21:35 <isd> Would folks be interested in a sandstorm port of the firefox sync server?

21:36 <isd> Something I've mentally toyed with before.

21:43 <TimMc> Interesting idea.

21:50 <samba__> uh

21:50 <samba__> nice idea

21:58 <ocdtr_web> isd: The Firefox Sync Server has been a longstanding wishlist item for me, and something I submitted to the app committee voting list a long time ago. :)

21:58 <ocdtr_web> Currently, I've placed a $100 bounty on it being ported and published to the app market. http://sandsheep.com/bb

22:00 <ocdtr_web> While perhaps not enough on it's own to justify the developer time of such a port in itself, hopefully it's a decent incentive/thank you for anyone who wants to tackle it.

22:03 <ocdtr_web> isd: You may also find it useful to note that someone described a bit of what work would be entailed here: https://github.com/mozilla-services/syncserver/issues/72

22:05 <isd> ocdtr_web: good to know, thanks. I figured the auth stuff would be finicky.

22:08 <ocdtr_web> I think rfk may have been suggesting using the Sync Server on Sandstorm, without hosting Firefox Accounts Server in some of that, but I'm not sure. I'd assume a Sandstorm app would just include both pieces of software.

22:08 <isd> That had been my thinking

22:08 <ocdtr_web> "just". Sorry, asheesh, even though you're not here.

22:09 <ocdtr_web> But yeah, the biggest question is if Firefox itself will be able to login to a server hosted on Sandstorm without heavy modification.

22:09 <strugee> where *is* asheesh?

22:09 <ocdtr_web> Because nobody's going to want to have to run a browser fork to get sync working.

22:09 <isd> Agree.

22:10 <isd> We'd either need to figure out how to avoid needing client-side changes, or push any changes upstream.

22:10 <ocdtr_web> strugee: "Hey all, I'm going to take a break from IRC for some days." -asheesh

22:11 <ocdtr_web> I imagine upstreaming changes to Firefox itself to support Sandstorm is probably really hard.

22:12 <strugee> ah, gotcha. good for him

22:12 <strugee> isd: or figure out how to "fix" that behavior in an extension

22:13 <isd> ocdtr_web: yeah, that's probably more than I'm up for.

22:15 <ocdtr_web> The Firefox Sync page mentions WSGI, which I am kinda assuming from my lack of knowledge is a protocol Sandstorm would need to support for this to work.

22:16 <isd> ocdtr_web: wsgi is just the standard python web-app middleware thing

22:16 <isd> think of it as a python-aware cgi

22:17 <isd> any python app in the app market is already going to be using it.

22:17 <ocdtr_web> As I said, "my lack of knowledge". :)

22:18 <ocdtr_web> I really should learn Python sometime.

22:24 <isd> It's a tool I'm glad to have on my belt, even though I'm not terribly in love with the language.

22:25 <ocdtr_web> I am pretty glad I know PHP. I'm not in love with it either. :D

22:26 <zarvox> FWIW I looked not-very-hard into supporting the firefox sync server at one point, but I think I punted on it once I realized I would have to land patches in Firefox

22:26 <ocdtr_web> I think there's actually a strong correlation between the languages people "don't love", and the languages that run basically everything and are universally pretty useful.

22:27 <zarvox> also I think you'd have to set up a firefox auth server, which was a bit undocumented? or use Mozilla's

22:27 <ocdtr_web> zarvox: Yeah, you'd have to run both Firefox Accounts Server and Firefox Sync Server in the Sandstorm app, presumably.

22:27 <zarvox> and if you wanted to set up your own FxA server then you also had to about:config the FxA server URL

22:28 <ocdtr_web> It'd be silly to still use Mozilla's auth server, while doing a self-hosting thing.

22:28 <zarvox> and basically that was all more yak-shaving than I wanted to deal with at the time :P

22:29 <ocdtr_web> I'm okay with the about:config, long as we don't have to do a fork. My question is... what code would need to be added to Firefox? It would seem to me if the auth and sync servers were both running in the Sandstorm app, and Sandstorm supported whatever network protocol it used... where would the issue be?

22:30 <zarvox> Last I checked the "firefox talking to the data server would have to learn how to speak bearer tokens for Sandstorm to route the requests to the appropriate backend" was the big issue

22:30 <strugee> ocdtr_web: I'd say it was unideal but not "silly"

22:30 <strugee> still better to own your data even if you don't own your auth

22:30 <zarvox> it's possible that since we made HTTP Basic auth work for api hosts, that might be feasible now with username:password in the URL

22:30 <strugee> though admittdly it's encrypted anyway so I suppose in a sense it isn't that much of an improvement

22:31 <zarvox> (maybe the user-agent would have to get adjusted? not really sure)

22:32 <isd> https://github.com/mozilla-services/syncserver readme suggests it encompasses the whole server side thing now. I seem to recall that not being there last I looked.

22:35 <ocdtr_web> Changelog doesn't indicate anything changed since the issue discussion above.

22:39 <isd> No promises: https://github.com/zenhack/firefox-sync-sandstorm

22:41 <isd> Ug, pythonpaste.org is *still* down.

22:41 <isd> I'm thinking it may just be unmaintained at this point

22:42 <isd> which is kind of a pain given how much stuff out there uses it (incl. syncserver)