We can always compute the store path from the content hash, but not
vice versa. Storing the content hash allows `hashPath()` to be
replaced by `fetchToStore(...FetchMode::DryRun...)`, which gets us
caching in lazy-trees mode.
86785fd9d1 was broken because it was
storing the full path in the MountedSourceAccessor as the `path` field
in the fetcher cache key (i.e. including the
/nix/store/... prefix). Especially in the case of lazy (virtual) store
paths, this didn't work at all because those paths are different every time.
This was broken because MountedSourceAccessor did not return a
fingerprint. Previously fingerprints were global to an accessor, but
with a MountedSourceAccessor the fingerprint can be different for each
mount point.
These got displayed as e.g.
«github:NixOS/nixpkgs/adaa24fbf46737f3f1b5497bf64bae750f82942e?narHash=sha256-qhFMmDkeJX9KJwr5H32f1r7Prs7XbQWtO0h3V0a0rFY%3D»/nix/store/x9wnkly3k1gkq580m90jjn32q9f05q2v-source/pkgs/stdenv/generic/source-stdenv.sh
Now we get
«github:NixOS/nixpkgs/adaa24fbf46737f3f1b5497bf64bae750f82942e?narHash=sha256-qhFMmDkeJX9KJwr5H32f1r7Prs7XbQWtO0h3V0a0rFY%3D»/pkgs/stdenv/generic/source-stdenv.sh
For example, instead of doing
#include "nix/store-config.hh"
#include "nix/derived-path.hh"
Now do
#include "nix/store/config.hh"
#include "nix/store/derived-path.hh"
This was originally planned in the issue, and also recent requested by
Eelco.
Most of the change is purely mechanical. There is just one small
additional issue. See how, in the example above, we took this
opportunity to also turn `<comp>-config.hh` into `<comp>/config.hh`.
Well, there was already a `nix/util/config.{cc,hh}`. Even though there
is not a public configuration header for libutil (which also would be
called `nix/util/config.{cc,hh}`) that's still confusing, To avoid any
such confusion, we renamed that to `nix/util/configuration.{cc,hh}`.
Finally, note that the libflake headers already did this, so we didn't
need to do anything to them. We wouldn't want to mistakenly get
`nix/flake/flake/flake.hh`!
Progress on #7876
(cherry picked from commit cc24766fa6)
For example, instead of doing
#include "nix/store-config.hh"
#include "nix/derived-path.hh"
Now do
#include "nix/store/config.hh"
#include "nix/store/derived-path.hh"
This was originally planned in the issue, and also recent requested by
Eelco.
Most of the change is purely mechanical. There is just one small
additional issue. See how, in the example above, we took this
opportunity to also turn `<comp>-config.hh` into `<comp>/config.hh`.
Well, there was already a `nix/util/config.{cc,hh}`. Even though there
is not a public configuration header for libutil (which also would be
called `nix/util/config.{cc,hh}`) that's still confusing, To avoid any
such confusion, we renamed that to `nix/util/configuration.{cc,hh}`.
Finally, note that the libflake headers already did this, so we didn't
need to do anything to them. We wouldn't want to mistakenly get
`nix/flake/flake/flake.hh`!
Progress on #7876
This makes paths in error messages behave similar to lazy-trees,
e.g. instead of store paths like
error: attribute 'foobar' missing
at /nix/store/ddzfiipzqlrh3gnprmqbadnsnrxsmc9i-source/machine/configuration.nix:209:7:
208|
209| pkgs.foobar
| ^
210| ];
you now get
error: attribute 'foobar' missing
at /home/eelco/Misc/eelco-configurations/machine/configuration.nix:209:7:
208|
209| pkgs.foobar
| ^
210| ];
The short answer for why we need to do this is so we can consistently do
`#include "nix/..."`. Without this change, there are ways to still make
that work, but they are hacky, and they have downsides such as making it
harder to make sure headers from the wrong Nix library (e..g.
`libnixexpr` headers in `libnixutil`) aren't being used.
The C API alraedy used `nix_api_*`, so its headers are *not* put in
subdirectories accordingly.
Progress on #7876
We resisted doing this for a while because it would be annoying to not
have the header source file pairs close by / easy to change file
path/name from one to the other. But I am ameliorating that with
symlinks in the next commit.
(cherry picked from commit f3e1c47f47)
The short answer for why we need to do this is so we can consistently do
`#include "nix/..."`. Without this change, there are ways to still make
that work, but they are hacky, and they have downsides such as making it
harder to make sure headers from the wrong Nix library (e..g.
`libnixexpr` headers in `libnixutil`) aren't being used.
The C API alraedy used `nix_api_*`, so its headers are *not* put in
subdirectories accordingly.
Progress on #7876
We resisted doing this for a while because it would be annoying to not
have the header source file pairs close by / easy to change file
path/name from one to the other. But I am ameliorating that with
symlinks in the next commit.
This allows writing lock files with dirty inputs, so long as they have
a NAR hash. (Currently they always have a NAR hash, but with lazy
trees that may not always be the case.)
Generally dirty locks are bad for reproducibility (we can detect if
the dirty input has changed, but we have no way to fetch it except
substitution). Hence we don't allow them by default.
Fixes#11181.
We now just check that the fetcher doesn't change any attributes in
the input, and return all the original attributes (i.e. discarding any
new attributes and keeping any attributes that the fetcher didn't
keep).
This fixes the error
'{"__final":true,"lastModified":1686592866,"narHash":"sha256-riGg89eWhXJcPNrQGcSwTEEm7CGxWC06oSX44hajeMw","owner":"nixos","repo":"nixpkgs","rev":"0eeebd64de89e4163f4d3cf34ffe925a5cf67a05","type":"github"}' resulted in different input
'{"__final":true,"lastModified":1686592866,"narHash":"sha256-riGg89eWhXJcPNrQGcSwTEEm7CGxWC06oSX44hajeMw=","owner":"nixos","repo":"nixpkgs","rev":"0eeebd64de89e4163f4d3cf34ffe925a5cf67a05","type":"github"}'
in flake-regressions/tests/nix-community/patsh/0.2.1 (note the lack of
a trailing '=' in the NAR hash in the lock file).
We haven't added the narHash attribute yet at that point. And if the
caller uses getAccesor() instead of fetchToStore() (e.g. in `nix
registry pin`), the narHash attribute will never be added. This could
lead to a mismatch.
The ability to substitute inputs was removed in #10612 because it was
broken: with user-specified inputs containing a `narHash` attribute,
substitution resulted in an input that lacked the attributes returned
by the real fetcher (such as `lastModified`).
To fix this, we introduce a new input attribute `final`. If `final =
true`, fetching the input cannot add or change any attributes.
We only attempt to substitute inputs that have `final = true`. This is
implied by lock file entries; we only write a lock file if all its
entries are "final".
The user can specified `final = true` in `fetchTree`, in which case it
is their responsibility to ensure that all attributes returned by the
fetcher are included in the `fetchTree` call. For example,
nix eval --impure --expr 'builtins.fetchTree { type = "github"; owner = "NixOS"; repo = "patchelf"; final = true; narHash = "sha256-FSoxTcRZMGHNJh8dNtKOkcUtjhmhU6yQXcZZfUPLhQM="; }'
succeeds in a store path with the specified NAR hash exists or is
substitutable, but fails with
error: fetching final input '{"final":true,"narHash":"sha256-FSoxTcRZMGHNJh8dNtKOkcUtjhmhU6yQXcZZfUPLhQM=","owner":"NixOS","repo":"patchelf","type":"github"}' resulted in different input '{"final":true,"lastModified":1718457448,"narHash":"sha256-FSoxTcRZMGHNJh8dNtKOkcUtjhmhU6yQXcZZfUPLhQM=","owner":"NixOS","repo":"patchelf","rev":"a0f54334df36770b335c051e540ba40afcbf8378","type":"github"}'
Known behavior changes:
- `MemorySourceAccessor`'s comparison operators no longer forget to
compare the `SourceAccessor` base class.
Progress on #10832
What remains for that issue is hopefully much easier!
Progress on #5638
There are still a global fetcher and eval settings, but they are pushed
down into `libnixcmd`, which is a lot less bad a place for this sort of
thing.
Continuing process pioneered in
52bfccf8d8.
Subflakes are flakes in the same tree, accessed in flake inputs via
relative paths (e.g. `inputs.foo.url = "path:./subdir"`). Previously
these didn't work very well because they would be separately copied to
the store, which is inefficient and makes references to parent
directories tricky or impossible. Furthermore, they had their own NAR
hash in the lock file, which is superfluous since the parent is
already locked.
Now subflakes are accessed via the accessor of the calling flake. This
avoids the unnecessary copy and makes it possible for subflakes to
depend on flakes in a parent directory (so long as they're in the same
tree).
Lock file nodes for relative flake inputs now have a new `parent` field:
{
"locked": {
"path": "./subdir",
"type": "path"
},
"original": {
"path": "./subdir",
"type": "path"
},
"parent": [
"foo",
"bar"
]
}
which denotes that `./subdir` is to be interpreted relative to the
directory of the `bar` input of the `foo` input of the root flake.
Extracted from the lazy-trees branch.
After the removal of the InputAccessor::fetchToStore() method, the
only remaining functionality in InputAccessor was `fingerprint` and
`getLastModified()`, and there is no reason to keep those in a
separate class.
Having a narHash doesn't mean that we have the other attributes
returned by the fetcher (such as lastModified or rev). For instance,
$ nix flake metadata github:NixOS/patchelf/7c2f768bf9601268a4e71c2ebe91e2011918a70f
Last modified: 2024-01-15 10:51:22
but
$ nix flake metadata github:NixOS/patchelf/7c2f768bf9601268a4e71c2ebe91e2011918a70f?narHash=sha256-PPXqKY2hJng4DBVE0I4xshv/vGLUskL7jl53roB8UdU%3D
(does not print a "Last modified")
The latter only happens if the store path already exists or is
substitutable, which made this impure behaviour unpredictable.
Fixes#10601.
