lix/src/libexpr/nixexpr.hh

510 lines
15 KiB
C++
Raw Normal View History

#pragma once
///@file
2003-10-30 16:11:24 +00:00
#include <map>
#include <vector>
#include "value.hh"
#include "symbol-table.hh"
2020-05-11 21:52:15 +00:00
#include "error.hh"
#include "position.hh"
#include "eval-error.hh"
#include "pos-idx.hh"
#include "pos-table.hh"
namespace nix {
2003-10-30 16:11:24 +00:00
struct Env;
struct Value;
2014-01-21 17:29:55 +00:00
class EvalState;
struct ExprWith;
struct StaticEnv;
/**
* An attribute path is a sequence of attribute names.
*/
struct AttrName
{
Symbol symbol;
std::unique_ptr<Expr> expr;
AttrName(Symbol s) : symbol(s) {};
AttrName(std::unique_ptr<Expr> e) : expr(std::move(e)) {};
};
typedef std::vector<AttrName> AttrPath;
std::string showAttrPath(const SymbolTable & symbols, const AttrPath & attrPath);
/* Abstract syntax of Nix expressions. */
struct Expr
{
protected:
Expr(Expr &&) = default;
Expr & operator=(Expr &&) = default;
public:
struct AstSymbols {
Symbol sub, lessThan, mul, div, or_, findFile, nixPath, body;
};
Expr() = default;
Expr(const Expr &) = delete;
Expr & operator=(const Expr &) = delete;
2014-01-21 17:29:55 +00:00
virtual ~Expr() { };
virtual void show(const SymbolTable & symbols, std::ostream & str) const;
virtual void bindVars(EvalState & es, const std::shared_ptr<const StaticEnv> & env);
2010-04-12 22:03:27 +00:00
virtual void eval(EvalState & state, Env & env, Value & v);
virtual Value * maybeThunk(EvalState & state, Env & env);
virtual void setName(Symbol name);
virtual PosIdx getPos() const { return noPos; }
};
#define COMMON_METHODS \
2022-06-02 14:55:28 +00:00
void show(const SymbolTable & symbols, std::ostream & str) const override; \
void eval(EvalState & state, Env & env, Value & v) override; \
void bindVars(EvalState & es, const std::shared_ptr<const StaticEnv> & env) override;
struct ExprInt : Expr
{
NixInt n;
Value v;
2022-01-04 17:40:39 +00:00
ExprInt(NixInt n) : n(n) { v.mkInt(n); };
language: cleanly ban integer overflows This also bans various sneaking of negative numbers from the language into unsuspecting builtins as was exposed while auditing the consequences of changing the Nix language integer type to a newtype. It's unlikely that this change comprehensively ensures correctness when passing integers out of the Nix language and we should probably add a checked-narrowing function or something similar, but that's out of scope for the immediate change. During the development of this I found a few fun facts about the language: - You could overflow integers by converting from unsigned JSON values. - You could overflow unsigned integers by converting negative numbers into them when going into Nix config, into fetchTree, and into flake inputs. The flake inputs and Nix config cannot actually be tested properly since they both ban thunks, however, we put in checks anyway because it's possible these could somehow be used to do such shenanigans some other way. Note that Lix has banned Nix language integer overflows since the very first public beta, but threw a SIGILL about them because we run with -fsanitize=signed-overflow -fsanitize-undefined-trap-on-error in production builds. Since the Nix language uses signed integers, overflow was simply undefined behaviour, and since we defined that to trap, it did. Trapping on it was a bad UX, but we didn't even entirely notice that we had done this at all until it was reported as a bug a couple of months later (which is, to be fair, that flag working as intended), and it's got enough production time that, aside from code that is IMHO buggy (and which is, in any case, not in nixpkgs) such as https://git.lix.systems/lix-project/lix/issues/445, we don't think anyone doing anything reasonable actually depends on wrapping overflow. Even for weird use cases such as doing funny bit crimes, it doesn't make sense IMO to have wrapping behaviour, since two's complement arithmetic overflow behaviour is so *aggressively* not what you want for *any* kind of mathematics/algorithms. The Nix language exists for package management, a domain where bit crimes are already only dubiously in scope to begin with, and it makes a lot more sense for that domain for the integers to never lose precision, either by throwing errors if they would, or by being arbitrary-precision. This change will be ported to CppNix as well, to maintain language consistency. Fixes: https://git.lix.systems/lix-project/lix/issues/423 Change-Id: I51f253840c4af2ea5422b8a420aa5fafbf8fae75
2024-07-12 14:22:34 +00:00
ExprInt(NixInt::Inner n) : n(n) { v.mkInt(n); };
2022-06-02 14:55:28 +00:00
Value * maybeThunk(EvalState & state, Env & env) override;
COMMON_METHODS
};
struct ExprFloat : Expr
{
NixFloat nf;
Value v;
2022-01-04 17:40:39 +00:00
ExprFloat(NixFloat nf) : nf(nf) { v.mkFloat(nf); };
2022-06-02 14:55:28 +00:00
Value * maybeThunk(EvalState & state, Env & env) override;
COMMON_METHODS
};
struct ExprString : Expr
{
std::string s;
Value v;
2023-02-10 23:34:31 +00:00
ExprString(std::string &&s) : s(std::move(s)) { v.mkString(this->s.data()); };
2022-06-02 14:55:28 +00:00
Value * maybeThunk(EvalState & state, Env & env) override;
COMMON_METHODS
};
struct ExprPath : Expr
{
std::string s;
Value v;
ExprPath(std::string s) : s(std::move(s)) { v.mkPath(this->s.c_str()); };
2022-06-02 14:55:28 +00:00
Value * maybeThunk(EvalState & state, Env & env) override;
COMMON_METHODS
};
2020-02-24 13:33:01 +00:00
typedef uint32_t Level;
typedef uint32_t Displacement;
2013-10-08 12:24:53 +00:00
struct ExprVar : Expr
{
PosIdx pos;
Symbol name;
/* Whether the variable comes from an environment (e.g. a rec, let
or function argument) or from a "with".
`nullptr`: Not from a `with`.
Valid pointer: the nearest, innermost `with` expression to query first. */
ExprWith * fromWith;
2013-09-02 14:29:15 +00:00
/* In the former case, the value is obtained by going `level`
levels up from the current environment and getting the
`displ`th value in that environment. In the latter case, the
value is obtained by getting the attribute named `name` from
the set stored in the environment that is `level` levels up
from the current one.*/
2020-02-24 13:33:01 +00:00
Level level;
Displacement displ;
ExprVar(Symbol name) : name(name) { };
ExprVar(const PosIdx & pos, Symbol name) : pos(pos), name(name) { };
2022-06-02 14:55:28 +00:00
Value * maybeThunk(EvalState & state, Env & env) override;
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
/**
* A pseudo-expression for the purpose of evaluating the `from` expression in `inherit (from)` syntax.
* Unlike normal variable references, the displacement is set during parsing, and always refers to
* `ExprAttrs::inheritFromExprs` (by itself or in `ExprLet`), whose values are put into their own `Env`.
*/
struct ExprInheritFrom : ExprVar
{
ref<Expr> fromExpr;
ExprInheritFrom(PosIdx pos, Displacement displ, ref<Expr> fromExpr)
: ExprVar(pos, {}), fromExpr(fromExpr)
{
this->level = 0;
this->displ = displ;
this->fromWith = nullptr;
}
void show(SymbolTable const & symbols, std::ostream & str) const override;
void bindVars(EvalState & es, const std::shared_ptr<const StaticEnv> & env) override;
};
struct ExprSelect : Expr
{
PosIdx pos;
/** The expression attributes are being selected on. e.g. `foo` in `foo.bar.baz`. */
std::unique_ptr<Expr> e;
/** A default value specified with `or`, if the selected attr doesn't exist.
* e.g. `bix` in `foo.bar.baz or bix`
*/
std::unique_ptr<Expr> def;
/** The path of attributes being selected. e.g. `bar.baz` in `foo.bar.baz.` */
AttrPath attrPath;
ExprSelect(const PosIdx & pos, std::unique_ptr<Expr> e, AttrPath attrPath, std::unique_ptr<Expr> def) : pos(pos), e(std::move(e)), def(std::move(def)), attrPath(std::move(attrPath)) { };
ExprSelect(const PosIdx & pos, std::unique_ptr<Expr> e, Symbol name) : pos(pos), e(std::move(e)) { attrPath.push_back(AttrName(name)); };
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
2010-04-12 21:21:24 +00:00
struct ExprOpHasAttr : Expr
{
std::unique_ptr<Expr> e;
AttrPath attrPath;
ExprOpHasAttr(std::unique_ptr<Expr> e, AttrPath attrPath) : e(std::move(e)), attrPath(std::move(attrPath)) { };
PosIdx getPos() const override { return e->getPos(); }
2010-04-12 21:21:24 +00:00
COMMON_METHODS
};
struct ExprAttrs : Expr
{
bool recursive;
PosIdx pos;
struct AttrDef {
enum class Kind {
/** `attr = expr;` */
Plain,
/** `inherit attr1 attrn;` */
Inherited,
/** `inherit (expr) attr1 attrn;` */
InheritedFrom,
};
Kind kind;
std::unique_ptr<Expr> e;
PosIdx pos;
2020-02-24 13:33:01 +00:00
Displacement displ; // displacement
AttrDef(std::unique_ptr<Expr> e, const PosIdx & pos, Kind kind = Kind::Plain)
: kind(kind), e(std::move(e)), pos(pos) { };
AttrDef() { };
template<typename T>
const T & chooseByKind(const T & plain, const T & inherited, const T & inheritedFrom) const
{
switch (kind) {
case Kind::Plain:
return plain;
case Kind::Inherited:
return inherited;
default:
case Kind::InheritedFrom:
return inheritedFrom;
}
}
};
typedef std::map<Symbol, AttrDef> AttrDefs;
AttrDefs attrs;
std::unique_ptr<std::vector<ref<Expr>>> inheritFromExprs;
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
struct DynamicAttrDef {
std::unique_ptr<Expr> nameExpr, valueExpr;
PosIdx pos;
DynamicAttrDef(std::unique_ptr<Expr> nameExpr, std::unique_ptr<Expr> valueExpr, const PosIdx & pos)
: nameExpr(std::move(nameExpr)), valueExpr(std::move(valueExpr)), pos(pos) { };
Dynamic attrs This adds new syntax for attribute names: * attrs."${name}" => getAttr name attrs * attrs ? "${name}" => isAttrs attrs && hasAttr attrs name * attrs."${name}" or def => if attrs ? "${name}" then attrs."${name}" else def * { "${name}" = value; } => listToAttrs [{ inherit name value; }] Of course, it's a bit more complicated than that. The attribute chains can be arbitrarily long and contain combinations of static and dynamic parts (e.g. attrs."${foo}".bar."${baz}" or qux), which is relatively straightforward for the getAttrs/hasAttrs cases but is more complex for the listToAttrs case due to rules about duplicate attribute definitions. For attribute sets with dynamic attribute names, duplicate static attributes are detected at parse time while duplicate dynamic attributes are detected when the attribute set is forced. So, for example, { a = null; a.b = null; "${"c"}" = true; } will be a parse-time error, while { a = {}; "${"a"}".b = null; c = true; } will be an eval-time error (technically that case could theoretically be detected at parse time, but the general case would require full evaluation). Moreover, duplicate dynamic attributes are not allowed even in cases where they would be with static attributes ({ a.b.d = true; a.b.c = false; } is legal, but { a."${"b"}".d = true; a."${"b"}".c = false; } is not). This restriction might be relaxed in the future in cases where the static variant would not be an error, but it is not obvious that that is desirable. Finally, recursive attribute sets with dynamic attributes have the static attributes in scope but not the dynamic ones. So rec { a = true; "${"b"}" = a; } is equivalent to { a = true; b = true; } but rec { "${"a"}" = true; b = a; } would be an error or use a from the surrounding scope if it exists. Note that the getAttr, getAttr or default, and hasAttr are all implemented purely in the parser as syntactic sugar, while attribute sets with dynamic attribute names required changes to the AST to be implemented cleanly. This is an alternative solution to and closes #167 Signed-off-by: Shea Levy <shea@shealevy.com>
2013-09-21 03:25:30 +00:00
};
typedef std::vector<DynamicAttrDef> DynamicAttrDefs;
DynamicAttrDefs dynamicAttrs;
ExprAttrs(const PosIdx &pos) : recursive(false), pos(pos) { };
ExprAttrs() : recursive(false) { };
PosIdx getPos() const override { return pos; }
COMMON_METHODS
std::shared_ptr<const StaticEnv> bindInheritSources(
EvalState & es, const std::shared_ptr<const StaticEnv> & env);
Env * buildInheritFromEnv(EvalState & state, Env & up);
void showBindings(const SymbolTable & symbols, std::ostream & str) const;
};
struct ExprList : Expr
{
std::vector<std::unique_ptr<Expr>> elems;
ExprList() { };
COMMON_METHODS
Value * maybeThunk(EvalState & state, Env & env) override;
PosIdx getPos() const override
{
return elems.empty() ? noPos : elems.front()->getPos();
}
};
2003-10-30 16:11:24 +00:00
struct Formal
{
PosIdx pos;
Symbol name;
std::unique_ptr<Expr> def;
};
/** Attribute set destructuring in arguments of a lambda, if present */
struct Formals
{
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
typedef std::vector<Formal> Formals_;
Formals_ formals;
bool ellipsis;
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
bool has(Symbol arg) const
{
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
auto it = std::lower_bound(formals.begin(), formals.end(), arg,
[] (const Formal & f, const Symbol & sym) { return f.name < sym; });
return it != formals.end() && it->name == arg;
}
std::vector<std::reference_wrapper<const Formal>> lexicographicOrder(const SymbolTable & symbols) const
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
{
std::vector<std::reference_wrapper<const Formal>> result(formals.begin(), formals.end());
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
std::sort(result.begin(), result.end(),
[&] (const Formal & a, const Formal & b) {
std::string_view sa = symbols[a.name], sb = symbols[b.name];
return sa < sb;
defer formals duplicate check for incresed efficiency all round if we defer the duplicate argument check for lambda formals we can use more efficient data structures for the formals set, and we can get rid of the duplication of formals names to boot. instead of a list of formals we've seen and a set of names we'll keep a vector instead and run a sort+dupcheck step before moving the parsed formals into a newly created lambda. this improves performance on search and rebuild by ~1%, pure parsing gains more (about 4%). this does reorder lambda arguments in the xml output, but the output is still stable. this shouldn't be a problem since argument order is not semantically important anyway. before nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.550 s ± 0.060 s [User: 6.470 s, System: 1.664 s] Range (min … max): 8.435 s … 8.666 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 346.7 ms ± 2.1 ms [User: 312.4 ms, System: 34.2 ms] Range (min … max): 343.8 ms … 353.4 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.720 s ± 0.031 s [User: 2.415 s, System: 0.231 s] Range (min … max): 2.662 s … 2.780 s 20 runs after nix search --no-eval-cache --offline ../nixpkgs hello Time (mean ± σ): 8.462 s ± 0.063 s [User: 6.398 s, System: 1.661 s] Range (min … max): 8.339 s … 8.542 s 20 runs nix eval -f ../nixpkgs/pkgs/development/haskell-modules/hackage-packages.nix Time (mean ± σ): 329.1 ms ± 1.4 ms [User: 296.8 ms, System: 32.3 ms] Range (min … max): 326.1 ms … 330.8 ms 20 runs nix eval --raw --impure --expr 'with import <nixpkgs/nixos> {}; system' Time (mean ± σ): 2.687 s ± 0.035 s [User: 2.392 s, System: 0.228 s] Range (min … max): 2.626 s … 2.754 s 20 runs
2022-01-19 15:49:02 +00:00
});
return result;
}
};
struct ExprLambda : Expr
{
/** Where the lambda is defined in Nix code. May be falsey if the
* position is not known. */
PosIdx pos;
/** Name of the lambda. This is set if the lambda is defined in a
* let-expression or an attribute set, such that there is a name.
* Lambdas may have a falsey symbol as the name if they are anonymous */
Symbol name;
/** The argument name of this particular lambda. Is a falsey symbol if there
* is no such argument. */
Symbol arg;
/** Formals are present when the lambda destructures an attr set as
* argument, with or without ellipsis */
std::unique_ptr<Formals> formals;
std::unique_ptr<Expr> body;
ExprLambda(PosIdx pos, Symbol arg, std::unique_ptr<Formals> formals, std::unique_ptr<Expr> body)
: pos(pos), arg(arg), formals(std::move(formals)), body(std::move(body))
{
2020-04-29 16:14:32 +00:00
};
ExprLambda(PosIdx pos, std::unique_ptr<Formals> formals, std::unique_ptr<Expr> body)
: pos(pos), formals(std::move(formals)), body(std::move(body))
{
}
2022-06-02 14:55:28 +00:00
void setName(Symbol name) override;
std::string showNamePos(const EvalState & state) const;
inline bool hasFormals() const { return formals != nullptr; }
PosIdx getPos() const override { return pos; }
/** Returns the name of the lambda,
* or "anonymous lambda" if it doesn't have one.
*/
inline std::string getName(SymbolTable const & symbols) const
{
if (this->name) {
return symbols[this->name];
}
return "anonymous lambda";
}
/** Returns the name of the lambda in single quotes,
* or "anonymous lambda" if it doesn't have one.
*/
inline std::string getQuotedName(SymbolTable const & symbols) const
{
if (this->name) {
return concatStrings("'", symbols[this->name], "'");
}
return "anonymous lambda";
}
COMMON_METHODS
};
struct ExprCall : Expr
{
std::unique_ptr<Expr> fun;
std::vector<std::unique_ptr<Expr>> args;
PosIdx pos;
ExprCall(const PosIdx & pos, std::unique_ptr<Expr> fun, std::vector<std::unique_ptr<Expr>> && args)
: fun(std::move(fun)), args(std::move(args)), pos(pos)
{ }
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
struct ExprLet : Expr
{
std::unique_ptr<ExprAttrs> attrs;
std::unique_ptr<Expr> body;
ExprLet(std::unique_ptr<ExprAttrs> attrs, std::unique_ptr<Expr> body) : attrs(std::move(attrs)), body(std::move(body)) { };
COMMON_METHODS
};
struct ExprWith : Expr
{
PosIdx pos;
std::unique_ptr<Expr> attrs, body;
2018-05-02 11:56:34 +00:00
size_t prevWith;
ExprWith * parentWith;
ExprWith(const PosIdx & pos, std::unique_ptr<Expr> attrs, std::unique_ptr<Expr> body) : pos(pos), attrs(std::move(attrs)), body(std::move(body)) { };
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
struct ExprIf : Expr
{
PosIdx pos;
std::unique_ptr<Expr> cond, then, else_;
ExprIf(const PosIdx & pos, std::unique_ptr<Expr> cond, std::unique_ptr<Expr> then, std::unique_ptr<Expr> else_) : pos(pos), cond(std::move(cond)), then(std::move(then)), else_(std::move(else_)) { };
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
2010-04-12 21:21:24 +00:00
struct ExprAssert : Expr
{
PosIdx pos;
std::unique_ptr<Expr> cond, body;
ExprAssert(const PosIdx & pos, std::unique_ptr<Expr> cond, std::unique_ptr<Expr> body) : pos(pos), cond(std::move(cond)), body(std::move(body)) { };
PosIdx getPos() const override { return pos; }
2010-04-12 21:21:24 +00:00
COMMON_METHODS
};
struct ExprOpNot : Expr
{
std::unique_ptr<Expr> e;
ExprOpNot(std::unique_ptr<Expr> e) : e(std::move(e)) { };
PosIdx getPos() const override { return e->getPos(); }
2010-04-12 21:21:24 +00:00
COMMON_METHODS
};
#define MakeBinOp(name, s) \
struct name : Expr \
{ \
PosIdx pos; \
std::unique_ptr<Expr> e1, e2; \
name(std::unique_ptr<Expr> e1, std::unique_ptr<Expr> e2) : e1(std::move(e1)), e2(std::move(e2)) { }; \
name(const PosIdx & pos, std::unique_ptr<Expr> e1, std::unique_ptr<Expr> e2) : pos(pos), e1(std::move(e1)), e2(std::move(e2)) { }; \
2022-06-02 14:55:28 +00:00
void show(const SymbolTable & symbols, std::ostream & str) const override \
{ \
str << "("; e1->show(symbols, str); str << " " s " "; e2->show(symbols, str); str << ")"; \
} \
2022-06-02 14:55:28 +00:00
void bindVars(EvalState & es, const std::shared_ptr<const StaticEnv> & env) override \
{ \
e1->bindVars(es, env); e2->bindVars(es, env); \
} \
2022-06-02 14:55:28 +00:00
void eval(EvalState & state, Env & env, Value & v) override; \
PosIdx getPos() const override { return pos; } \
};
MakeBinOp(ExprOpEq, "==")
MakeBinOp(ExprOpNEq, "!=")
MakeBinOp(ExprOpAnd, "&&")
MakeBinOp(ExprOpOr, "||")
MakeBinOp(ExprOpImpl, "->")
MakeBinOp(ExprOpUpdate, "//")
MakeBinOp(ExprOpConcatLists, "++")
2010-04-12 21:21:24 +00:00
struct ExprConcatStrings : Expr
{
PosIdx pos;
bool forceString;
std::vector<std::pair<PosIdx, std::unique_ptr<Expr>>> es;
ExprConcatStrings(const PosIdx & pos, bool forceString, std::vector<std::pair<PosIdx, std::unique_ptr<Expr>>> es)
: pos(pos), forceString(forceString), es(std::move(es)) { };
PosIdx getPos() const override { return pos; }
2010-04-12 21:21:24 +00:00
COMMON_METHODS
};
struct ExprPos : Expr
{
PosIdx pos;
ExprPos(const PosIdx & pos) : pos(pos) { };
PosIdx getPos() const override { return pos; }
COMMON_METHODS
};
/* only used to mark thunks as black holes. */
struct ExprBlackHole : Expr
{
void show(const SymbolTable & symbols, std::ostream & str) const override {}
void eval(EvalState & state, Env & env, Value & v) override;
void bindVars(EvalState & es, const std::shared_ptr<const StaticEnv> & env) override {}
};
extern ExprBlackHole eBlackHole;
2010-04-08 11:41:19 +00:00
/* Static environments are used to map variable names onto (level,
displacement) pairs used to obtain the value of the variable at
runtime. */
struct StaticEnv
{
ExprWith * isWith;
const StaticEnv * up;
// Note: these must be in sorted order.
2020-02-24 13:33:01 +00:00
typedef std::vector<std::pair<Symbol, Displacement>> Vars;
Vars vars;
StaticEnv(ExprWith * isWith, const StaticEnv * up, size_t expectedSize = 0) : isWith(isWith), up(up) {
vars.reserve(expectedSize);
};
void sort()
{
std::stable_sort(vars.begin(), vars.end(),
[](const Vars::value_type & a, const Vars::value_type & b) { return a.first < b.first; });
}
void deduplicate()
{
auto it = vars.begin(), jt = it, end = vars.end();
while (jt != end) {
*it = *jt++;
while (jt != end && it->first == jt->first) *it = *jt++;
it++;
}
vars.erase(it, end);
}
Vars::const_iterator find(Symbol name) const
{
Vars::value_type key(name, 0);
auto i = std::lower_bound(vars.begin(), vars.end(), key);
if (i != vars.end() && i->first == name) return i;
return vars.end();
}
};
}