this gives about 20% performance improvements on pure parsing. obviously
it will be less on full eval, but depending on how much parsing is to be
done (e.g. including hackage-packages.nix or not) it's more like 4%-10%.
this has been tested (with thousands of core hours of fuzzing) to ensure
that the ASTs produced by the new parser are exactly the same as the old
one would have produced. error messages will change (sometimes by a lot)
and are not yet perfect, but we would rather leave this as is for later.
test results for running only the parser (excluding the variable binding
code) in a tight loop with inputs and parameters as given are promising:
- 40% faster on lix's package.nix at 10000 iterations
- 1.3% faster on nixpkgs all-packages.nix at 1000 iterations
- equivalent on all of nixpkgs concatenated at 100 iterations
(excluding invalid files, each file surrounded with parens)
more realistic benchmarks are somewhere in between the extremes, parsing
once again getting the largest uplift. other realistic workloads improve
by a few percentage points as well, notably system builds are 4% faster.
Benchmarks summary (from ./bench/summarize.jq bench/bench-*.json)
old/bin/nix --extra-experimental-features 'nix-command flakes' eval -f bench/nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix
mean: 0.408s ± 0.025s
user: 0.355s | system: 0.033s
median: 0.389s
range: 0.388s ... 0.442s
relative: 1
new/bin/nix --extra-experimental-features 'nix-command flakes' eval -f bench/nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix
mean: 0.332s ± 0.024s
user: 0.279s | system: 0.033s
median: 0.314s
range: 0.313s ... 0.361s
relative: 0.814
---
old/bin/nix --extra-experimental-features 'nix-command flakes' eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
mean: 6.133s ± 0.022s
user: 5.395s | system: 0.437s
median: 6.128s
range: 6.099s ... 6.183s
relative: 1
new/bin/nix --extra-experimental-features 'nix-command flakes' eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
mean: 5.925s ± 0.025s
user: 5.176s | system: 0.456s
median: 5.934s
range: 5.861s ... 5.943s
relative: 0.966
---
GC_INITIAL_HEAP_SIZE=10g old/bin/nix eval --extra-experimental-features 'nix-command flakes' --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
mean: 4.503s ± 0.027s
user: 3.731s | system: 0.547s
median: 4.499s
range: 4.478s ... 4.541s
relative: 1
GC_INITIAL_HEAP_SIZE=10g new/bin/nix eval --extra-experimental-features 'nix-command flakes' --raw --impure --expr 'with import <nixpkgs/nixos> {}; system'
mean: 4.285s ± 0.031s
user: 3.504s | system: 0.571s
median: 4.281s
range: 4.221s ... 4.328s
relative: 0.951
---
old/bin/nix --extra-experimental-features 'nix-command flakes' search --no-eval-cache github:nixos/nixpkgs/e1fa12d4f6c6fe19ccb59cac54b5b3f25e160870 hello
mean: 16.475s ± 0.07s
user: 14.088s | system: 1.572s
median: 16.495s
range: 16.351s ... 16.536s
relative: 1
new/bin/nix --extra-experimental-features 'nix-command flakes' search --no-eval-cache github:nixos/nixpkgs/e1fa12d4f6c6fe19ccb59cac54b5b3f25e160870 hello
mean: 15.973s ± 0.013s
user: 13.558s | system: 1.615s
median: 15.973s
range: 15.946s ... 15.99s
relative: 0.97
---
Change-Id: Ie66ec2d045dec964632c6541e25f8f0797319ee2
Here's my guide so far:
$ rg '((?!(recursive).*) Nix
(?!(daemon|store|expression|Rocks!|Packages|language|derivation|archive|account|user|sandbox|flake).*))'
-g '!doc/' --pcre2
All items from this query have been tackled. For the documentation side:
that's for lix-project/lix#162.
Additionally, all remaining references to github.com/NixOS/nix which
were not relevant were also replaced.
Fixes: lix-project/lix#148.
Fixes: lix-project/lix#162.
Change-Id: Ib3451fae5cb8ab8cd9ac9e4e4551284ee6794545
Signed-off-by: Raito Bezarius <raito@lix.systems>
These were mistakenly labeled `eval-fail-*`.
Note that the `lang.sh` runner passes `parse-fail-*` tests on stdin, so
filenames are removed from error messages.
Change-Id: I7f3a0d78b6cfa87af29aaa1b7af19d5a57fd4ade
we now keep not a table of all positions, but a table of all origins and
their sizes. position indices are now direct pointers into the virtual
concatenation of all parsed contents. this slightly reduces memory usage
and time spent in the parser, at the cost of not being able to report
positions if the total input size exceeds 4GiB. this limit is not unique
to nix though, rustc and clang also limit their input to 4GiB (although
at least clang refuses to process inputs that are larger, we will not).
this new 4GiB limit probably will not cause any problems for quite a
while, all of nixpkgs together is less than 100MiB in size and already
needs over 700MiB of memory and multiple seconds just to parse. 4GiB
worth of input will easily take multiple minutes and over 30GiB of
memory without even evaluating anything. if problems *do* arise we can
probably recover the old table-based system by adding some tracking to
Pos::Origin (or increasing the size of PosIdx outright), but for time
being this looks like more complexity than it's worth.
since we now need to read the entire input again to determine the
line/column of a position we'll make unsafeGetAttrPos slightly lazy:
mostly the set it returns is only used to determine the file of origin
of an attribute, not its exact location. the thunks do not add
measurable runtime overhead.
notably this change is necessary to allow changing the parser since
apparently nothing supports nix's very idiosyncratic line ending choice
of "anything goes", making it very hard to calculate line/column
positions in the parser (while byte offsets are very easy).
(cherry picked from commit 5d9fdab3de0ee17c71369ad05806b9ea06dfceda)
Change-Id: Ie0b2430cb120c09097afa8c0101884d94f4bbf34
this needs a string comparison because there seems to be no other way to
get that information out of bison. usually the location info is going to
be correct (pointing at a bad token), but since EOF isn't a token as
such it'll be wrong in that this case.
this hasn't shown up much so far because a single line ending *is* a
token, so any file formatted in the usual manner (ie, ending in a line
ending) would have its EOF position reported correctly.
(cherry picked from commit 855fd5a1bb781e4f722c1d757ba43e866d370132)
Change-Id: I120c56a962f4286b1ae3b71da7b71ce8ec3e0535
the parser treats a plain \r as a newline, error reports do not. this
can lead to interesting divergences if anything makes use of this
feature, with error reports pointing to wrong locations in the input (or
even outside the input altogether).
(cherry picked from commit 2be6b143289e5479cc4a2667bb84e879116c2447)
Change-Id: Ieb7f7655bac8cb0cf5734c60bd41723388f2973c
previously we reported the error at the beginning of the binding
block (for plain inherits) or the beginning of the attr list (for
inherit-from), effectively hiding where exactly the error happened.
this also carries over to runtime positions of attributes in sets as
reported by unsafeGetAttrPos. we're not worried about this changing
observable eval behavior because it *is* marked unsafe, and the new
behavior is much more useful.
(cherry picked from commit 1edd6fada53553b89847ac3981ac28025857ca02)
Change-Id: I2f50eb9f3dc3977db4eb3e3da96f1cb37ccd5174
we already normalize attr order to lexicographic, doing the same for
formals makes sense. doubly so because the order of formals would
otherwise depend on the context of the expression, which is not quite as
useful as one might expect.
(cherry picked from commit 4147ecfb1c51f3fe3b4adcbd4e753fd487dab645)
Change-Id: I3fd0dbdef3ac7447a3a03ff20bb514a0d0f23fb1
the parser modifies its inputs, which means that sharing them between
the error context reporting system and the parser itself can confuse the
reporting system. usually this led to early truncation of error context
reports which, while not dangerous, can be quite confusing.
(cherry picked from commit d384ecd553aa997270b79ee98d02f7cf7e1849e6)
Change-Id: I677646b5675b12b2faa787943646aa36dc6e6ee3
diagnose attr duplication at the path the duplication was detected, not
at the path the current attribute wanted to place. doing the latter is
only correct if a leaf attribute was duplicated, not if an attrpath was
set to a non-attrset in one binding and a (potentially implied) attrset
in another binding.
fixes#124
Change-Id: Ic4aa9cc12a9874d4e7897c6f64408f10aa36fc82
using the total-attrs-printed and total-list-items-printed counters to
calculate how many attrs were elided only works properly if no nesting
is involved. once things do nest the global counter can exceed the size
of the currently printed object, leading to unsigned wrapping and great
overestimation of elided counts. counting locally in addition to global
counts fixes this.
these are functional tests because creating these objects requires the
evaluator to not be a huge amount of code, and we also want defaults to
be tested for cli usage.
fixes#14
Change-Id: Icb9a0cb21b2f4bacbc5e9dcdd8c0b9055b4088a7
desugaring inherit-from to syntactic duplication of the source expr also
duplicates side effects of the source expr (such as trace calls) and
expensive computations (such as derivationStrict).
(cherry picked from commit cefd0302b55b3360dbca59cfcb4bf6a750d6cdcf)
Change-Id: Iff519f991adef2e51683ba2c552d37a3df7a179e
for plain inherits this is really just a stylistic choice, but for
inherit-from it actually fixes an exponential size increase problem
during expr printing (as may happen during assertion failure reporting,
on during duplicate attr detection in the parser)
(cherry picked from commit ecf8b12d60ad2929f9998666cf0966475b91e291)
Change-Id: Ie55f0cb01a37e766414c31f8d40f51c2c7d106b0
this also has the effect of sorting let bindings lexicographically
rather than by symbol creation order as was previously done, giving a
better canonicalization in the process.
(cherry picked from commit 6c08fba533ef31cad2bdc03ba72ecf58dc8ee5a0)
Change-Id: Ia887f629305645bb8a165fbbc0d32e620912595a
Do not skip any stack frames when `--show-trace` is given
(cherry picked from commit 0b47783d0a879875d558f0b56e49584f25ceb2d0)
Change-Id: Ia0f18266dbcf97543110110c655c219c7a3e3270
Pretty-print values in the REPL by printing each item in a list or
attrset on a separate line. When possible, single-item lists and
attrsets are printed on one line, as long as they don't contain a nested
list, attrset, or thunk.
Before:
```
{ attrs = { a = { b = { c = { }; }; }; }; list = [ 1 ]; list' = [ 1 2 3 ]; }
```
After:
```
{
attrs = {
a = {
b = {
c = { };
};
};
};
list = [ 1 ];
list' = [
1
2
3
];
}
```
(cherry picked from commit c0a15fb7d03dfb8f53bc6726c414bc88aa362592)
Change-Id: Ia2b41849165a5ddb63f7a8c272a2476b3e4292df
While preparing PRs like #9753, I've had to change error messages in
dozens of code paths. It would be nice if instead of
EvalError("expected 'boolean' but found '%1%'", showType(v))
we could write
TypeError(v, "boolean")
or similar. Then, changing the error message could be a mechanical
refactor with the compiler pointing out places the constructor needs to
be changed, rather than the error-prone process of grepping through the
codebase. Structured errors would also help prevent the "same" error
from having multiple slightly different messages, and could be a first
step towards error codes / an error index.
This PR reworks the exception infrastructure in `libexpr` to
support exception types with different constructor signatures than
`BaseError`. Actually refactoring the exceptions to use structured data
will come in a future PR (this one is big enough already, as it has to
touch every exception in `libexpr`).
The core design is in `eval-error.hh`. Generally, errors like this:
state.error("'%s' is not a string", getAttrPathStr())
.debugThrow<TypeError>()
are transformed like this:
state.error<TypeError>("'%s' is not a string", getAttrPathStr())
.debugThrow()
The type annotation has moved from `ErrorBuilder::debugThrow` to
`EvalState::error`.
(cherry picked from commit c6a89c1a1659b31694c0fbcd21d78a6dd521c732)
Change-Id: Iced91ba4e00ca9e801518071fb43798936cbd05a
Print the value in `error: cannot coerce` messages
(cherry picked from commit 5b7bfd2d6b89d7dd5f54c1ca6c8072358d31a84e)
===
test taken from 6e8d5983143ae576e3f4b1d2954a5267f2943a49; it was added
previously (and not backported because its pr was a mostly-revert), but
it's useful to have around.
Change-Id: Icbd14b55e3610ce7b774667bf14b82e6dc717982
Print the value in `value is X while a Y is expected` error
(cherry picked from commit 5f72a97092da6af28a7d2b2a50d74e9d34fae7e1)
Change-Id: Idb4bc903ae59a0f5b6fb3b1da4d47970fe0a6efe
Previously, there were two mostly-identical value printers -- one in
`libexpr/eval.cc` (which didn't force values) and one in
`libcmd/repl.cc` (which did force values and also printed ANSI color
codes).
This PR unifies both of these printers into `print.cc` and provides a
`PrintOptions` struct for controlling the output, which allows for
toggling whether values are forced, whether repeated values are tracked,
and whether ANSI color codes are displayed.
Additionally, `PrintOptions` allows tuning the maximum number of
attributes, list items, and bytes in a string that will be displayed;
this makes it ideal for contexts where printing too much output (e.g.
all of Nixpkgs) is distracting. (As requested by @roberth in
https://github.com/NixOS/nix/pull/9554#issuecomment-1845095735)
Please read the tests for example output.
Future work:
- It would be nice to provide this function as a builtin, perhaps
`builtins.toStringDebug` -- a printing function that never fails would
be useful when debugging Nix code.
- It would be nice to support customizing `PrintOptions` members on the
command line, e.g. `--option to-string-max-attrs 1000`.
(cherry picked from commit 0fa08b451682fb3311fe58112ff05c4fe5bee3a4, )
===
Restore ambiguous value printer for `nix-instantiate`
The Nix team has requested that this output format remain unchanged.
I've added a warning to the man page explaining that `nix-instantiate
--eval` output will not parse correctly in many situations.
(cherry picked from commit df84dd4d8dd3fd6381ac2ca3064432ab31a16b79)
Change-Id: I7cca6b4b53cd0642f2d49af657d5676a8554c9f8
Restore `builtins.pathExists` behavior on broken symlinks
(cherry picked from commit d53c8901ef7f2033855dd99063522e3d56a19dab)
===
note that this variant differs markedly from the source commit because
we haven't endured quite as much lazy trees.
Change-Id: I0facf282f21fe0db4134be5c65a8368c1b3a06fc
Color `diff` output in `tests/functional/lang` tests
(cherry picked from commit 1dc55c0f2f034bce6e3de4a5cda96d686b10a7f8)
Change-Id: Ie9b3fc3446bd3caa0fd8885de88639516a2ff862
tests/functional/lang: Test substring with negative length
(cherry picked from commit 86156d05dd33f856d8804f89669a7fe9b81f1a0d)
Change-Id: I2e2086027a43f8111ba5068ac16590eaa0b798d4
Fix performance of builtins.substring for empty substrings
(cherry picked from commit b2deff1947c2fe57fdbf1a472eb9003eb407f8d3)
Change-Id: I4ddfc8d26a4781c9520fff9807849a073ee7bed8
Fix segfault on infinite recursion in some cases
(cherry picked from commit bf1b294bd81ca76c5ec9fe3ecd52196bf52a8300)
Change-Id: Id137541426ec8536567835953fccf986a3aebf16
I think it is bad for these reasons when `tests/` contains a mix of
functional and integration tests
- Concepts is harder to understand, the documentation makes a good
unit vs functional vs integration distinction, but when the
integration tests are just two subdirs within `tests/` this is not
clear.
- Source filtering in the `flake.nix` is more complex. We need to
filter out some of the dirs from `tests/`, rather than simply pick
the dirs we want and take all of them. This is a good sign the
structure of what we are trying to do is not matching the structure
of the files.
With this change we have a clean:
```shell-session
$ git show 'HEAD:tests'
tree HEAD:tests
functional/
installer/
nixos/
```
(cherry picked from commit 68c81c737571794f7246db53fb4774e94fcf4b7e)
