<ocdtrekkie>
Again, I could answer, but he is once again... not here.
<crab>
was the "…" a pause to put on sunglasses? :-)
larjona has joined #sandstorm
<ocdtrekkie>
No. Just a generic dramatic pause.
<ocdtrekkie>
No puns.
harish has joined #sandstorm
harish has quit [Ping timeout: 248 seconds]
harish has joined #sandstorm
tobald has joined #sandstorm
n8a has quit [Ping timeout: 246 seconds]
harish has quit [Ping timeout: 248 seconds]
<hannes[m]>
in the etherpad grain, how can i enable the fancy inline character-by-character author coloring?
<hannes[m]>
just by line is not sufficient...
harish has joined #sandstorm
xet7 has joined #sandstorm
n8a has joined #sandstorm
jemc has joined #sandstorm
jemc has quit [Ping timeout: 240 seconds]
tobald has quit [Quit: Ex-Chat]
ill_logic has quit [Remote host closed the connection]
n8a has quit [Ping timeout: 276 seconds]
n8a has joined #sandstorm
xet7 has quit [Ping timeout: 260 seconds]
ill_logic has joined #sandstorm
eloyesp has joined #sandstorm
gwillen has joined #sandstorm
<eloyesp>
Hi, I'm starting with sandstorm (I've installed it three weeks ago and were working on some developments specifically for it)
<eloyesp>
I was also reading the source to check if I was able to contribute somehow, but I found that sandstorm itself is licensed with "apache license"
<xj9[m]>
what's wrong with Apache?
<sknebel>
...and?
<eloyesp>
and it sounded a little contradictory to me, as sandstorm is supposed to promote "software libre"
<eloyesp>
why it doesn't use GPL or AGPL?
<maurer>
Practically, it probably has to do with the fact that the company that was writing the open source project was also writing non-open source extensions
<maurer>
(Specifically, they were producing a multi-chassis hosting setup for sandstorm grains)
<eloyesp>
what is the the point of allowing privative forks of sandstorm?
<xj9[m]>
I mean, GPL is better but Apache is definitely free software license
<maurer>
Eh, "better" depends on the goals of your free software
<maurer>
but if you're wondering what the original reason is, the original reason is probably "because we were building a paid product that incorporated this code"
<maurer>
With a side of "If we make it GPL, it will scare off developers/businesses"
<xj9[m]>
kek
<maurer>
I mean, that is a real concern
<maurer>
I do research at a university, code I write for a grant is copyright the university
<maurer>
they will let me release under some free software licenses, but GPL is forbidden
<maurer>
as a result, even though I'm writing code with the intention of freely giving it away, I can't use any libraries that are GPL licensed
<maurer>
I'm almost certainly not the only one
* TimMc
rolls eyes
<eloyesp>
that is scary
<maurer>
In any case, tl;dr, there are reasons to use licenses other than the GPL, usually pertaining to real world concerns e.g. "How will the developers get paid" and "We want group X to be able to use our software"
<eloyesp>
maurer: so the university allows you to work on apache licensed projects, but prohibits you to contribute on GPLed ones?
<maurer>
eloyesp: I can contribute on my own time, but research projects I do for the university (and so have the university copyright) can only be BSD-3/MIT, or a couple others
<maurer>
eloyesp: No copyleft, no patent grants
<maurer>
However, linking against apache2 doesn't force my code to be apache2
<maurer>
Thus the ability to still use apache2 libraries
<ocdtr_web>
I don't know if Sandstorm's goal has ever been explicitly to "promote software libre" in the FSF-type definition. Particularly since it also supports closed source apps (it just doesn't have any right now).
<ocdtr_web>
Making small/niche web apps viable is definitely a key goal, and many of them are open source.
<ocdtr_web>
Arguably, Sandstorm may be one of the best places to run closed source apps, in fact, when you consider that they run in a sandbox that can make sure that they behave themselves.
<ocdtr_web>
And they are gone, but for the chat log, I feel like "what is the the point of allowing privative forks of sandstorm?" must be answered with:
<ocdtr_web>
How can you call it "software libre" when you're trying to disallow people from doing things? o.O
ocdtr_web has quit [Quit: Page closed]
<xj9[m]>
there is a distinction between user and developer freedom
<xj9[m]>
free software and the FSF are explicitly about promoting freedom for users
frigginglorious has quit [Ping timeout: 252 seconds]
<xj9[m]>
by allowing proprietary forks, you are allowing certain parties to deprive users of their freedoms
<xj9[m]>
a matter of perspective and priorities
gambatte has joined #sandstorm
<isd>
fwiw, for something like sandstorm (core) I would probably have gone AGPLv3 if I'd started the project.
<simpson>
TBH that would have closed off something I'm considering, which is to have my LLC host an Oasis at unreasonably-thin margins as a community service.
<isd>
I can't speak to the original team's reasoning. I suspect blackrock was relevant to it (then-proprietary scalable backend), but that's moot now that blackrock has been released as foss.
n8a has quit [Ping timeout: 246 seconds]
<simpson>
But I don't have customers lined up, so it's kind of moot.
<isd>
simpson: How would that be a problem?
<simpson>
isd: AGPL
<isd>
simpson: yeah, right. I don't follow.
<simpson>
It's just not something I want within my business until it's been tested in a USA court.
<isd>
In any case, I showed up to a pre-existing project that was doing good work, and decided it wasn't worth sweating the license.
<isd>
simpson: why? Assume for the sake of argument everything in it is enforceable -- what's the problem with just having a link to a github fork?
<isd>
If you're just hosting a thing as a community service, I don't see how it's at all relevant
<isd>
Unless you actually planned on making proprietary modifications.
<simpson>
isd: It's not clear whether or not it affects unrelated trade-secret code running nearby under the same umbrella of ownership, and until I know for sure, I am not interested in taking on that risk.
n8a has joined #sandstorm
Telesight has quit [Remote host closed the connection]
<isd>
Which bits are unclear? IIRC the patent grants are basically the same thing as in the apache license.
<xj9[m]>
AGPL is GPL 3 + a clause that specified that network access is considered distribution
<isd>
correct.
<mokomull>
IMHO, the unclear bit is defining a "derivative work", which affects any flavor of the GPL
<isd>
If you're just hosting it, you're not going to hit any grey areas there.
<mokomull>
In my experience hosting things with open-source software, you're eventually going to hit _something_ that you need to monkeypatch.
<simpson>
isd: How much of a k8s cluster is affected by hosting, say, MongoDB in a pod?
pie_ has quit [Remote host closed the connection]
pie_ has joined #sandstorm
pie__ has joined #sandstorm
pie__ has quit [Remote host closed the connection]
<xj9[m]>
how is derivative work unclear?
<xj9[m]>
if you include GPL code in your program, the whole program has to be GPL
<maurer>
Yeah, but if it's AGPL
<maurer>
Then are communicating networked services linked?
<maurer>
etc
<xj9[m]>
AGPL only adds that network access is considered distribution
<simpson>
The Mongo bone's connected to the mongo_exporter bone, the mongo_exporter bone's connected to the Prometheus bone, the Prometheus bone's connected to the Kubernetes bone, the Kubernetes bone's connected to the rest of the cluster, etc.
<isd>
xj9[m]: in fairness, with out that none of this comes up at all when you're just hosting.
<simpson>
Bingo. Apache 2 is fine. I like Apache 2. I don't care about the patent clauses or anti-tivoization clauses. Heck, I like them.
<isd>
simpson: clarifying question: would plain old GPLv3 cause any trouble, or is it *just* the hosting-is-distribution issue?
<mokomull>
xj9[m]: GNU believes that linking against a .so even if you've never so much as seen the source code creates a "derivative work" encompassing the whole result. Others don't
<xj9[m]>
LGPL explicitly grants that ability
<simpson>
isd: GPLv3 is fine, e.g. I have gcc in my NixOS closures and it doesn't bother me.
<mokomull>
Sure, as do a bunch of other projects (e.g. Linux clarifies that using int $0x80 doesn't create a derivative work)
<maurer>
xj9[m]: The notion of what is a derivative work is sufficiently fuzzy to me as a programmer that I won't use any GPL, let alone AGPL, unless my entire project is intended to be GPL and is fairly compartmentalized from anything that needs to not be
<maurer>
xj9[m]: AGPL is awkward because it explicitly brings the network into scope
<maurer>
xj9[m]: I'm not a lawyer, but when the license starts talking about network behavior, it would make me nervous if I had something I didn't want to distribute
<mokomull>
I can't so much think of an AGPL'ed piece of software that's actually used in the real world. Perhaps MongoDB, but that has a license exception IIRC.
<mokomull>
Not our little open-source Freenode corner of the world.
<isd>
mokomull: Clarify then. 500k users seems like a pretty reasonable justification.
<isd>
What is "the real world"
<xj9[m]>
most of the users are japanese actually
<xj9[m]>
not free software people
<mokomull>
ah, interesting, that I didn't know
<xj9[m]>
real world enough bitch?
<isd>
xj9[m]: keep it civil.
<isd>
I actually tend to agree that the notion of derivative work is pretty murky.
<mokomull>
For context, I'm an infrastructure nerd. Hence why I hang out in #sandstorm. One wouldn't generally build a company atop an AGPL'ed infrastructure that's going to cause their /whole/ application to be in-scope.
<isd>
In any case, like I said, I came in to an existing community and didn't think it was worth making a fuss over.
<isd>
but, to be blunt, I personally basically have no sympathy for the use case you're discussing.
<xj9[m]>
i would chose AGPL specifically to prevent you from using my IP + your secret sauce to make money
<xj9[m]>
you can make money, i have no problem with that
<xj9[m]>
but if you are going to make money with my tech, i'm definitely going to want either: money OR any improvements/changes you make to the IP i've created
<mokomull>
That's your right as an author. That severely limits the number of engineers that are legally able to contribute to your software, though.
<xj9[m]>
so what? is it actually better that you can take my tech, make money with it, and i get nothing?
<isd>
Meh, in practice not many people are hacking on core anyway.
<mokomull>
That's where the fuzziness of "derivative work" comes in. Does my webapp that, say, links against your templating library become a whole derivative work?
<mokomull>
You want my contributions to the example templating library. I don't want to use it if it means I have to give you the rest of it.
<isd>
Honestly the number of people I've heard from who have been turned off from the apache license is pretty substantial. I wonder if it really makes sense from a community standpoint to worry that much about contributors who feel the opposite way; copyleft fans seem like they'd be more excited about something like sandstorm in general anyway.
<xj9[m]>
generally speaking, you'd license a library LGPL which explicitly grants you the ability to link to it without invoking the viral distribution clause
<isd>
And, frankly, as far as who's hacking on core now: it's basically kentonv and an occasional drive-by patch.
<isd>
It's not like the apache license is attracting a ton of folks.
<xj9[m]>
but, if it was vanilla GPL, which doesn't have that explicit grant, you'd need to license the whole thing GPL
<mokomull>
xj9[m]: Except a _lot_ of libraries are _not_ LGPL'ed -- some explicitly to artificially limit their user base, some just out of pure "didn't bother thinking about it"
<isd>
The FSF also specifically doesn't recommend defaulting to LGPL for libraries -- hence the name change. They keep pointing to readline as an example of it pulling more code into the pool of free software.
<mokomull>
I think that's what annoys me the most, artificially alienating a project's potential users in the hopes that a magical Libre world will appear.
<isd>
mokomull's facts are basically dead on. I just... don't really care.
<isd>
fwiw, I don't think "strongest copyleft possible" is necessarily the right option in every case. I've got a bunch of apache licensed stuff on my github.
<mokomull>
Yeah, as you said ... if what has Actually Happened™ didn't bring a huge contributor base, then it's probably not worth thinking about the theoreticals.
<mokomull>
What's actually happened is more concrete data than I have :)
<isd>
musl actually relicensed lgpl -> mit because static linking was such a standard use case for the library that it was a pain to work with; the folks arguing for it by and large weren't doing proprietary anything, they just didn't want to deal with the hassle.
<xj9[m]>
the only goal for free software is to promote/protect software freedom
<isd>
It really depends on what makes sense for the community.
<isd>
If it *is* going to drive everybody away, then maybe having the software not go anywhere because of copyleft license defeats the point.
<isd>
but yeah, that's not what the data suggests in this cas.e
<xj9[m]>
mokomull is clearly a corporate shill who really wants to leech value from the free/libre community
<xj9[m]>
that is their prerogative of course
<mokomull>
xj9[m]: Actually working in an environment where we contribute everything back to an Apache project :)
<isd>
xj9[m]: please, keep it civil. knock it off with the insults.
<isd>
Really not ok.
<mokomull>
like, I don't even have a local repo. Everything's on the Apache JIRA waiting for code review :P
<mokomull>
but yes, I work for a company than makes money and pays me
<mokomull>
that*
<TimMc>
mokomull: Whoa there, that's just a little too wild for me. :-P
<isd>
FWIW, most of my income comes from hacking on a FOSS project, which is apache licensed.
<isd>
(some of it is internal tools from other clients, and sysadmin works)
<TimMc>
(working for money so you can buy food? that's not the FOSS spirit!)
<isd>
(but I've never written a line of proprietary product code).
<xj9[m]>
that's actually not what it means to be a shill
<isd>
The characterization that anyone advocating the apache license is a corporate shill just trying to make the code available so they can hijack it and turn it into a proprietary product really isn't fair.
<isd>
It's a lot more complex than that.
<xj9[m]>
i've written plenty of proprietary code in my career, but i don't run around promoting ways to make it easier for corporations to make money off of other people's work
<isd>
I do think there is a certain amount of FUD going around re: the real risks of using the GPL if you're a company that otherwise does proprietary stuff. But it's not that simple either.
<isd>
I've seen quite a lot of corporate contributions to permissively licensed projects. We can go around in circles talking about theory, but I can't just ignore that fact. There is something deeper going on.
<xj9[m]>
of course GPL is toxic in a proprietary environment. that's the point
<xj9[m]>
that's because permissive licenses work in their favor
<xj9[m]>
however, the free bits basically never make it to the user
<isd>
These projects tend to be infrastructure/developer oriented things, so that's not terribly surprising.
<isd>
FWIW, re: things not making it to the user: most of the action with proprietary products these days seems to be SaaS; I think the lack of foss alternatives has less to do with licensing and more to do with feasibility of users actually running stuff themselves.
<isd>
hence, sandstorm.
<isd>
A side point, mongo db is agpl, so if there's real worry about supporting infrastructure being tainted by hosting a thing, that probably affects sandstorm anyway. The usual interpretation is that it doesn't though, and it seems like that would easily extend to operating systems, standard deployment scripts, etc. But I get not wanting to mess with it before it's been tested in court.
<isd>
I'm going to step out for a bit. Will probably be back on in 20-30 min.
isd has quit [Quit: Leaving.]
xet7 has quit [Quit: Leaving]
pie__ has joined #sandstorm
pie_ has quit [Remote host closed the connection]
isd has joined #sandstorm
pie_ has joined #sandstorm
pie__ has quit [Read error: Connection reset by peer]
<ocdtrekkie>
xj9[m]: The free bits always can make it to use the user: People can always download and use the original, open source code.
<ocdtrekkie>
Copyleft is toxic because it tells people what they must do with their code, which is, to me, generally unacceptable even without any profit motive.
<ocdtrekkie>
While I open source my code (MIT license, generally), and don't mind offering up contributions to other people, that's my choice. I took my effort and my code and decided to offer it.
<ocdtrekkie>
It is morally and ethically wrong to dictate terms over other people's work. Note that I do not see a difference between user and developer freedom.
<ocdtrekkie>
Because I am a user of most of the projects I look at, and sometimes also a developer.
<ocdtrekkie>
In an ideal collaborative world, that would quite often be the case.
<isd>
ocdtrekkie: the gpl doesn't preclude you from using your own code in other ways -- you can't combine it with the gpl work without licensing it under that license. You can still use the bit that you wrote elsewhere, under different (even incompatible) licenses.
<isd>
It doesn't restrict you at all, except in the context where you're combining it with gpl code.
<ocdtrekkie>
I get that, but if I create a modification to another piece of software, that modification, as far as I'm concerned, is my code. I'm not opposed to sharing it (usually), but I don't think anyone claiming they own my work inherently has merit.
<ocdtrekkie>
I've yet to have a particular practical situation where I decided away from something due to license, but I view copyleft as a generally toxic behavior: Trying to force others to adopt your values. It's like religion, it's evangelical. (Which is why arguments about it get heated.)
pie_ has quit [Read error: Connection reset by peer]
<isd>
The *only* context in which the gpl restricts what you can do with the code is if you're looking to redistribute gpl'd code (possibly in object form) under a *more* restrictive license.
pie_ has joined #sandstorm
<isd>
(Obviously you can't downgrade to MIT or such)
<isd>
(but, the only thing MIT let's you do that GPL doesn't is add restrictions)
<isd>
Like, if the GPL actually imposed anything on you except not turning the GPL code into a proprietary product, I might agree with you.
<isd>
Again, if your additions are at all separable, you can still distribute the separable bits independently as part of another thing, however you want.
<ocdtrekkie>
I see nothing inherently evil about someone deciding not to open source their work. I prefer if they do, but don't think it's fair to require it.
<isd>
Again, it's only required if you want to base it on the GPL'd work. I think you can argue that it's reasonable to not open source your work. I *don't* think it's reasonable to complain about someone asking that *if you base your work on theirs,* you pay it forward.
<isd>
If it's okay for people to not open source their work, asking folks to pay it forward is *strictly* more generous.
<ocdtrekkie>
It's not "asking", it's a license requirement.
<ocdtrekkie>
It's not really "free" if it comes with significant restrictive terms.
<isd>
So are the things in an EULA. I don't see how it as at all reasonable to argue that EULAs are OK but the GPL isn't.
<ocdtrekkie>
I think EULAs are awful, FWIW.
<isd>
To clarify: the thing I think is inconsistent is arguing that you shouldn't use copyleft, because it lets other people do something even more restrictive. I don't think you can reasonably say proprietary code is okay but copyleft isn't. There's certainly room for disagreement, but I don't see how that's consistent at all.
<isd>
Whether it's 'really "free"' is just inviting an argument about semantics, which I'm not particuarly interested in.
<ocdtrekkie>
Proprietary code doesn't impose it's philosophy upon me as a requirement. Proprietary code has never required I offer my own code in it's license.
<isd>
It typically just says you can't use it in derivitive works *at all*
<isd>
It's imposing *strictly more* than the GPL is.
<isd>
Remember, the GPL only imposes these restrictions *if you make a derivitive work*
<isd>
which, with proprietary code, you generally are just not allowed to do, period.
<isd>
(Or if you redistribute, but the same argument applies there as well).
<ocdtrekkie>
Fair, though if I do not have the code, I couldn't do anything with it anyways. In effect, the license does not control me.
<isd>
I don't see how that's any better. You are still "controlled" to at least the extend you would be if you had the code.
<isd>
Whether that's because someone has taken technical measures or employed a lawyer strikes me as irrelevant.
TC01 has quit [Remote host closed the connection]
<ocdtrekkie>
I would argue employing lawyers is always worse. ;)