<domenkozar[m]>
it has many drawbacks that you only discover when it's too late
<domenkozar[m]>
I use ASL as it does the right thing
<domenkozar[m]>
most people love MIT/BSD3
<domenkozar[m]>
but lawyers do face-palm when they see it
<domenkozar[m]>
because it says nothing about all the issues
<domenkozar[m]>
so interpretation is left to the country of origin
<domenkozar[m]>
if you want to sell your livers to lawyers that's the way to do it
<domenkozar[m]>
but then all in all, you have to decide on the goals, but it takes some learning to understand which ones are worth pursuing
<domenkozar[m]>
I'm lucky to have hanged around silver_hook for long time :) he's one of the best FOSS lawyers out there in EU and trust me the stories you hear, many lessons learned.
<drakonis>
domenkozar[m]: have you thought about providing the ability to browse cachix?
<drakonis>
a public cache index
<domenkozar[m]>
listing of nars?
<domenkozar[m]>
there will be one for the owner, but having public listing I'm not convinced it's useful
<domenkozar[m]>
although it would be implicilty useful once it has command-not-found support
<domenkozar[m]>
drakonis: what's your use case? :)
<drakonis>
it is useful for code that isn't ready for nixpkgs
<drakonis>
or some kind of repository that cannot exist in nixpkgs
<drakonis>
its useful for specific edge cases
<drakonis>
providing public builds is my use case
drakonis has quit [Quit: WeeChat 2.4]
Synthetica has joined #nixos-dev
<domenkozar[m]>
but do you want to list: caches, derivations or files within?
init_6 has joined #nixos-dev
<domenkozar[m]>
@drak
<domenkozar[m]>
drakonis: ^^
init_6 has quit [Ping timeout: 246 seconds]
sam_d has quit [Ping timeout: 246 seconds]
sam_d has joined #nixos-dev
obadz has quit [Quit: WeeChat 2.4]
obadz has joined #nixos-dev
justanotheruser has quit [Ping timeout: 268 seconds]
<qyliss^work>
MPL is nice -- I think of it as halfway between GPL and ASL
<Taneb>
American Sign language?
<Taneb>
Oh, Apache Software License?
<Profpatsch>
domenkozar[m]: What are other drawbacks of the GPL apart from companies not using your code?
<Profpatsch>
Which might not be a bad thing honestly :)
<domenkozar[m]>
did you read the twitter thread?
<domenkozar[m]>
because I give a bunch of reasons there :)
<domenkozar[m]>
actually he published that just a few hours ago after our discussion with Jake :)
<domenkozar[m]>
s/ago//
<domenkozar[m]>
I think his blog post should be summarized by Tim Minchin quote
<domenkozar[m]>
Science adjusts its views based on what's observed
<domenkozar[m]>
Faith is the denial of observation so that belief can be preserved.
<Profpatsch>
Free Software was not always about the licenses. I mean it still isn’t, it’s about the four freedoms.
<Profpatsch>
But the FSF is a bit stuck in their ways I guess.
<Profpatsch>
And who can blame them, it has worked for them so far
<Profpatsch>
I think a huge part of the struggle is (as everywhere else) about the (well-founded) fear of companies taking the commons and privatizing them
<domenkozar[m]>
you mean the motivation for using GPL?
<Profpatsch>
Though of course, coming from an inverse copyright-discussion angle, proprietized MIT/APL/MPL code does not un-opensource the existing code.
<Profpatsch>
The proprietary parts are just lost to the commons is all
<Profpatsch>
domenkozar[m]: I guess
<domenkozar[m]>
long story to explain how that doesn't wor
<domenkozar[m]>
k
<domenkozar[m]>
not really in the mood to go down that path :)
orivej has joined #nixos-dev
<ryantm>
Thanks for pointing out the issues with nixpkgs-update licensing. I'll work on making it better.
<ryantm>
If I can get all the contributors to dedicate their work to the public domain, then I can license it under any number of other licenses that might make people happy.
<gchristensen>
"...They may even prohibit any attempt by copyright owners to surrender rights automatically..." a problem the book mentioned is that all licenses like this expire after a certain number of years
<tilpner>
But what if the catch is that it's not legally possible to dedicate something to the public domain?
<tilpner>
Then you make it seem like your creation is safe to use, while indeed nobody else has any right to use it
<tilpner>
If you have looked into this in detail, I'll shut up right now :)
<ryantm>
I haven't looked into it too in-depth.
<tilpner>
I can't safely say either way whether public domain is possible or not for you
<tilpner>
But I can say that there is some FUD around it that may cause people to shy away from your creation while they wouldn't have worried with a safer license
<tilpner>
Again, I lack any qualification to say whether it's a thing or not
* gchristensen
doesn't know
<tilpner>
But just the existence of the question whether it is would make me decide for a license
<ryantm>
My concern for a license is that one of the contributors or their heirs might try to use the government to force other people to do something.
<ryantm>
I don't want to be promoting that kind of thing.
<gchristensen>
sounds like you need a laywer :X
<ryantm>
gchristensen: Thanks for the book reference.
<gchristensen>
yep!
Synthetica has quit [Quit: Connection closed for inactivity]
<ryantm>
Oh, cool. I reside in the jurisdiction of the 9th circuit, which is where the case dbj mentions happened.
<Profpatsch>
domenkozar[m]: You mean how the GPL doesn’t help with keeping code in the commons?
<Profpatsch>
I guess what the article you linked says is true, you can’t prevent bad actors from being bad actors.
justanotheruser has joined #nixos-dev
<domenkozar[m]>
Honestly I'm not sure if you're trolling or you made that conclusion to fit your bias :)
<domenkozar[m]>
but let's say that is the conclusion, wasn't GPL whole point to prevent bad actors?
<domenkozar[m]>
maybe that discussion should happen on a different channel : )
<domenkozar[m]>
btw, I was thinking we'd make nix-prefetch* scripts not require write access to store
<domenkozar[m]>
so they can be used during build time
<domenkozar[m]>
I mean, optionally not require.
drakonis has joined #nixos-dev
<ryantm>
My impression is the nix-prefetch scripts are useless if you want them to work generally, because they don't work the same way as nix-build.
<ryantm>
I had to abandon using nix-prefetch-url -A in nixpkgs-update because I kept running into corner cases it doesn't handle.
<ryantm>
Now I just use nix-build and parse the output of it to find the hash it expects.
<domenkozar[m]>
in this case you can't use nix-build withing a build so that's not an option :)
<samueldr>
and been that way since a good while back
<manveru>
:)
<samueldr>
I asked at the time and had no "but omg wtf?" reactions so I guess it's fine
<manveru>
ok, was just surprised
<manveru>
wanted to make a nix-shell script that writes nix-shell scripts, and hit that
<samueldr>
and, AFAIK, it won't be "fixed", as lua would require other hacks
<samueldr>
at least, that's something I figured out when thinking about it
<samueldr>
oh, I guess one could write a quiney thing then using that feature :)
<manveru>
jup :D
<samueldr>
and while I said lua, that's not limited to lua; any language that doesn't use `#` as comments delimiter would need to somehow wrangle those instructions into a multi-line comment
<samueldr>
so imagine a C++ interpreter (no #) `/*\n#! nix-shell -p ...\n*/`
<manveru>
yeah, i understand the reasoning
<manveru>
was just wondering, since i didn't immediately get it :)
<gchristensen>
ryantm: you might want to read up on https://nlnet.nl/PET/faq/ before soliciting contributors to make specific license changes
<gchristensen>
to make sure it is compatible
<thoughtpolice>
Speaking of that, I actually did have a nix-shell-cc prototype somewhere I've forgotten that could be used as a shebang with nix-shell. Needless to say: absolutely horrifying.
<thoughtpolice>
In my defense though, I really didn't want to type out the 5 extra words to run 'gcc' in nix-shell.
armijn has joined #nixos-dev
<armijn>
hello, I was asked by gchristensen to join this channel
<Taneb>
Hi, armijn
<gchristensen>
ryantm: armijn works with nlnet on the grants and what not
<{^_^}>
ryantm/nixpkgs-update#115 (by ryantm, 1 hour ago, open): indicate that contributors also waive their copyrights
<armijn>
I understand what ryantm wants to do but I don't think that this is the right approach
<armijn>
the difficult part is this:
<armijn>
"To the extent possible under law"
<armijn>
it isn't defined what happens if it is *not* possible
<armijn>
and it might also not be possible to waive copyrights in certain jurisdictions
<armijn>
therefore I would very strongly recommend to use a standard very permissive license, assuming that this is the intention
<armijn>
ryantm: what is your intention with this particular license?
<armijn>
ah, there are logs :)
<armijn>
"My concern for a license is that one of the contributors or their heirs might try to use the government to force other people to do something."
<armijn>
yes, it is possible that in certain jurisdictions copyrights will eventually end up in the hands of a government.
<armijn>
but by making an *explicit* license grant (example: MIT) you are taking away that threat: code cannot be retroactively relicensed
<armijn>
the copyright holders would still have the copyright (which cannot easily be waived in certain jurisdictions) but since you would have a license it doesn't matter who owns the copyright in the code
<armijn>
the current language leaves that up in the air
drakonis_ has joined #nixos-dev
drakonis has quit [Ping timeout: 258 seconds]
drakonis has joined #nixos-dev
Synthetica has joined #nixos-dev
drakonis_ has quit [Ping timeout: 248 seconds]
<armijn>
ryantm: can we talk?
andi- has quit [Ping timeout: 250 seconds]
andi- has joined #nixos-dev
andi- has quit [Excess Flood]
<armijn>
ryantm: I have reached out to some lawyers in the US, hoping to have a contact for you to talk to soon
andi- has joined #nixos-dev
<ryantm>
armijn: I direct messaged you on IRC.
drakonis1 has joined #nixos-dev
drakonis1 has quit [Quit: WeeChat 2.4]
andi- has quit [Ping timeout: 246 seconds]
drakonis_ has joined #nixos-dev
drakonis has quit [Ping timeout: 252 seconds]
andi- has joined #nixos-dev
drakonis_ has quit [Read error: Connection reset by peer]
drakonis_ has joined #nixos-dev
drakonis has joined #nixos-dev
drakonis_ has quit [Ping timeout: 276 seconds]
drakonis_ has joined #nixos-dev
drakonis_ has quit [Ping timeout: 258 seconds]
drakonis_ has joined #nixos-dev
drakonis_ has quit [Ping timeout: 252 seconds]
<gchristensen>
When running a `--check` build with `--keep-failed`, Nix puts a differing output dir at `/nix/store/...-foo.check`. this path is not registered, and so is not safe from GC
<gchristensen>
I think it would be interesting to have this path be GC rooted so that (a) I don't have to do string concatenation to find it, (b) it doesn't get GC rooted. also, i think it wouldbe interesting in these cases to provide an output which contains the build directory ("/build/...") so I can get a good reference to that, too.
<gchristensen>
any ideas on how to do this? the part I'm unsure of is if there is a way for at the end of the build to instantiate "bogus" drvs for these to be capttured
<gchristensen>
one method I considered is using the diff hook mechanism (extending this to support passing in the temp dir), although this is still racy
orivej has quit [Ping timeout: 244 seconds]
drakonis has quit [Ping timeout: 268 seconds]
drakonis_ has joined #nixos-dev
<sphalerite>
gchristensen: I assume that (b) is supposed to be "it doesn't get GCd"?
<gchristensen>
yep
<LnL>
non linux builds can't use a .check bindmount so they use hash rewriting, in which case the check output is a completely different path
<gchristensen>
interesting
<LnL>
I think that path could be registered as a valid output along with an indirect result-checked root