lix/src/libexpr/eval.hh

798 lines
23 KiB
C++
Raw Normal View History

#pragma once
///@file
#include "attr-set.hh"
#include "eval-error.hh"
#include "types.hh"
#include "value.hh"
#include "nixexpr.hh"
#include "symbol-table.hh"
#include "config.hh"
#include "experimental-features.hh"
#include "search-path.hh"
#include "repl-exit-status.hh"
#include <gc/gc_allocator.h>
#include <map>
2019-10-27 09:15:51 +00:00
#include <optional>
#include <unordered_map>
#include <functional>
namespace nix {
/**
* We put a limit on primop arity because it lets us use a fixed size array on
* the stack. 8 is already an impractical number of arguments. Use an attrset
* argument for such overly complicated functions.
*/
constexpr size_t maxPrimOpArity = 8;
class Store;
class EvalState;
class StorePath;
struct SingleDerivedPath;
enum RepairFlag : bool;
struct MemoryInputAccessor;
namespace eval_cache {
class EvalCache;
}
/** Alias for std::map which uses boehmgc's allocator conditional on us actually
* using boehmgc in this build.
*/
#if HAVE_BOEHMGC
template<typename KeyT, typename ValueT>
using GcMap = std::map<
KeyT,
ValueT,
std::less<KeyT>,
traceable_allocator<std::pair<KeyT const, ValueT>>
>;
#else
using GcMap = std::map<KeyT, ValueT>
#endif
/**
* Function that implements a primop.
*/
using PrimOpImpl = void(EvalState & state, PosIdx pos, Value ** args, Value & v);
/**
* Info about a primitive operation, and its implementation
*/
struct PrimOp
{
/**
* Name of the primop. `__` prefix is treated specially.
*/
std::string name;
/**
* Names of the parameters of a primop, for primops that take a
* fixed number of arguments to be substituted for these parameters.
*/
std::vector<std::string> args;
/**
* Aritiy of the primop.
*
* If `args` is not empty, this field will be computed from that
* field instead, so it doesn't need to be manually set.
*/
size_t arity = 0;
/**
* Optional free-form documentation about the primop.
*/
const char * doc = nullptr;
/**
* Implementation of the primop.
*/
std::function<PrimOpImpl> fun;
/**
* Optional experimental for this to be gated on.
*/
std::optional<ExperimentalFeature> experimentalFeature;
/**
* Validity check to be performed by functions that introduce primops,
* such as RegisterPrimOp() and Value::mkPrimOp().
*/
void check();
};
Unify and refactor value printing 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
2024-03-08 02:05:47 +00:00
std::ostream & operator<<(std::ostream & output, PrimOp & primOp);
/**
* Info about a constant
*/
struct Constant
{
/**
* Optional type of the constant (known since it is a fixed value).
*
* @todo we should use an enum for this.
*/
ValueType type = nThunk;
/**
* Optional free-form documentation about the constant.
*/
const char * doc = nullptr;
/**
* Whether the constant is impure, and not available in pure mode.
*/
bool impureOnly = false;
};
using ValMap = GcMap<std::string, Value *>;
struct Env
{
Env * up;
Value * values[0];
};
2022-05-19 23:01:23 +00:00
void printEnvBindings(const EvalState &es, const Expr & expr, const Env & env);
void printEnvBindings(const SymbolTable & st, const StaticEnv & se, const Env & env, int lvl = 0);
2022-05-05 21:43:23 +00:00
std::unique_ptr<ValMap> mapStaticEnvBindings(const SymbolTable & st, const StaticEnv & se, const Env & env);
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
void copyContext(const Value & v, NixStringContext & context);
2010-03-30 09:22:33 +00:00
Unify and refactor value printing 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
2024-03-08 02:05:47 +00:00
std::string printValue(EvalState & state, Value & v);
std::ostream & operator << (std::ostream & os, const ValueType t);
/**
* Initialise the Boehm GC, if applicable.
*/
void initGC();
struct RegexCache;
std::shared_ptr<RegexCache> makeRegexCache();
2021-12-23 20:36:39 +00:00
struct DebugTrace {
std::shared_ptr<Pos> pos;
const Expr & expr;
const Env & env;
HintFmt hint;
bool isError;
2021-12-23 20:36:39 +00:00
};
2022-05-25 16:21:20 +00:00
class EvalState : public std::enable_shared_from_this<EvalState>
{
2010-03-31 15:38:03 +00:00
public:
SymbolTable symbols;
PosTable positions;
2013-10-28 06:34:44 +00:00
const Symbol sWith, sOutPath, sDrvPath, sType, sMeta, sName, sValue,
sSystem, sOverrides, sOutputs, sOutputName, sIgnoreNulls,
sFile, sLine, sColumn, sFunctor, sToString,
sRight, sWrong, sStructuredAttrs, sBuilder, sArgs,
sContentAddressed, sImpure,
2018-11-29 18:18:36 +00:00
sOutputHash, sOutputHashAlgo, sOutputHashMode,
sRecurseForDerivations,
sDescription, sSelf, sEpsilon, sStartSet, sOperator, sKey, sPath,
sPrefix,
sOutputSpecified;
const Expr::AstSymbols exprSymbols;
/**
* If set, force copying files to the Nix store even if they
* already exist there.
*/
RepairFlag repair;
/**
* The allowed filesystem paths in restricted or pure evaluation
* mode.
*/
2019-02-12 12:43:32 +00:00
std::optional<PathSet> allowedPaths;
2022-01-04 18:23:11 +00:00
Bindings emptyBindings;
/**
* Empty list constant.
*/
Value vEmptyList;
2023-04-06 13:25:06 +00:00
const SourcePath derivationInternal;
/**
* Store used to materialise .drv files.
*/
const ref<Store> store;
/**
* Store used to build stuff.
*/
2021-06-29 19:09:48 +00:00
const ref<Store> buildStore;
RootValue vCallFlake = nullptr;
RootValue vImportedDrvToDerivation = nullptr;
2019-03-21 08:30:16 +00:00
/**
* Debugger
*/
std::function<ReplExitStatus(ref<EvalState> es, ValMap const & extraEnv)> debugRepl;
bool debugStop;
bool inDebugger = false;
int trylevel;
2021-12-23 20:36:39 +00:00
std::list<DebugTrace> debugTraces;
2022-05-25 10:32:22 +00:00
std::map<const Expr*, const std::shared_ptr<const StaticEnv>> exprEnvs;
const std::shared_ptr<const StaticEnv> getStaticEnv(const Expr & expr) const
{
auto i = exprEnvs.find(&expr);
2022-05-25 10:32:22 +00:00
if (i != exprEnvs.end())
return i->second;
else
return std::shared_ptr<const StaticEnv>();;
}
2022-05-20 16:33:50 +00:00
void runDebugRepl(const Error * error, const Env & env, const Expr & expr);
2022-05-06 14:47:21 +00:00
template<class T, typename... Args>
[[nodiscard, gnu::noinline]]
EvalErrorBuilder<T> & error(const Args & ... args) {
// `EvalErrorBuilder::debugThrow` performs the corresponding `delete`.
return *new EvalErrorBuilder<T>(*this, args...);
}
2022-05-19 23:01:23 +00:00
/**
* A cache for evaluation caches, so as to reuse the same root value if possible
*/
std::map<const Hash, ref<eval_cache::EvalCache>> evalCaches;
2010-03-31 15:38:03 +00:00
private:
/* Cache for calls to addToStore(); maps source paths to the store
paths. */
std::map<SourcePath, StorePath> srcToStore;
/**
* A cache from path names to parse trees.
*/
using FileParseCache = GcMap<SourcePath, Expr *>;
FileParseCache fileParseCache;
/**
* A cache from path names to values.
*/
using FileEvalCache = GcMap<SourcePath, Value>;
FileEvalCache fileEvalCache;
SearchPath searchPath;
std::map<std::string, std::optional<std::string>> searchPathResolved;
/**
* Cache used by checkSourcePath().
*/
std::unordered_map<Path, SourcePath> resolvedPaths;
/**
* Cache used by prim_match().
*/
std::shared_ptr<RegexCache> regexCache;
#if HAVE_BOEHMGC
/**
* Allocation cache for GC'd Value objects.
*/
std::shared_ptr<void *> valueAllocCache;
/**
* Allocation cache for size-1 Env objects.
*/
std::shared_ptr<void *> env1AllocCache;
#endif
2010-03-30 15:18:20 +00:00
public:
2013-09-02 14:29:15 +00:00
2021-06-29 19:09:48 +00:00
EvalState(
const SearchPath & _searchPath,
2021-06-29 19:09:48 +00:00
ref<Store> store,
std::shared_ptr<Store> buildStore = nullptr);
~EvalState();
SearchPath getSearchPath() { return searchPath; }
/**
* Return a `SourcePath` that refers to `path` in the root
* filesystem.
*/
SourcePath rootPath(CanonPath path);
/**
* Allow access to a path.
*/
2021-10-07 10:11:00 +00:00
void allowPath(const Path & path);
/**
* Allow access to a store path. Note that this gets remapped to
* the real store path if `store` is a chroot store.
*/
void allowPath(const StorePath & storePath);
/**
* Allow access to a store path and return it as a string.
*/
void allowAndSetStorePathString(const StorePath & storePath, Value & v);
/**
* Check whether access to a path is allowed and throw an error if
* not. Otherwise return the canonicalised path.
*/
SourcePath checkSourcePath(const SourcePath & path);
void checkURI(const std::string & uri);
/**
* When using a diverted store and 'path' is in the Nix store, map
* 'path' to the diverted location (e.g. /nix/store/foo is mapped
* to /home/alice/my-nix/nix/store/foo). However, this is only
* done if the context is not empty, since otherwise we're
* probably trying to read from the actual /nix/store. This is
* intended to distinguish between import-from-derivation and
* sources stored in the actual /nix/store.
*/
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
Path toRealPath(const Path & path, const NixStringContext & context);
/**
* Parse a Nix expression from the specified file.
*/
Expr & parseExprFromFile(const SourcePath & path);
Expr & parseExprFromFile(const SourcePath & path, std::shared_ptr<StaticEnv> & staticEnv);
/**
* Parse a Nix expression from the specified string.
*/
Expr & parseExprFromString(std::string s, const SourcePath & basePath, std::shared_ptr<StaticEnv> & staticEnv);
Expr & parseExprFromString(std::string s, const SourcePath & basePath);
2013-09-02 14:29:15 +00:00
Expr & parseStdin();
/**
* Evaluate an expression read from the given file to normal
* form. Optionally enforce that the top-level expression is
* trivial (i.e. doesn't require arbitrary computation).
*/
void evalFile(const SourcePath & path, Value & v, bool mustBeTrivial = false);
2010-03-30 09:22:33 +00:00
/**
* Like `evalFile`, but with an already parsed expression.
*/
void cacheFile(
const SourcePath & path,
const SourcePath & resolvedPath,
Expr * e,
Value & v,
bool mustBeTrivial = false);
2013-09-02 16:34:04 +00:00
void resetFileCache();
/**
* Look up a file in the search path.
*/
SourcePath findFile(const std::string_view path);
SourcePath findFile(const SearchPath & searchPath, const std::string_view path, const PosIdx pos = noPos);
/**
* Try to resolve a search path value (not the optinal key part)
*
* If the specified search path element is a URI, download it.
*
* If it is not found, return `std::nullopt`
*/
std::optional<std::string> resolveSearchPathPath(const SearchPath::Path & path);
/**
* Evaluate an expression to normal form
*
* @param [out] v The resulting is stored here.
*/
void eval(Expr & e, Value & v);
/**
* Evaluation the expression, then verify that it has the expected
* type.
*/
inline bool evalBool(Env & env, Expr & e);
inline bool evalBool(Env & env, Expr & e, const PosIdx pos, std::string_view errorCtx);
inline void evalAttrs(Env & env, Expr & e, Value & v, const PosIdx pos, std::string_view errorCtx);
/**
* If `v` is a thunk, enter it and overwrite `v` with the result
* of the evaluation of the thunk. If `v` is a delayed function
* application, call the function and overwrite `v` with the
* result. Otherwise, this is a no-op.
*/
inline void forceValue(Value & v, const PosIdx pos);
void tryFixupBlackHolePos(Value & v, PosIdx pos);
/**
* Force a value, then recursively force list elements and
* attributes.
*/
void forceValueDeep(Value & v);
/**
* Force `v`, and then verify that it has the expected type.
*/
NixInt forceInt(Value & v, const PosIdx pos, std::string_view errorCtx);
NixFloat forceFloat(Value & v, const PosIdx pos, std::string_view errorCtx);
bool forceBool(Value & v, const PosIdx pos, std::string_view errorCtx);
2022-02-03 23:31:33 +00:00
void forceAttrs(Value & v, const PosIdx pos, std::string_view errorCtx);
2022-02-03 23:31:33 +00:00
template <typename Callable>
inline void forceAttrs(Value & v, Callable getPos, std::string_view errorCtx);
inline void forceList(Value & v, const PosIdx pos, std::string_view errorCtx);
/**
* @param v either lambda or primop
*/
void forceFunction(Value & v, const PosIdx pos, std::string_view errorCtx);
std::string_view forceString(Value & v, const PosIdx pos, std::string_view errorCtx);
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
std::string_view forceString(Value & v, NixStringContext & context, const PosIdx pos, std::string_view errorCtx);
std::string_view forceStringNoCtx(Value & v, const PosIdx pos, std::string_view errorCtx);
template<typename... Args>
[[gnu::noinline]]
void addErrorTrace(Error & e, const Args & ... formatArgs) const;
template<typename... Args>
[[gnu::noinline]]
void addErrorTrace(Error & e, const PosIdx pos, const Args & ... formatArgs) const;
public:
/**
* @return true iff the value `v` denotes a derivation (i.e. a
* set with attribute `type = "derivation"`).
*/
bool isDerivation(Value & v);
std::optional<std::string> tryAttrsToString(const PosIdx pos, Value & v,
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
NixStringContext & context, bool coerceMore = false, bool copyToStore = true);
2019-10-27 09:15:51 +00:00
/**
* String coercion.
*
* Converts strings, paths and derivations to a
* string. If `coerceMore` is set, also converts nulls, integers,
* booleans and lists to a string. If `copyToStore` is set,
* referenced paths are copied to the Nix store as a side effect.
*/
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
BackedStringView coerceToString(const PosIdx pos, Value & v, NixStringContext & context,
2022-11-28 23:25:36 +00:00
std::string_view errorCtx,
2021-07-29 16:03:07 +00:00
bool coerceMore = false, bool copyToStore = true,
2022-11-28 23:25:36 +00:00
bool canonicalizePath = true);
2010-03-30 09:22:33 +00:00
StorePath copyPathToStore(NixStringContext & context, const SourcePath & path);
2013-11-18 23:03:11 +00:00
/**
* Path coercion.
*
* Converts strings, paths and derivations to a
* path. The result is guaranteed to be a canonicalised, absolute
* path. Nothing is copied to the store.
*/
SourcePath coerceToPath(const PosIdx pos, Value & v, NixStringContext & context, std::string_view errorCtx);
2010-03-30 09:22:33 +00:00
/**
* Like coerceToPath, but the result must be a store path.
*/
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
StorePath coerceToStorePath(const PosIdx pos, Value & v, NixStringContext & context, std::string_view errorCtx);
/**
* Part of `coerceToSingleDerivedPath()` without any store IO which is exposed for unit testing only.
*/
std::pair<SingleDerivedPath, std::string_view> coerceToSingleDerivedPathUnchecked(const PosIdx pos, Value & v, std::string_view errorCtx);
/**
* Coerce to `SingleDerivedPath`.
*
* Must be a string which is either a literal store path or a
* "placeholder (see `DownstreamPlaceholder`).
*
* Even more importantly, the string context must be exactly one
* element, which is either a `NixStringContextElem::Opaque` or
* `NixStringContextElem::Built`. (`NixStringContextEleme::DrvDeep`
* is not permitted).
*
* The string is parsed based on the context --- the context is the
* source of truth, and ultimately tells us what we want, and then
* we ensure the string corresponds to it.
*/
SingleDerivedPath coerceToSingleDerivedPath(const PosIdx pos, Value & v, std::string_view errorCtx);
2013-09-02 16:34:04 +00:00
public:
/**
* The base environment, containing the builtin functions and
* values.
*/
Env & baseEnv;
/**
* The same, but used during parsing to resolve variables.
*/
2021-09-14 16:49:22 +00:00
std::shared_ptr<StaticEnv> staticBaseEnv; // !!! should be private
2010-04-14 22:59:39 +00:00
/**
* Name and documentation about every constant.
*
* Constants from primops are hard to crawl, and their docs will go
* here too.
*/
std::vector<std::pair<std::string, Constant>> constantInfos;
2010-04-14 22:59:39 +00:00
private:
2013-09-02 14:29:15 +00:00
2015-07-23 21:14:07 +00:00
unsigned int baseEnvDispl = 0;
2013-09-02 16:34:04 +00:00
void createBaseEnv();
2013-09-02 14:29:15 +00:00
Value * addConstant(const std::string & name, Value & v, Constant info);
2020-03-02 17:15:06 +00:00
void addConstant(const std::string & name, Value * v, Constant info);
Value * addPrimOp(PrimOp && primOp);
public:
Value & getBuiltin(const std::string & name);
2020-08-25 11:31:11 +00:00
struct Doc
{
Pos pos;
std::optional<std::string> name;
2020-08-25 11:31:11 +00:00
size_t arity;
std::vector<std::string> args;
/**
* Unlike the other `doc` fields in this file, this one should never be
* `null`.
*/
2020-08-25 11:31:11 +00:00
const char * doc;
};
std::optional<Doc> getDoc(Value & v);
private:
2013-10-08 12:24:53 +00:00
inline Value * lookupVar(Env * env, const ExprVar & var, bool noEval);
2013-09-02 14:29:15 +00:00
2014-01-21 17:29:55 +00:00
friend struct ExprVar;
friend struct ExprAttrs;
friend struct ExprLet;
Expr * parse(
char * text,
size_t length,
Pos::Origin origin,
const SourcePath & basePath,
std::shared_ptr<StaticEnv> & staticEnv);
/**
* Current Nix call stack depth, used with `max-call-depth` setting to throw stack overflow hopefully before we run out of system stack.
*/
size_t callDepth = 0;
public:
2013-09-02 14:29:15 +00:00
/**
* Do a deep equality test between two values. That is, list
* elements and attributes are compared recursively.
*/
bool eqValues(Value & v1, Value & v2, const PosIdx pos, std::string_view errorCtx);
2015-09-06 23:03:23 +00:00
bool isFunctor(Value & fun);
// FIXME: use std::span
void callFunction(Value & fun, size_t nrArgs, Value * * args, Value & vRes, const PosIdx pos);
void callFunction(Value & fun, Value & arg, Value & vRes, const PosIdx pos)
{
Value * args[] = {&arg};
callFunction(fun, 1, args, vRes, pos);
}
/**
* Automatically call a function for which each argument has a
* default value or has a binding in the `args` map.
*/
void autoCallFunction(Bindings & args, Value & fun, Value & res);
2013-09-02 14:29:15 +00:00
/**
* Allocation primitives.
*/
inline Value * allocValue();
inline Env & allocEnv(size_t size);
2010-03-30 14:39:27 +00:00
Value * allocAttr(Value & vAttrs, Symbol name);
Value * allocAttr(Value & vAttrs, std::string_view name);
2018-05-02 11:56:34 +00:00
Bindings * allocBindings(size_t capacity);
BindingsBuilder buildBindings(size_t capacity)
{
return BindingsBuilder(*this, allocBindings(capacity));
}
2018-05-02 11:56:34 +00:00
void mkList(Value & v, size_t length);
void mkThunk_(Value & v, Expr & expr);
void mkPos(Value & v, PosIdx pos);
/**
* Create a string representing a store path.
*
* The string is the printed store path with a context containing a
* single `NixStringContextElem::Opaque` element of that store path.
*/
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
void mkStorePathString(const StorePath & storePath, Value & v);
/**
* Create a string representing a `SingleDerivedPath::Built`.
*
* The string is the printed store path with a context containing a
* single `NixStringContextElem::Built` element of the drv path and
* output name.
*
* @param value Value we are settings
*
* @param b the drv whose output we are making a string for, and the
* output
*
* @param optStaticOutputPath Optional output path for that string.
* Must be passed if and only if output store object is
* input-addressed or fixed output. Will be printed to form string
* if passed, otherwise a placeholder will be used (see
* `DownstreamPlaceholder`).
*
* @param xpSettings Stop-gap to avoid globals during unit tests.
*/
void mkOutputString(
Value & value,
const SingleDerivedPath::Built & b,
std::optional<StorePath> optStaticOutputPath,
const ExperimentalFeatureSettings & xpSettings = experimentalFeatureSettings);
/**
* Create a string representing a `SingleDerivedPath`.
*
* A combination of `mkStorePathString` and `mkOutputString`.
*/
void mkSingleDerivedPathString(
const SingleDerivedPath & p,
Value & v);
void concatLists(Value & v, size_t nrLists, Value * * lists, const PosIdx pos, std::string_view errorCtx);
/**
* Print statistics, if enabled.
*
* Performs a full memory GC before printing the statistics, so that the
* GC statistics are more accurate.
*/
void maybePrintStats();
/**
* Print statistics, unconditionally, cheaply, without performing a GC first.
*/
void printStatistics();
/**
* Perform a full memory garbage collection - not incremental.
*
* @return true if Nix was built with GC and a GC was performed, false if not.
* The return value is currently not thread safe - just the return value.
*/
bool fullGC();
/**
* Realise the given context, and return a mapping from the placeholders
* used to construct the associated value to their final store path
*/
Use `std::set<StringContextElem>` not `PathSet` for string contexts Motivation `PathSet` is not correct because string contexts have other forms (`Built` and `DrvDeep`) that are not rendered as plain store paths. Instead of wrongly using `PathSet`, or "stringly typed" using `StringSet`, use `std::std<StringContextElem>`. ----- In support of this change, `NixStringContext` is now defined as `std::std<StringContextElem>` not `std:vector<StringContextElem>`. The old definition was just used by a `getContext` method which was only used by the eval cache. It can be deleted altogether since the types are now unified and the preexisting `copyContext` function already suffices. Summarizing the previous paragraph: Old: - `value/context.hh`: `NixStringContext = std::vector<StringContextElem>` - `value.hh`: `NixStringContext Value::getContext(...)` - `value.hh`: `copyContext(...)` New: - `value/context.hh`: `NixStringContext = std::set<StringContextElem>` - `value.hh`: `copyContext(...)` ---- The string representation of string context elements no longer contains the store dir. The diff of `src/libexpr/tests/value/context.cc` should make clear what the new representation is, so we recommend reviewing that file first. This was done for two reasons: Less API churn: `Value::mkString` and friends did not take a `Store` before. But if `NixStringContextElem::{parse, to_string}` *do* take a store (as they did before), then we cannot have the `Value` functions use them (in order to work with the fully-structured `NixStringContext`) without adding that argument. That would have been a lot of churn of threading the store, and this diff is already large enough, so the easier and less invasive thing to do was simply make the element `parse` and `to_string` functions not take the `Store` reference, and the easiest way to do that was to simply drop the store dir. Space usage: Dropping the `/nix/store/` (or similar) from the internal representation will safe space in the heap of the Nix programming being interpreted. If the heap contains many strings with non-trivial contexts, the saving could add up to something significant. ---- The eval cache version is bumped. The eval cache serialization uses `NixStringContextElem::{parse, to_string}`, and since those functions are changed per the above, that means the on-disk representation is also changed. This is simply done by changing the name of the used for the eval cache from `eval-cache-v4` to eval-cache-v5`. ---- To avoid some duplication `EvalCache::mkPathString` is added to abstract over the simple case of turning a store path to a string with just that string in the context. Context This PR picks up where #7543 left off. That one introduced the fully structured `NixStringContextElem` data type, but kept `PathSet context` as an awkward middle ground between internal `char[][]` interpreter heap string contexts and `NixStringContext` fully parsed string contexts. The infelicity of `PathSet context` was specifically called out during Nix team group review, but it was agreeing that fixing it could be left as future work. This is that future work. A possible follow-up step would be to get rid of the `char[][]` evaluator heap representation, too, but it is not yet clear how to do that. To use `NixStringContextElem` there we would need to get the STL containers to GC pointers in the GC build, and I am not sure how to do that. ---- PR #7543 effectively is writing the inverse of a `mkPathString`, `mkOutputString`, and one more such function for the `DrvDeep` case. I would like that PR to have property tests ensuring it is actually the inverse as expected. This PR sets things up nicely so that reworking that PR to be in that more elegant and better tested way is possible. Co-authored-by: Théophane Hufschmitt <7226587+thufschmitt@users.noreply.github.com>
2023-01-29 01:31:10 +00:00
[[nodiscard]] StringMap realiseContext(const NixStringContext & context);
private:
2012-08-13 03:41:48 +00:00
/**
* Like `mkOutputString` but just creates a raw string, not an
* string Value, which would also have a string context.
*/
std::string mkOutputStringRaw(
const SingleDerivedPath::Built & b,
std::optional<StorePath> optStaticOutputPath,
const ExperimentalFeatureSettings & xpSettings = experimentalFeatureSettings);
/**
* Like `mkSingleDerivedPathStringRaw` but just creates a raw string
* Value, which would also have a string context.
*/
std::string mkSingleDerivedPathStringRaw(
const SingleDerivedPath & p);
2015-07-23 21:14:07 +00:00
unsigned long nrEnvs = 0;
unsigned long nrValuesInEnvs = 0;
unsigned long nrValues = 0;
unsigned long nrListElems = 0;
unsigned long nrLookups = 0;
2015-07-23 21:14:07 +00:00
unsigned long nrAttrsets = 0;
unsigned long nrAttrsInAttrsets = 0;
unsigned long nrAvoided = 0;
2015-07-23 21:14:07 +00:00
unsigned long nrOpUpdates = 0;
unsigned long nrOpUpdateValuesCopied = 0;
unsigned long nrListConcats = 0;
unsigned long nrPrimOpCalls = 0;
unsigned long nrFunctionCalls = 0;
bool countCalls;
using PrimOpCalls = std::map<std::string, size_t>;
PrimOpCalls primOpCalls;
using FunctionCalls = std::map<ExprLambda *, size_t>;
FunctionCalls functionCalls;
2013-11-07 17:04:36 +00:00
void incrFunctionCall(ExprLambda * fun);
using AttrSelects = std::map<PosIdx, size_t>;
AttrSelects attrSelects;
2014-01-21 17:29:55 +00:00
friend struct ExprOpUpdate;
friend struct ExprOpConcatLists;
friend struct ExprVar;
friend struct ExprString;
friend struct ExprInt;
friend struct ExprFloat;
friend struct ExprPath;
2014-01-21 17:29:55 +00:00
friend struct ExprSelect;
friend void prim_getAttr(EvalState & state, const PosIdx pos, Value * * args, Value & v);
friend void prim_match(EvalState & state, const PosIdx pos, Value * * args, Value & v);
friend void prim_split(EvalState & state, const PosIdx pos, Value * * args, Value & v);
2022-01-04 19:29:17 +00:00
friend struct Value;
};
struct DebugTraceStacker {
DebugTraceStacker(EvalState & evalState, DebugTrace t);
~DebugTraceStacker()
{
evalState.debugTraces.pop_front();
}
EvalState & evalState;
DebugTrace trace;
2021-12-28 00:35:27 +00:00
};
/**
* @return A string representing the type of the value `v`.
2023-06-21 20:06:16 +00:00
*
* @param withArticle Whether to begin with an english article, e.g. "an
* integer" vs "integer".
*/
2023-06-21 20:06:16 +00:00
std::string_view showType(ValueType type, bool withArticle = true);
std::string showType(const Value & v);
/**
* If `path` refers to a directory, then append "/default.nix".
*/
SourcePath resolveExprPath(SourcePath path);
static constexpr std::string_view corepkgsPrefix{"/__corepkgs__/"};
}
#include "eval-inline.hh"