Nix/src/libexpr/primops.cc

838 lines
26 KiB
C++
Raw Normal View History

#include "build.hh"
#include "misc.hh"
#include "eval.hh"
#include "globals.hh"
#include "store.hh"
#include "util.hh"
#include "expr-to-xml.hh"
#include "nixexpr-ast.hh"
#include <algorithm>
namespace nix {
static Expr primBuiltins(EvalState & state, const ATermVector & args)
{
/* Return an attribute set containing all primops. This allows
Nix expressions to test for new primops and take appropriate
action if they're not available. For instance, rather than
calling a primop `foo' directly, they could say `if builtins ?
foo then builtins.foo ... else ...'. */
ATermMap builtins(128);
for (ATermMap::const_iterator i = state.primOps.begin();
i != state.primOps.end(); ++i)
{
string name = aterm2String(i->key);
if (string(name, 0, 2) == "__")
name = string(name, 2);
/* !!! should use makePrimOp here, I guess. */
builtins.set(toATerm(name), makeAttrRHS(makeVar(i->key), makeNoPos()));
}
return makeAttrs(builtins);
}
/* Load and evaluate an expression from path specified by the
argument. */
static Expr primImport(EvalState & state, const ATermVector & args)
{
ATermList es;
Path path;
Expr arg = evalExpr(state, args[0]), arg2;
if (matchPath(arg, arg2))
path = aterm2String(arg2);
else if (matchStr(arg, arg2)) {
path = aterm2String(arg2);
if (path == "" || path[0] != '/')
throw EvalError("`import' requires an absolute path name");
path = canonPath(path);
}
else if (matchAttrs(arg, es)) {
Expr a = queryAttr(arg, "type");
/* If it is a derivation, we have to realise it and load the
Nix expression created at the derivation's output path. */
if (a && evalString(state, a) == "derivation") {
a = queryAttr(arg, "drvPath");
if (!a) throw EvalError("bad derivation in import");
Path drvPath = evalPath(state, a);
buildDerivations(singleton<PathSet>(drvPath));
a = queryAttr(arg, "outPath");
if (!a) throw EvalError("bad derivation in import");
path = evalPath(state, a);
}
}
else throw TypeError("`import' requires a path or derivation as its argument");
return evalFile(state, path);
}
static void flattenList(EvalState & state, Expr e, ATermList & result)
{
ATermList es;
e = evalExpr(state, e);
if (matchList(e, es))
for (ATermIterator i(es); i; ++i)
flattenList(state, *i, result);
else
result = ATinsert(result, e);
}
ATermList flattenList(EvalState & state, Expr e)
{
ATermList result = ATempty;
flattenList(state, e, result);
return ATreverse(result);
}
void toString(EvalState & state, Expr e,
ATermList & context, string & result)
{
e = evalExpr(state, e);
ATerm s;
ATermList es;
int n;
Expr e2;
while (matchContext(e, es, e2)) {
e = e2;
for (ATermIterator i(es); i; ++i)
context = ATinsert(context, *i);
}
/* Note that `false' is represented as an empty string for shell
scripting convenience, just like `null'. */
if (matchStr(e, s)) result += aterm2String(s);
else if (matchUri(e, s)) result += aterm2String(s);
else if (e == eTrue) result += "1";
else if (e == eFalse) ;
else if (matchInt(e, n)) result += int2String(n);
else if (matchNull(e)) ;
else if (matchAttrs(e, es)) {
Expr a = queryAttr(e, "type");
if (a && evalString(state, a) == "derivation") {
Expr a2 = queryAttr(e, "outPath");
if (!a2) throw EvalError("output path missing");
result += evalPath(state, a2);
context = ATinsert(context, e);
}
else throw TypeError("cannot convert an attribute set to a string");
}
else if (matchPath(e, s)) {
Path path(canonPath(aterm2String(s)));
if (!isStorePath(path)) {
if (isDerivation(path))
throw EvalError(format("file names are not allowed to end in `%1%'")
% drvExtension);
Path dstPath;
if (state.srcToStore[path] != "")
dstPath = state.srcToStore[path];
else {
dstPath = addToStore(path);
state.srcToStore[path] = dstPath;
printMsg(lvlChatty, format("copied source `%1%' -> `%2%'")
% path % dstPath);
}
path = dstPath;
}
result += path;
context = ATinsert(context, makePath(toATerm(path)));
}
else if (matchList(e, es)) {
es = flattenList(state, e);
bool first = true;
for (ATermIterator i(es); i; ++i) {
if (!first) result += " "; else first = false;
toString(state, *i, context, result);
}
}
2006-08-29 17:29:38 +02:00
else throw TypeError(format("cannot convert %1% to a string") % showType(e));
}
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* Returns the hash of a derivation modulo fixed-output
subderivations. A fixed-output derivation is a derivation with one
output (`out') for which an expected hash and hash algorithm are
specified (using the `outputHash' and `outputHashAlgo'
attributes). We don't want changes to such derivations to
propagate upwards through the dependency graph, changing output
paths everywhere.
For instance, if we change the url in a call to the `fetchurl'
function, we do not want to rebuild everything depending on it
(after all, (the hash of) the file being downloaded is unchanged).
So the *output paths* should not change. On the other hand, the
*derivation store expression paths* should change to reflect the
new dependency graph.
That's what this function does: it returns a hash which is just the
of the derivation ATerm, except that any input store expression
paths have been replaced by the result of a recursive call to this
function, and that for fixed-output derivations we return
(basically) its outputHash. */
static Hash hashDerivationModulo(EvalState & state, Derivation drv)
{
/* Return a fixed hash for fixed-output derivations. */
if (drv.outputs.size() == 1) {
DerivationOutputs::const_iterator i = drv.outputs.begin();
if (i->first == "out" &&
i->second.hash != "")
{
return hashString(htSHA256, "fixed:out:"
+ i->second.hashAlgo + ":"
+ i->second.hash + ":"
+ i->second.path);
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
}
}
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* For other derivations, replace the inputs paths with recursive
calls to this function.*/
DerivationInputs inputs2;
for (DerivationInputs::iterator i = drv.inputDrvs.begin();
i != drv.inputDrvs.end(); ++i)
{
Hash h = state.drvHashes[i->first];
if (h.type == htUnknown) {
Derivation drv2 = derivationFromPath(i->first);
h = hashDerivationModulo(state, drv2);
state.drvHashes[i->first] = h;
}
inputs2[printHash(h)] = i->second;
}
drv.inputDrvs = inputs2;
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
return hashTerm(unparseDerivation(drv));
}
/* Construct (as a unobservable side effect) a Nix derivation
expression that performs the derivation described by the argument
set. Returns the original set extended with the following
attributes: `outPath' containing the primary output path of the
derivation; `drvPath' containing the path of the Nix expression;
and `type' set to `derivation' to indicate that this is a
derivation. */
static Expr primDerivationStrict(EvalState & state, const ATermVector & args)
{
startNest(nest, lvlVomit, "evaluating derivation");
2006-05-04 14:21:08 +02:00
ATermMap attrs(128); /* !!! */
queryAllAttrs(evalExpr(state, args[0]), attrs, true);
/* Figure out the name already (for stack backtraces). */
2006-05-04 14:21:08 +02:00
Expr eDrvName = attrs.get(toATerm("name"));
if (!eDrvName)
throw EvalError("required attribute `name' missing");
ATerm posDrvName;
if (!matchAttrRHS(eDrvName, eDrvName, posDrvName)) abort();
string drvName = evalString(state, eDrvName);
/* Build the derivation expression by processing the attributes. */
Derivation drv;
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
string outputHash;
string outputHashAlgo;
bool outputHashRecursive = false;
2006-05-04 14:21:08 +02:00
for (ATermMap::const_iterator i = attrs.begin(); i != attrs.end(); ++i) {
string key = aterm2String(i->key);
ATerm value;
Expr pos;
2006-05-04 14:21:08 +02:00
ATerm rhs = i->value;
if (!matchAttrRHS(rhs, value, pos)) abort();
startNest(nest, lvlVomit, format("processing attribute `%1%'") % key);
try {
ATermList context = ATempty;
/* The `args' attribute is special: it supplies the
command-line arguments to the builder. */
if (key == "args") {
ATermList es;
value = evalExpr(state, value);
if (!matchList(value, es)) {
static bool haveWarned = false;
warnOnce(haveWarned, "the `args' attribute should evaluate to a list");
es = flattenList(state, value);
}
for (ATermIterator i(es); i; ++i) {
string s;
toString(state, *i, context, s);
drv.args.push_back(s);
}
}
/* All other attributes are passed to the builder through
the environment. */
else {
string s;
toString(state, value, context, s);
drv.env[key] = s;
if (key == "builder") drv.builder = s;
else if (key == "system") drv.platform = s;
else if (key == "name") drvName = s;
else if (key == "outputHash") outputHash = s;
else if (key == "outputHashAlgo") outputHashAlgo = s;
else if (key == "outputHashMode") {
if (s == "recursive") outputHashRecursive = true;
else if (s == "flat") outputHashRecursive = false;
else throw EvalError(format("invalid value `%1%' for `outputHashMode' attribute") % s);
}
}
/* Everything in the context of the expression should be
added as dependencies of the resulting derivation. */
for (ATermIterator i(context); i; ++i) {
ATerm s;
ATermList as;
if (matchPath(*i, s)) {
assert(isStorePath(aterm2String(s)));
drv.inputSrcs.insert(aterm2String(s));
}
else if (matchAttrs(*i, as)) {
Expr a = queryAttr(*i, "type");
assert(a && evalString(state, a) == "derivation");
Expr a2 = queryAttr(*i, "drvPath");
if (!a2) throw EvalError("derivation path missing");
drv.inputDrvs[evalPath(state, a2)] = singleton<StringSet>("out");
}
else abort();
}
} catch (Error & e) {
e.addPrefix(format("while processing the derivation attribute `%1%' at %2%:\n")
% key % showPos(pos));
e.addPrefix(format("while instantiating the derivation named `%1%' at %2%:\n")
% drvName % showPos(posDrvName));
throw;
}
}
/* Do we have all required attributes? */
if (drv.builder == "")
throw EvalError("required attribute `builder' missing");
if (drv.platform == "")
throw EvalError("required attribute `system' missing");
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* If an output hash was given, check it. */
if (outputHash == "")
outputHashAlgo = "";
else {
HashType ht = parseHashType(outputHashAlgo);
if (ht == htUnknown)
throw EvalError(format("unknown hash algorithm `%1%'") % outputHashAlgo);
Hash h(ht);
if (outputHash.size() == h.hashSize * 2)
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* hexadecimal representation */
h = parseHash(ht, outputHash);
else if (outputHash.size() == hashLength32(h))
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* base-32 representation */
h = parseHash32(ht, outputHash);
else
throw Error(format("hash `%1%' has wrong length for hash type `%2%'")
% outputHash % outputHashAlgo);
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
string s = outputHash;
outputHash = printHash(h);
if (outputHashRecursive) outputHashAlgo = "r:" + outputHashAlgo;
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
}
2006-09-21 20:52:05 +02:00
/* Check whether the derivation name is valid. */
checkStoreName(drvName);
if (isDerivation(drvName))
throw EvalError(format("derivation names are not allowed to end in `%1%'")
% drvExtension);
/* Construct the "masked" derivation store expression, which is
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
the final one except that in the list of outputs, the output
paths are empty, and the corresponding environment variables
have an empty value. This ensures that changes in the set of
output names do get reflected in the hash. */
drv.env["out"] = "";
drv.outputs["out"] =
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
DerivationOutput("", outputHashAlgo, outputHash);
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* Use the masked derivation expression to compute the output
path. */
Path outPath = makeStorePath("output:out",
hashDerivationModulo(state, drv), drvName);
/* Construct the final derivation store expression. */
drv.env["out"] = outPath;
drv.outputs["out"] =
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
DerivationOutput(outPath, outputHashAlgo, outputHash);
/* Write the resulting term into the Nix store directory. */
Path drvPath = writeDerivation(drv, drvName);
printMsg(lvlChatty, format("instantiated `%1%' -> `%2%'")
% drvName % drvPath);
2005-01-18 12:15:50 +01:00
/* Optimisation, but required in read-only mode! because in that
case we don't actually write store expressions, so we can't
read them later. */
state.drvHashes[drvPath] = hashDerivationModulo(state, drv);
2005-01-18 12:15:50 +01:00
* Removed the `id' attribute hack. * Formalise the notion of fixed-output derivations, i.e., derivations for which a cryptographic hash of the output is known in advance. Changes to such derivations should not propagate upwards through the dependency graph. Previously this was done by specifying the hash component of the output path through the `id' attribute, but this is insecure since you can lie about it (i.e., you can specify any hash and then produce a completely different output). Now the responsibility for checking the output is moved from the builder to Nix itself. A fixed-output derivation can be created by specifying the `outputHash' and `outputHashAlgo' attributes, the latter taking values `md5', `sha1', and `sha256', and the former specifying the actual hash in hexadecimal or in base-32 (auto-detected by looking at the length of the attribute value). MD5 is included for compatibility but should be considered deprecated. * Removed the `drvPath' pseudo-attribute in derivation results. It's no longer necessary. * Cleaned up the support for multiple output paths in derivation store expressions. Each output now has a unique identifier (e.g., `out', `devel', `docs'). Previously there was no way to tell output paths apart at the store expression level. * `nix-hash' now has a flag `--base32' to specify that the hash should be printed in base-32 notation. * `fetchurl' accepts parameters `sha256' and `sha1' in addition to `md5'. * `nix-prefetch-url' now prints out a SHA-1 hash in base-32. (TODO: a flag to specify the hash.)
2005-01-17 17:55:19 +01:00
/* !!! assumes a single output */
2006-05-04 14:21:08 +02:00
ATermMap outAttrs(2);
outAttrs.set(toATerm("outPath"),
makeAttrRHS(makePath(toATerm(outPath)), makeNoPos()));
outAttrs.set(toATerm("drvPath"),
makeAttrRHS(makePath(toATerm(drvPath)), makeNoPos()));
return makeAttrs(outAttrs);
}
static Expr primDerivationLazy(EvalState & state, const ATermVector & args)
{
Expr eAttrs = evalExpr(state, args[0]);
2006-05-04 14:21:08 +02:00
ATermMap attrs(128); /* !!! */
queryAllAttrs(eAttrs, attrs, true);
2006-05-04 14:21:08 +02:00
attrs.set(toATerm("type"),
makeAttrRHS(makeStr(toATerm("derivation")), makeNoPos()));
Expr drvStrict = makeCall(makeVar(toATerm("derivation!")), eAttrs);
2006-05-04 14:21:08 +02:00
attrs.set(toATerm("outPath"),
makeAttrRHS(makeSelect(drvStrict, toATerm("outPath")), makeNoPos()));
attrs.set(toATerm("drvPath"),
makeAttrRHS(makeSelect(drvStrict, toATerm("drvPath")), makeNoPos()));
return makeAttrs(attrs);
}
2003-11-02 17:31:35 +01:00
/* Return the base name of the given string, i.e., everything
following the last slash. */
static Expr primBaseNameOf(EvalState & state, const ATermVector & args)
2003-11-02 17:31:35 +01:00
{
2004-11-03 19:12:03 +01:00
return makeStr(toATerm(baseNameOf(evalString(state, args[0]))));
2003-11-02 17:31:35 +01:00
}
/* Return the directory of the given path, i.e., everything before the
last slash. */
static Expr primDirOf(EvalState & state, const ATermVector & args)
{
return makePath(toATerm(dirOf(evalPath(state, args[0]))));
}
ATerm coerceToString(Expr e)
{
ATerm s;
if (matchStr(e, s) || matchPath(e, s) || matchUri(e, s))
return s;
return 0;
}
/* Convert the argument (which can be a path or a uri) to a string. */
static Expr primToString(EvalState & state, const ATermVector & args)
2003-11-02 17:31:35 +01:00
{
ATermList context = ATempty;
bool dummy;
string s = coerceToStringWithContext(state, context, args[0], dummy);
return wrapInContext(context, makeStr(toATerm(s)));
2003-11-02 17:31:35 +01:00
}
/* Convert the argument (which can be any Nix expression) to an XML
representation returned in a string. Not all Nix expressions can
be sensibly or completely represented (e.g., functions). */
static Expr primToXML(EvalState & state, const ATermVector & args)
{
std::ostringstream out;
printTermAsXML(strictEvalExpr(state, args[0]), out);
return makeStr(toATerm(out.str()));
}
/* Store a string in the Nix store as a source file that can be used
as an input by derivations. */
static Expr primToFile(EvalState & state, const ATermVector & args)
{
string s = evalString(state, args[0]);
Path storePath = addTextToStore("", s, PathSet());
return makePath(toATerm(storePath));
}
/* Boolean constructors. */
static Expr primTrue(EvalState & state, const ATermVector & args)
{
return eTrue;
}
static Expr primFalse(EvalState & state, const ATermVector & args)
{
return eFalse;
}
/* Return the null value. */
static Expr primNull(EvalState & state, const ATermVector & args)
{
return makeNull();
}
/* Determine whether the argument is the null value. */
static Expr primIsNull(EvalState & state, const ATermVector & args)
{
return makeBool(matchNull(evalExpr(state, args[0])));
}
/* Determine whether the argument is a list. */
static Expr primIsList(EvalState & state, const ATermVector & args)
{
ATermList list;
return makeBool(matchList(evalExpr(state, args[0]), list));
}
static Path findDependency(Path dir, string dep)
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
{
if (dep[0] == '/') throw EvalError(
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
format("illegal absolute dependency `%1%'") % dep);
Path p = canonPath(dir + "/" + dep);
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
if (pathExists(p))
return p;
else
return "";
}
/* Make path `p' relative to directory `pivot'. E.g.,
relativise("/a/b/c", "a/b/x/y") => "../x/y". Both input paths
should be in absolute canonical form. */
static string relativise(Path pivot, Path p)
{
assert(pivot.size() > 0 && pivot[0] == '/');
assert(p.size() > 0 && p[0] == '/');
if (pivot == p) return ".";
/* `p' is in `pivot'? */
Path pivot2 = pivot + "/";
if (p.substr(0, pivot2.size()) == pivot2) {
return p.substr(pivot2.size());
}
/* Otherwise, `p' is in a parent of `pivot'. Find up till which
path component `p' and `pivot' match, and add an appropriate
number of `..' components. */
string::size_type i = 1;
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
while (1) {
string::size_type j = pivot.find('/', i);
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
if (j == string::npos) break;
j++;
if (pivot.substr(0, j) != p.substr(0, j)) break;
i = j;
}
string prefix;
unsigned int slashes = count(pivot.begin() + i, pivot.end(), '/') + 1;
while (slashes--) {
prefix += "../";
}
return prefix + p.substr(i);
}
static Expr primDependencyClosure(EvalState & state, const ATermVector & args)
{
startNest(nest, lvlDebug, "finding dependencies");
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
Expr attrs = evalExpr(state, args[0]);
/* Get the start set. */
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
Expr startSet = queryAttr(attrs, "startSet");
if (!startSet) throw EvalError("attribute `startSet' required");
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
ATermList startSet2 = evalList(state, startSet);
Path pivot;
PathSet workSet;
for (ATermIterator i(startSet2); i; ++i) {
Path p = evalPath(state, *i);
workSet.insert(p);
pivot = dirOf(p);
}
/* Get the search path. */
PathSet searchPath;
Expr e = queryAttr(attrs, "searchPath");
if (e) {
ATermList list = evalList(state, e);
for (ATermIterator i(list); i; ++i) {
Path p = evalPath(state, *i);
searchPath.insert(p);
}
}
Expr scanner = queryAttr(attrs, "scanner");
if (!scanner) throw EvalError("attribute `scanner' required");
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
/* Construct the dependency closure by querying the dependency of
each path in `workSet', adding the dependencies to
`workSet'. */
PathSet doneSet;
while (!workSet.empty()) {
Path path = *(workSet.begin());
workSet.erase(path);
if (doneSet.find(path) != doneSet.end()) continue;
doneSet.insert(path);
try {
/* Call the `scanner' function with `path' as argument. */
debug(format("finding dependencies in `%1%'") % path);
ATermList deps = evalList(state, makeCall(scanner, makePath(toATerm(path))));
/* Try to find the dependencies relative to the `path'. */
for (ATermIterator i(deps); i; ++i) {
string s = evalString(state, *i);
Path dep = findDependency(dirOf(path), s);
if (dep == "") {
for (PathSet::iterator j = searchPath.begin();
j != searchPath.end(); ++j)
{
dep = findDependency(*j, s);
if (dep != "") break;
}
}
if (dep == "")
debug(format("did NOT find dependency `%1%'") % s);
else {
debug(format("found dependency `%1%'") % dep);
workSet.insert(dep);
}
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
}
} catch (Error & e) {
e.addPrefix(format("while finding dependencies in `%1%':\n")
% path);
throw;
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
}
}
/* Return a list of the dependencies we've just found. */
ATermList deps = ATempty;
for (PathSet::iterator i = doneSet.begin(); i != doneSet.end(); ++i) {
deps = ATinsert(deps, makeStr(toATerm(relativise(pivot, *i))));
deps = ATinsert(deps, makePath(toATerm(*i)));
}
debug(format("dependency list is `%1%'") % makeList(deps));
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
return makeList(deps);
}
2006-08-23 17:46:00 +02:00
static Expr primAbort(EvalState & state, const ATermVector & args)
{
2006-08-23 17:46:27 +02:00
throw Abort(format("evaluation aborted with the following error message: `%1%'") %
2006-08-23 17:46:00 +02:00
evalString(state, args[0]));
}
/* Return the first element of a list. */
static Expr primHead(EvalState & state, const ATermVector & args)
{
ATermList list = evalList(state, args[0]);
if (ATisEmpty(list))
throw Error("`head' called on an empty list");
return evalExpr(state, ATgetFirst(list));
}
/* Return a list consisting of everything but the the first element of
a list. */
static Expr primTail(EvalState & state, const ATermVector & args)
{
ATermList list = evalList(state, args[0]);
if (ATisEmpty(list))
throw Error("`tail' called on an empty list");
return makeList(ATgetNext(list));
}
/* Return an environment variable. Use with care. */
static Expr primGetEnv(EvalState & state, const ATermVector & args)
{
string name = evalString(state, args[0]);
return makeStr(toATerm(getEnv(name)));
}
/* Apply a function to every element of a list. */
static Expr primMap(EvalState & state, const ATermVector & args)
{
Expr fun = evalExpr(state, args[0]);
ATermList list = evalList(state, args[1]);
ATermList res = ATempty;
for (ATermIterator i(list); i; ++i)
res = ATinsert(res, makeCall(fun, *i));
return makeList(ATreverse(res));
}
/* Return a string constant representing the current platform. Note!
that differs between platforms, so Nix expressions using
`__currentSystem' can evaluate to different values on different
platforms. */
static Expr primCurrentSystem(EvalState & state, const ATermVector & args)
{
return makeStr(toATerm(thisSystem));
}
static Expr primCurrentTime(EvalState & state, const ATermVector & args)
{
return ATmake("Int(<int>)", time(0));
}
/* Dynamic version of the `.' operator. */
static Expr primGetAttr(EvalState & state, const ATermVector & args)
{
string attr = evalString(state, args[0]);
return evalExpr(state, makeSelect(args[1], toATerm(attr)));
}
/* Dynamic version of the `?' operator. */
static Expr primHasAttr(EvalState & state, const ATermVector & args)
{
string attr = evalString(state, args[0]);
return evalExpr(state, makeOpHasAttr(args[1], toATerm(attr)));
}
static Expr primRemoveAttrs(EvalState & state, const ATermVector & args)
{
2006-05-04 14:21:08 +02:00
ATermMap attrs(128); /* !!! */
queryAllAttrs(evalExpr(state, args[0]), attrs, true);
ATermList list = evalList(state, args[1]);
for (ATermIterator i(list); i; ++i)
/* It's not an error for *i not to exist. */
2006-05-04 14:21:08 +02:00
attrs.remove(toATerm(evalString(state, *i)));
return makeAttrs(attrs);
}
static Expr primRelativise(EvalState & state, const ATermVector & args)
{
Path pivot = evalPath(state, args[0]);
Path path = evalPath(state, args[1]);
return makeStr(toATerm(relativise(pivot, path)));
}
static Expr primAdd(EvalState & state, const ATermVector & args)
{
int i1 = evalInt(state, args[0]);
int i2 = evalInt(state, args[1]);
return makeInt(i1 + i2);
}
static Expr primLessThan(EvalState & state, const ATermVector & args)
{
int i1 = evalInt(state, args[0]);
int i2 = evalInt(state, args[1]);
return makeBool(i1 < i2);
}
void EvalState::addPrimOps()
{
addPrimOp("builtins", 0, primBuiltins);
addPrimOp("true", 0, primTrue);
addPrimOp("false", 0, primFalse);
addPrimOp("null", 0, primNull);
addPrimOp("__currentSystem", 0, primCurrentSystem);
addPrimOp("__currentTime", 0, primCurrentTime);
addPrimOp("import", 1, primImport);
addPrimOp("derivation!", 1, primDerivationStrict);
addPrimOp("derivation", 1, primDerivationLazy);
addPrimOp("baseNameOf", 1, primBaseNameOf);
addPrimOp("dirOf", 1, primDirOf);
addPrimOp("toString", 1, primToString);
addPrimOp("__toXML", 1, primToXML);
addPrimOp("__toFile", 1, primToFile);
addPrimOp("isNull", 1, primIsNull);
addPrimOp("__isList", 1, primIsList);
* A primitive operation `dependencyClosure' to do automatic dependency determination (e.g., finding the header files dependencies of a C file) in Nix low-level builds automatically. For instance, in the function `compileC' in make/lib/default.nix, we find the header file dependencies of C file `main' as follows: localIncludes = dependencyClosure { scanner = file: import (findIncludes { inherit file; }); startSet = [main]; }; The function works by "growing" the set of dependencies, starting with the set `startSet', and calling the function `scanner' for each file to get its dependencies (which should yield a list of strings representing relative paths). For instance, when `scanner' is called on a file `foo.c' that includes the line #include "../bar/fnord.h" then `scanner' should yield ["../bar/fnord.h"]. This list of dependencies is absolutised relative to the including file and added to the set of dependencies. The process continues until no more dependencies are found (hence its a closure). `dependencyClosure' yields a list that contains in alternation a dependency, and its relative path to the directory of the start file, e.g., [ /bla/bla/foo.c "foo.c" /bla/bar/fnord.h "../bar/fnord.h" ] These relative paths are necessary for the builder that compiles foo.c to reconstruct the relative directory structure expected by foo.c. The advantage of `dependencyClosure' over the old approach (using the impure `__currentTime') is that it's completely pure, and more efficient because it only rescans for dependencies (i.e., by building the derivations yielded by `scanner') if sources have actually changed. The old approach rescanned every time.
2005-08-14 14:38:47 +02:00
addPrimOp("dependencyClosure", 1, primDependencyClosure);
2006-08-23 17:46:00 +02:00
addPrimOp("abort", 1, primAbort);
addPrimOp("__head", 1, primHead);
addPrimOp("__tail", 1, primTail);
addPrimOp("__getEnv", 1, primGetEnv);
addPrimOp("map", 2, primMap);
addPrimOp("__getAttr", 2, primGetAttr);
addPrimOp("__hasAttr", 2, primHasAttr);
addPrimOp("removeAttrs", 2, primRemoveAttrs);
addPrimOp("relativise", 2, primRelativise);
addPrimOp("__add", 2, primAdd);
addPrimOp("__lessThan", 2, primLessThan);
}
}