lix/src/libstore/worker-protocol.hh

227 lines
6.3 KiB
C++
Raw Normal View History

#pragma once
///@file
#include "serialise.hh"
namespace nix {
#define WORKER_MAGIC_1 0x6e697863
#define WORKER_MAGIC_2 0x6478696f
#define PROTOCOL_VERSION (1 << 8 | 35)
#define GET_PROTOCOL_MAJOR(x) ((x) & 0xff00)
#define GET_PROTOCOL_MINOR(x) ((x) & 0x00ff)
#define STDERR_NEXT 0x6f6c6d67
#define STDERR_READ 0x64617461 // data needed from source
#define STDERR_WRITE 0x64617416 // data for sink
#define STDERR_LAST 0x616c7473
#define STDERR_ERROR 0x63787470
#define STDERR_START_ACTIVITY 0x53545254
#define STDERR_STOP_ACTIVITY 0x53544f50
#define STDERR_RESULT 0x52534c54
class Store;
struct Source;
// items being serialised
struct DerivedPath;
struct DrvOutput;
struct Realisation;
struct BuildResult;
struct KeyedBuildResult;
enum TrustedFlag : bool;
/**
* The "worker protocol", used by unix:// and ssh-ng:// stores.
*
* This `struct` is basically just a `namespace`; We use a type rather
* than a namespace just so we can use it as a template argument.
*/
struct WorkerProto
{
/**
* Enumeration of all the request types for the protocol.
*/
enum struct Op : uint64_t;
/**
* Data type for canonical pairs of serialisers for the worker protocol.
*
* See https://en.cppreference.com/w/cpp/language/adl for the broader
* concept of what is going on here.
*/
template<typename T>
struct Serialise;
// This is the definition of `Serialise` we *want* to put here, but
// do not do so.
//
// The problem is that if we do so, C++ will think we have
// seralisers for *all* types. We don't, of course, but that won't
// cause an error until link time. That makes for long debug cycles
// when there is a missing serialiser.
//
// By not defining it globally, and instead letting individual
// serialisers specialise the type, we get back the compile-time
// errors we would like. When no serialiser exists, C++ sees an
// abstract "incomplete" type with no definition, and any attempt to
// use `to` or `from` static methods is a compile-time error because
// they don't exist on an incomplete type.
//
// This makes for a quicker debug cycle, as desired.
#if 0
{
static T read(const Store & store, Source & from);
static void write(const Store & store, Sink & out, const T & t);
};
#endif
/**
* Wrapper function around `WorkerProto::Serialise<T>::write` that allows us to
* infer the type instead of having to write it down explicitly.
*/
template<typename T>
static void write(const Store & store, Sink & out, const T & t)
{
WorkerProto::Serialise<T>::write(store, out, t);
}
};
enum struct WorkerProto::Op : uint64_t
{
IsValidPath = 1,
HasSubstitutes = 3,
QueryPathHash = 4, // obsolete
QueryReferences = 5, // obsolete
QueryReferrers = 6,
AddToStore = 7,
AddTextToStore = 8, // obsolete since 1.25, Nix 3.0. Use WorkerProto::Op::AddToStore
BuildPaths = 9,
EnsurePath = 10,
AddTempRoot = 11,
AddIndirectRoot = 12,
SyncWithGC = 13,
FindRoots = 14,
ExportPath = 16, // obsolete
QueryDeriver = 18, // obsolete
SetOptions = 19,
CollectGarbage = 20,
QuerySubstitutablePathInfo = 21,
QueryDerivationOutputs = 22, // obsolete
QueryAllValidPaths = 23,
QueryFailedPaths = 24,
ClearFailedPaths = 25,
QueryPathInfo = 26,
ImportPaths = 27, // obsolete
QueryDerivationOutputNames = 28, // obsolete
QueryPathFromHashPart = 29,
QuerySubstitutablePathInfos = 30,
QueryValidPaths = 31,
QuerySubstitutablePaths = 32,
QueryValidDerivers = 33,
OptimiseStore = 34,
VerifyStore = 35,
BuildDerivation = 36,
AddSignatures = 37,
NarFromPath = 38,
AddToStoreNar = 39,
QueryMissing = 40,
QueryDerivationOutputMap = 41,
RegisterDrvOutput = 42,
QueryRealisation = 43,
AddMultipleToStore = 44,
AddBuildLog = 45,
BuildPathsWithResults = 46,
};
/**
* Convenience for sending operation codes.
*
* @todo Switch to using `WorkerProto::Serialise` instead probably. But
* this was not done at this time so there would be less churn.
*/
inline Sink & operator << (Sink & sink, WorkerProto::Op op)
{
return sink << (uint64_t) op;
}
/**
* Convenience for debugging.
*
* @todo Perhaps render known opcodes more nicely.
*/
inline std::ostream & operator << (std::ostream & s, WorkerProto::Op op)
{
return s << (uint64_t) op;
}
/**
* Declare a canonical serialiser pair for the worker protocol.
*
* We specialise the struct merely to indicate that we are implementing
* the function for the given type.
*
* Some sort of `template<...>` must be used with the caller for this to
* be legal specialization syntax. See below for what that looks like in
* practice.
*/
#define MAKE_WORKER_PROTO(T) \
struct WorkerProto::Serialise< T > { \
static T read(const Store & store, Source & from); \
static void write(const Store & store, Sink & out, const T & t); \
};
template<>
MAKE_WORKER_PROTO(std::string);
template<>
MAKE_WORKER_PROTO(StorePath);
template<>
MAKE_WORKER_PROTO(ContentAddress);
template<>
MAKE_WORKER_PROTO(DerivedPath);
template<>
MAKE_WORKER_PROTO(Realisation);
template<>
MAKE_WORKER_PROTO(DrvOutput);
template<>
MAKE_WORKER_PROTO(BuildResult);
template<>
MAKE_WORKER_PROTO(KeyedBuildResult);
template<>
MAKE_WORKER_PROTO(std::optional<TrustedFlag>);
template<typename T>
MAKE_WORKER_PROTO(std::vector<T>);
template<typename T>
MAKE_WORKER_PROTO(std::set<T>);
template<typename K, typename V>
#define X_ std::map<K, V>
MAKE_WORKER_PROTO(X_);
#undef X_
/**
* These use the empty string for the null case, relying on the fact
* that the underlying types never serialise to the empty string.
*
* We do this instead of a generic std::optional<T> instance because
* ordinal tags (0 or 1, here) are a bit of a compatability hazard. For
* the same reason, we don't have a std::variant<T..> instances (ordinal
* tags 0...n).
*
* We could the generic instances and then these as specializations for
* compatability, but that's proven a bit finnicky, and also makes the
* worker protocol harder to implement in other languages where such
* specializations may not be allowed.
*/
template<>
MAKE_WORKER_PROTO(std::optional<StorePath>);
template<>
MAKE_WORKER_PROTO(std::optional<ContentAddress>);
}