<niksnut>
authors should just be required to submit a nix closure of their software :-)
<gchristensen>
yes!
<pie_>
^
matthewbauer has joined #nixos-dev
<matthewbauer>
does anyone know how to override cabal2nix to a newer version? i am running into issues with unsupported platforms that has been fixed in cabal2nix master but not in the one in nixpkgs.
<Profpatsch>
haddock crashes way too often on current master.
<Profpatsch>
segfaults, that is.
<Profpatsch>
matthewbauer: Do you have a local nixpkgs checkout?
<Profpatsch>
matthewbauer: It depends on which cabal2nix you are calling.
<Profpatsch>
And I know what issue you mean. :p
<Profpatsch>
But if it’s the one from all-packages, the following should maybe do the trick: let cabal2nix = pkgs.haskell.lib.appendPatch pkgs.cabal2nix (pkgs.fetchpatch …) in …
jtojnar has joined #nixos-dev
matthewbauer has quit [Ping timeout: 248 seconds]
ben has joined #nixos-dev
matthewbauer has joined #nixos-dev
joepie91 has quit [Excess Flood]
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-dev
__Sander__ has quit [*.net *.split]
teto1 has quit [*.net *.split]
thefloweringash[ has quit [*.net *.split]
adisbladis[m] has quit [*.net *.split]
dtz has quit [*.net *.split]
bgamari has quit [*.net *.split]
Profpatsch has quit [*.net *.split]
sphalerite has quit [*.net *.split]
__Sander__ has joined #nixos-dev
teto1 has joined #nixos-dev
adisbladis[m] has joined #nixos-dev
sphalerite has joined #nixos-dev
bgamari has joined #nixos-dev
Profpatsch has joined #nixos-dev
dtz has joined #nixos-dev
thefloweringash[ has joined #nixos-dev
crystalgamma[m] has quit [Ping timeout: 240 seconds]
Drakonis[m] has quit [Ping timeout: 245 seconds]
rycee has quit [Ping timeout: 245 seconds]
copumpkin has quit [Ping timeout: 245 seconds]
primeos[m] has quit [Ping timeout: 245 seconds]
pstn has quit [Ping timeout: 245 seconds]
codyopel has quit [Ping timeout: 245 seconds]
hl has quit [Ping timeout: 245 seconds]
nocent has quit [Ping timeout: 240 seconds]
thefloweringash[ has quit [Ping timeout: 240 seconds]
regnat[m] has quit [Ping timeout: 240 seconds]
olejorgenb[m] has quit [Ping timeout: 240 seconds]
hedning[m] has quit [Ping timeout: 240 seconds]
adisbladis[m] has quit [Ping timeout: 240 seconds]
dtz has quit [Ping timeout: 240 seconds]
peterhoeg has quit [Ping timeout: 255 seconds]
florianjacob has quit [Ping timeout: 247 seconds]
timokau[m] has quit [Ping timeout: 256 seconds]
sphalerit has quit [Ping timeout: 260 seconds]
__Sander__ has quit [Quit: Konversation terminated!]
<Profpatsch>
Mic92: Oh, you’re announcing it soon?
<Mic92>
Profpatsch: yes. It has all features it needs to make it usuable.
<Profpatsch>
Mic92: Nice. I read the README and I like the minimalism very much.
<Profpatsch>
Perfect design I’d say.
<Profpatsch>
How often do you update the NUR?
<Profpatsch>
I’m guessing one does it centrally every n days?
<infinisil>
Mic92: I'd favor it to not use builtins.fetchGit without a hash, I don't want it to always download new stuff
<Profpatsch>
Oh, it does?
<infinisil>
It's in the readme
<Profpatsch>
Shouldn’t NUR just be a simple cache of hashes to repos?
<Mic92>
infinisil: every 15 min or so. You can provide a hash
<Mic92>
but only for the top-level repo
<Mic92>
the underlying repos have hashes
<Profpatsch>
So it has to use IFD, of course.
<Mic92>
you can also use NUR from a git checkout
<Profpatsch>
(import from derivation)
<samueldr>
👍 namespacing per-repo for the attribute names
<infinisil>
Hmm yeah, nice idea though, I like it :)
<Profpatsch>
Now I need to add my packages from our vuizvui package set. :)
<Profpatsch>
Good thing nix is lazily evaluated …
<infinisil>
Mic92: Is it possible to use a subdirectory of a repo?
<Mic92>
samueldr: would you build a search page, when I create the needed the data?
<Profpatsch>
Mic92: So what’s the story for discovering packages
florianjacob has joined #nixos-dev
<Mic92>
Profpatsch: running an indexer over all repositories.
<Profpatsch>
I mean I need to IFD every subrepo to get a list of all repositories. Based on the checks NUR does on push, they are all required to evaluate.
<Mic92>
but not yet implemented
<samueldr>
Mic92: probably can, hit me up this week-end if you have questions for the format
<Profpatsch>
Mic92: I’m assuming all subrepos need to specify a full hash?
<Profpatsch>
Because fetgit without hash breaks your evaluation guarantees.
<Mic92>
Profpatsch: references/hashes are updated automatically by a cronjob.
<Profpatsch>
(I still don’t know why nix incorporated fixed inputs without hash, that’s just silly)
<infinisil>
Although, what's the point in an overlay for that?
<Profpatsch>
+2
ben has joined #nixos-dev
<infinisil>
Nothing needs to get overridden
<infinisil>
It's just self that you need
<gchristensen>
it uses a more understandable mechanism, is mostly it, imo
<gchristensen>
and overrides are less nice compared to overlays
<Mic92>
many ways lead to Rome
<gchristensen>
like semantically I think it is more ... what it is ... than an override
* infinisil
agrees with overlays
<infinisil>
Related: If somebody wants to supply some overlay, they could do nur.repos.infinisil.myfancyoverlay = self: super: { fancy = changes; }
<gchristensen>
anyway, mostly a semantic question and I like the project (though it spooks me a little :)
<Profpatsch>
gchristensen: Well, you develop your packages privately in NUR, then, once they are ready to move, you put them in nixpkgs and reference them in NUR.
<infinisil>
When should something go in nixpkgs and not NUR?
<Profpatsch>
Some packages are only personal anyway, or are tools that are not big or general enough to go to nixpkgs.
<Mic92>
infinisil: if it benefits from the binary cache.
<Profpatsch>
I mean everybody has some helper tools.
<infinisil>
Could you put that in a section in the readme as well? What belongs and doesn't belong here/in nixpkgs
<samueldr>
one could also realistically build a backports collection I guess
<Mic92>
speaking of backports, maybe I should put an upper limit on evaluation time.
<Mic92>
I don't think we should nixpkgs forks there.
<Mic92>
they are too big.
<Mic92>
*should allow
<samueldr>
Mic92: that would be *hard* to do using time, imagine e.g. travis uses different CPU kinds or some build servers are heavily CPU-starved
<samueldr>
this may make the limit quite fuzzy, sometimes passing other times not
pstn has joined #nixos-dev
crystalgamma[m] has joined #nixos-dev
sphalerit has joined #nixos-dev
adisbladis[m] has joined #nixos-dev
timokau[m] has joined #nixos-dev
hedning[m] has joined #nixos-dev
nocent has joined #nixos-dev
regnat[m] has joined #nixos-dev
primeos[m] has joined #nixos-dev
Drakonis[m] has joined #nixos-dev
hl has joined #nixos-dev
olejorgenb[m] has joined #nixos-dev
dtz has joined #nixos-dev
codyopel has joined #nixos-dev
copumpkin has joined #nixos-dev
thefloweringash[ has joined #nixos-dev
rycee has joined #nixos-dev
peterhoeg has joined #nixos-dev
MichaelRaskin has joined #nixos-dev
matthewbauer has quit [Ping timeout: 260 seconds]
<infinisil>
Mic92: Is NUR supposed to be for packages exclusively?
<Mic92>
infinisil: at moment yes. I have not thought about a concept how a good module structure could look like.
<infinisil>
Other things would be: useful nix functions, modules
<Mic92>
*at the moment
<Mic92>
I first want to implement a package search.
<infinisil>
What do you mean by module structure?
<Mic92>
How modules would be exposed and discovered by the users.
<Mic92>
and used
<infinisil>
Just by telling people about it
<infinisil>
Having it restricted to packages only feels a bit arbitrary just because they are the only easily listable things
<Mic92>
infinisil: ideally the would be seperated from packages by a convention/scheme.
<infinisil>
How about just doing it how nixpkgs does it
<infinisil>
Packages are all top-level attributes for which isDerivation is true, recurse with recurseIntoAttrs
<infinisil>
Would also make it easy for NUR packages to be discovered with nix-env automatically
<Mic92>
I don't understand, how this is different from what it currently does.
elvishjerricco has quit [Ping timeout: 256 seconds]
<infinisil>
I'm just saying that I don't think it makes sense to enforce a "only packages" rule. Just have it work like nixpkgs, which allows arbitrary things, and use the nix-env way to discover packages, no convention/scheme needed
elvishjerricco has joined #nixos-dev
<infinisil>
(Maybe we're talking past each other here?)
<crystalgamma[m]>
I have a nix expression that outputs <LAMBDA> when displayed over builtins.trace but causes "attempt to call something which is not a function but a set" when called … any idea what could be happening here?
<Mic92>
crystalgamma[m]: this happens if you not give a function enough arguments, so returns its result.
<crystalgamma[m]>
(but if I call it with wrong arguments it still outputs the missing mandatory parameter error)
<infinisil>
crystalgamma[m]: It would be good to know what the expression is..
<infinisil>
Also, more of a question for #nixos really
<crystalgamma[m]>
and if I evaluate it from nix-build -E instead of deep in nixpkgs, it works as intended
<infinisil>
crystalgamma[m]: Can you ask in #nixos?
__Sander__ has quit [Quit: Konversation terminated!]