant spk
<dwrensha>
paulproteus: but that's still something that sounds scary to me
<dwrensha>
because gitlab already has lots of moving parts
<paulproteus>
Yeah, it's not clear to me that migrating existing apps is going to be much fun for anyone, nor necessarily worth anyone's time. Although it might be, who knows.
<dwrensha>
paulproteus: yeah, I'm thinking gitlab should probably not be my *first* vagrant-spk app
<paulproteus>
But I think any tool like this is most useful for new things than for things that "could be migrated I guess but my current setup works OK".
<paulproteus>
FWIW we face a similar inertia hurdle getting people to migrate to our Etherpad package, even though Etherpad users likely want to use Sandstorm!
<erikoeurch>
just a question -- how are Sandstorm apps updated? Are there any docs about it?
<erikoeurch>
I mean when some particular app has a new version
<erikoeurch>
paulproteus, you're the docs master, right? :)
<paulproteus>
I _suppose_
<paulproteus>
I don't want to hold that title forever!
<paulproteus>
If you decide you want to try writing a doc in the docs/ directory and submitting a pull request erikoeurch I would be super grateful. If you don't have time, that's fine too!
<paulproteus>
"The Chaos Computer Club is, by its chapter and by common consent, a galactic organization of all life forms, regardless of their age, gender or upbringing." <https://events.ccc.de/camp/2015/wiki/Static:Volunteers>
<paulproteus>
I am learning things!
<erikoeurch>
paulproteus, could help out a little with the docs, but not in the very near future as I'm hard at work on things needed to graduate...
<paulproteus>
(-:
<paulproteus>
Where are you working on graduating from?
<erikoeurch>
Linköping University, not overly famous :)
<paulproteus>
Cool (-: I met some people from the Gothenburg Linux Users Group, I believe, while I was at FOSDEM.
<paulproteus>
I know it's not the same place, but it reminds me.
joshbuddy has joined #sandstorm
<erikoeurch>
well, it's not a very big country anyway :)
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
jadewang has quit [Ping timeout: 250 seconds]
jadewang has joined #sandstorm
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
jadewang has quit [Ping timeout: 265 seconds]
jadewang has joined #sandstorm
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
amyers has quit [Ping timeout: 240 seconds]
amyers has joined #sandstorm
<YuviPanda>
paulproteus: heard you are going to ccc camp!
amyers has quit [Read error: Connection reset by peer]
amyers has joined #sandstorm
<paulproteus>
Yeah! What about you?
<paulproteus>
Also how'd you hear that!??
pouledodue has joined #sandstorm
augustl has joined #sandstorm
<augustl>
hey folks. Trying to figure out how I can get SSL for my self hosted *.sandcats.io instance
<augustl>
are there any guides out there?
<patrickod>
didn't you mail nb-discuss about it ? :)
jadewang has quit [Remote host closed the connection]
amyers has quit [Ping timeout: 248 seconds]
jadewang has joined #sandstorm
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
englishm_ has joined #sandstorm
<paulproteus>
augustl: Hi!
englishm_ has quit []
<paulproteus>
augustl: w/r/t https -- you're going to need a wildcard cert for yourname.sandcats.io.
<paulproteus>
"Sometime soon/eventually" I hope to have something different to tell you.
<paulproteus>
But promises are sweet.
<mcpherrin>
I know somebody's asked about this before, but will something like Let's Encrypt be able to issue certs fast enough to not need a wildcard?
<mcpherrin>
Or would having certs with fixed names leak the secret magic juice?
<paulproteus>
They'd leak the secret magic juice, since Let's Encrypt will want to publish every hostname as soon as the issue a cert, and we want the hostname to not be easily discoverable public information.
amyers has joined #sandstorm
amyers has quit [Ping timeout: 244 seconds]
<zarvox>
from elsewhere:
<zarvox>
14:52 < staticvec> some rando is presenting about sandstorm.io at gophercon right now