kentonv changed the topic of #sandstorm to: Welcome to #sandstorm: home of all things sandstorm.io. Say hi! | Have a question but no one is here? Try asking in the discussion group: https://groups.google.com/group/sandstorm-dev
<crab>
ocdtrekkie: what's this "gitweb pages" you were referring to on the call?
<DanC>
did Kenton make it to office hours on Jan 12?
ocdtrek_web has joined #sandstorm
<ocdtrek_web>
DanC: Yes
<ocdtrek_web>
Also https://sandstormcommunity.org is my potshot at providing a quickly editable place for organizing community efforts, making grain links for it mildly findable, scheduling Office Hours, etc. Opinions welcome.
<DanC>
the github wiki has the advantage that it's backed by git, so you can edit offline etc.
<DanC>
did those present Jan 12 discuss a next meeting?
<ocdtrek_web>
Based on the poll I did, right now we're just doing the same scheduled time on Sunday afternoons.
<DanC>
when copying stuff to sandstormcommunity.org , please be sure to leave a link to where it came from. Especially in case the original gets updated
<ocdtrek_web>
(Not my favorite time, by any means, but it won the poll.)
<DanC>
I'm aware of the poll data; I'm interested to know whether those present discussed a future meeting.
<ocdtrek_web>
'Tis a good idea. Though IMHO after an Office Hours is concluded, hopefully nobody is going back and editing the notes, because people are unlikely to find it.
<DanC>
in particular, did anyone offer to organize an agenda? Did Kenton confirm?
<DanC>
Let's try to get a clear owner for the next step on each topic. I see "Dan Krol can take point on contacting companies" ; should I expect to hear from him about Cloudron.io ? within the next week?
<ocdtrek_web>
Well, right now you have to not only go into the wiki, but expand the list of all pages, and know what you're looking for. ;)
<ocdtrek_web>
This site wasn't discussed/proposed during the meeting, I assembled it mostly as an experiment, FYI. Nobody's committed to it.
<ocdtrek_web>
I presented Cloudron.io's offers as an example of something we could do. They're a direct "competitor" of Sandstorm, so contacting them would be silly. ;)
<ocdtrek_web>
Nobody has explicitly laid out an agenda for next week at this point, I think. There are details about what different people are going to attempt to do, and presumably each week we should see what progress people have made.
Mitar has quit [Ping timeout: 268 seconds]
Mitar has joined #sandstorm
<DanC>
right... but if one person in particular looks over things and sends out an agenda around T-24hrs, that can help a lot.
<DanC>
interesting that Kenton prefers 1pm Sunday to 4pm Saturday. I wonder what he would think of 1pm Saturday
<DanC>
meanwhile, I wonder how many $9/month subscriptions are available to fund development. I don't see evidence of very many.
<DanC>
re Perkeep in sandstorm ... (a) Perkeep is unfortunately kinda anti-ocap. Its API is identity-based; once you login, you can enumerate all the objects.
<DanC>
and (b) perkeep is only interesting (to me) if it provides redundancy. How would a sandstorm grain (or set of them) provide redundancy/
<ocdtrek_web>
DanC: Are you going to be available next Sunday, and would you want to kick off an agenda grain so we can assemble the agenda in advance? Also, I sent you a PM.
<DanC>
I'm no more available next Sunday than any other. "I'd like to avoid Sundays." -- yours truly, Jan 6.
* DanC
heads to a meeting...
ocdtrek_web has left #sandstorm [#sandstorm]
ocdtrek_web has joined #sandstorm
<ocdtrek_web>
Webchat wasn't showing me anything you wrote. :/
<ocdtrek_web>
If you responded to my PM, I'm sure it didn't show me that either. :/
<ocdtrek_web>
At least here I can use Whitequark.
<ocdtrek_web>
For me, that Sunday 1 PM time isn't exactly ideal either. I had proposed it to narrowly avoid church without torching the rest of my day when someone wasn't available Saturday. But it torches my after-church schedule with the family.
<ocdtrek_web>
Lyre never voted up in the poll, I wonder whether or not it would tie up the ideal-voted time. If we have enough interest in multiple times, we could maybe alternate back and forth to ensure more people can be involved.
xet7 has joined #sandstorm
<ocdtrek_web>
I probably can't make this weekend's either way probably.
<Guest43589>
<ocdtrek_web "Lyre never voted up in the poll,"> Wait, there was another poll? I'm a bit behind here. Link?
<ocdtrek_web>
Other weekday presents a fun and exciting problem where we then have to drill down and figure out which weekday. :P
<ocdtrek_web>
But it might be good to have those as the two options for like an alternating schedule meeting or something.
<DanC>
I did reply to your PM, ocdtrek_web . hm.
<DanC>
"Other Weekday After 7pm" is sorta cheating. If it gets chosen, then what? we still have to pick which day.
<DanC>
let's see if there's enough on the agenda to make it worth further scheduling pain (for me)... have we made any progress on the cron PR? who has the ball? Kenton? Did he say any more about when he could look at it?
<ocdtrek_web>
Kenton is going to try to look at the Cron PR. I think most of the question left on it is regarding testing. Also there are a couple other issues blocking use of the Cron API even once implemented, that Ian designed fixes for, but Kenton needs to review.
<ocdtrek_web>
The Freenode webchat appears to be utter trash. :/ If you want to respond to my PM, just send it to ocdtrekkie and I'll see it later.
<ocdtrek_web>
The options in the poll were mostly just "what people had suggested on the email list". And yeah, if we actually want to do any weekday meetings that would need further discussion. (Didn't expect it to win!)
<ocdtrek_web>
DanC: Yes it is.
<DanC>
ocdtrek_web, I replied to your PM by mail
<DanC>
I'm struggling to find where Kenton said something about reviewing the cron PR the week of Jan 11; does anyone else remember that?
<ocdtrek_web>
Got that. Freenode's webchat is so bad that I can type in it right now, and you can see it. But to see any responses, I have to switch tabs to Whitequark and look at the logs.
<DanC>
oh. my email archives contain very little of sandstorm-dev...
<DanC>
"One big reason not to use GitHub issues right now is that nobody can clean it up/manage it yet. " <- there's a straightforward fix for that
<ocdtrek_web>
DanC: xet7 is the maintainer of Wekan (and may be interested in doing frontend contributions to Sandstorm). JamborJan is no longer actively using Sandstorm.
<DanC>
ok, so that's another relevant github org with a bus number of 1
<ocdtrek_web>
Well, apparently mine's set to private, but I'm actually part of SandstormPorts too.
<DanC>
ah
<ocdtrek_web>
It was mostly just a place to host forks of Sandstorm app ports that people wanted to try updating apart from the original app porter. The only one AFAIK that led to an updated release is WordPress.
<ocdtrek_web>
Most Sandstorm apps have a bus factor of 1 due to the app publishing keys. (Apart from the small pile of them I've amassed. :/) App replacements requires of course, a Sandstorm release to replace the key, if we want to circumvent the original publisher in the app market.
<ocdtrek_web>
By the way, this is the other big issue pending the Cron PR: https://github.com/sandstorm-io/sandstorm/pull/3176 Specifically that the way most apps are built for Sandstorm, they can't actually call the Cron API even once it's merged.
<DanC>
"Spec out a solution to #3027" is a pretty obscure title. :-/
<ocdtrek_web>
DanC: Yes, but you can read issue 3027. ;)
<DanC>
all this would be more clear to me with concrete examples. ideally, pull/3176 would include a test case to demonstrate
<DanC>
in any case, the only path forward is thru Kenton? noone else is in a position to help?
<DanC>
why circumvent an app publisher? App ids are big random numbers, right? anyone can make one of those. It's not like "wordpress" is taken in the app market, is it?
<ocdtrek_web>
DanC: AppIds are a public key. In order to publish to them, you must have the private key.
<ocdtrek_web>
In this way, you can be sure an app update is either from the same developer or someone that app developer has entrusted their key to.
<DanC>
right, but I can do my own port of wordpress and publish it without any cooperation of the private key holder
<ocdtrek_web>
Someone can rekey the app and release their own version, but people won't see it as an update, and won't be able to update their existing grains to the new version without hackery. The replacement key methodology basically tells Sandstorm to treat one key as a valid replacement for another key.
<DanC>
sure; it's not an update. It's just a new app.
<ocdtrek_web>
So the two paths to updating an app abandoned by the author is to either ask them to pass on their private key for the app, or to add a PR to Sandstorm designating a replacement key, which of course, generally comes with some scrutiny.
<DanC>
path #3: don't. i.e. let the abandoned app be abandoned and publish a competitor.
<ocdtrek_web>
Or yes, just to publish anew and make people start over with their grains and such.
<DanC>
"Do you, Programmer,
<DanC>
take this Object to be part of the persistent state of your application,
<ocdtrek_web>
From the app market standpoint, I don't think we'd want to approve a WordPress 4, WordPress 5, and WordPress 6 that doesn't let you update between them, if at all possible to avoid.
<DanC>
though in this case, it's "do you, grain owner, ..."
<DanC>
i.e. portable data formats are a Good Thing.
<DanC>
but yeah, it sucks to have to restart your grain
<DanC>
sandstormports could set up some Shamir secret sharing stuff.
<ocdtrek_web>
Re: only path forward thru Kenton, the answer would be that Kenton arguably is Sandstorm's BDFL, so unless there was a lot of interest in hard forking the project, renaming it, etc. Yes.
<DanC>
typically the BDFL isn't required for every movement forward on every PR.
<ocdtrek_web>
I do really like apps that use simple/standard data formats. And many do use txt files or single sqlite files in the background. But in order to switch from appId A to appId B manually, you'd have to download your grain backup, edit the metadata to use the new grainId, and then reupload your backup.
<ocdtrek_web>
EtherCalc is a single JSON file as its data portion.
<DanC>
yes, well, all grains are a single string of bits :)
<DanC>
that is: JSON is not, by itself, a widely supported spreadsheet format
<ocdtrek_web>
The point where Kenton becomes the critical point is design questions. Which is where #3176 is. Ian designed a solution, but it's not really worth him putting a lot of effort into implementing the design if Kenton isn't going to approve of how it's implemented.
<ocdtrek_web>
A lot of what Ian is working on is really fundamental code to how Sandstorm interacts with apps, so it's pretty hard to undo once it's in place and apps start using it. And as an additional point, very few of us have any idea how that works to begin with. (It's mostly just magic to me.)
<DanC>
really? that seems pretty inefficient. it seems to me that it's cost-effective to put quite a bit of effort into fleshing out at least one, if not more than one, solution to any substantial design decision before it can't make further progress without the BFDL
<DanC>
"very few of us have any idea how that works to begin with" <- THAT is my point. If everything has to go through Kenton all the time, we're hurtin. But before an issue is blocked on Guido or Linus, any number of lieutenants has typically weighed in on it
<ocdtrek_web>
Cost-effective for the project, perhaps. Cost-effective for the freelance developer who could presumably use that time to make money on other projects, probably not.
<DanC>
if there really are so few people that grok this issue, then I'd be inclined to put some effort into knowledge sharing; that is: have Kenton and the developer discuss the PR with a number of other devs present so that next time some other devs and do more of what Kenton does
<ocdtrek_web>
DanC: Sure, it'd be nice to have more people who understand that part of the platform. But right now we don't have any.
<ocdtrek_web>
Linux has not just a lot of people who use it, but a lot of corporations who literally pay people full salaries to work on it full-time. It'd be nice to have that luxury. ;)
<DanC>
at some point in its history, linux was no larger than sandstorm
<DanC>
granted, Linus didn't get hired to do something else for most of his time
<DanC>
sandstorm raised $500k at one point. It could be that the market window is just passed. ("you never get a second chance to make a first impression") but I'd like to think that it hasn't.
<ocdtrek_web>
I would arguably say Sandstorm is an easier sell today than it was in 2014.
<ocdtrek_web>
When Sandstorm launched, and I told people Google was a dangerous entity to entrust all your data to, people called me a conspiracy theorist or a nut or whatever.
<ocdtrek_web>
In 2019, I think most people would agree peoples' perspectives have changed a fair bit.
<DanC>
one problem is that sandstorm is an app platform, and >80% of the work on such things is ugly compatibility crap; not the sort of thing that attracts hobby developers
ocdtrek_web has quit [Remote host closed the connection]
<DanC>
I note the sandstorm logo appears there; is that a licensed use?
<DanC>
Jade Wang is still listed on https://sandstorm.io/about ; is she still around? (I found this while looking for copyright / trademark holder info)
<DanC>
hmm... the indiegogo contributors list is available
ocdtr_web has joined #sandstorm
<ocdtr_web>
I think the reason I left my membership as private was I didn't want the logo there to misrepresent me as being part of Sandstorm.io.
<ocdtr_web>
I still need to modernize my doorbell.
<ocdtr_web>
Also most of my house. I moved recently, and a lot of the electrical I want to upgrade is a bit outside my skill level, so I need to set aside some money for an electrician to spend a couple days there.
<DanC>
hmm... how did she do this with twilio? "data from inside my home stay inside my home, and specifically not live in someone else’s servers."
<ocdtr_web>
Yeah, I wouldn't use Twilio for anything in my house personally. But it looks like the end goal there was to send a text, and it's mostly impossible to do without a third party.
<ocdtr_web>
Well, unless you email your phone number, which usually works.
<ocdtr_web>
I believe there are some strict rules about handling and retaining SMS messages though, so I'm not sure Twilio can store the data from them.
<ocdtr_web>
But like, I know Verizon is like 5555555555@vtext.com and it'll text the contents to the phone number in question, so I feel like avoiding Twilio should be possible.
ocdtr_web has quit [Remote host closed the connection]
<DanC>
a doorbell is, traditionally, a completely local notification
ocdtr_web has joined #sandstorm
<ocdtr_web>
Sure, but she wanted the notification on her watch/phone, which either requires you write your own app for local notifications while on Wi-Fi, or bounce across some remote server. Personally, I'd rather bounce any server need off my own web server or Sandstorm server.
<ocdtr_web>
In some ideal future world, in my case, I want my home automation system to have access points/consoles around my whole house, so I'm always in audible range of one. But I don't have a lot of time (or money) to invest in my home automation system right now.
<ocdtr_web>
(I also started work on a methodology for my home automation system(s) to pass messages back and forth via a server, with the goal of encrypting the payload so that the web server doesn't know what they're talking about. But I never finished it. Ideally I could have my phone talk directly to that server too.
<ocdtr_web>
It IS nice to know if someone presses your doorbell when you aren't home, after all.
cbaines_ is now known as cbaines
abliss has joined #sandstorm
<abliss>
is there a way to tell, from a grain, what app it came from?
<abliss>
(i mean, as a user, through the webui, without being an admin or even an owner of the grain)
<ocdtr_web>
abliss: If you hover over the app icon in the grain list, it'll tell you the name of the app.
<ocdtr_web>
Even if it's a private app you don't have installed.
<abliss>
thanks, guess i'll go try to dig up some old pointing device that has a hover semantic...
<ocdtr_web>
For instance, on Alpha, if I hover the app icon for docs.sandstorm.io, I can see it's called GitWeb Pages, but I can't see the version or appId or any of that.
<ocdtr_web>
LOL, what are you using?
<ocdtr_web>
You could probably inspect element if you can "right click" by holding down there.
<abliss>
ocdtr_web: do you already have push rights to the docs git repo?
<abliss>
any idea why all the git commits in that repo have the commit message "autocommit on $date" instead of real messages?
<ocdtr_web>
I have that grain, yes. I haven't tried to push to it yet. Mostly because there's PRs I'd want to see merged on GitHub before trying to refresh it.
<ocdtr_web>
My assumption is someone had docs pushes automated at some point in the past?
<ocdtr_web>
Well, actually, even after jenkins was out of the mix there, Kenton's say Autocommit too.
<ocdtr_web>
Interesting!
<abliss>
oh, i get it -- that's inserted by 'generate.sh'
<abliss>
you mentioned on the call that the deployed version is behind even the last merged PR, so why not refresh it?
<abliss>
last autocommit is 2018-03-11, last change to docs/* 2018-12-26
<ocdtr_web>
That's a valid point. Though I am hoping to get the merge ability sooner rather than later. And I will not complain if by then there's a copy of the SPK I can test push to.
ocdtr_web has quit [Remote host closed the connection]