nix-super/doc/manual/writing-nix-expressions.xml

<chapter xmlns="http://docbook.org/ns/docbook"
         xmlns:xlink="http://www.w3.org/1999/xlink"
         xml:id='chap-writing-nix-expressions'
         xmlns:xi="http://www.w3.org/2001/XInclude">

<title>Writing Nix Expressions</title>


<para>This chapter shows you how to write Nix expressions, which are
the things that tell Nix how to build packages.  It starts with a
simple example (a Nix expression for GNU Hello), and then moves
on to a more in-depth look at the Nix expression language.</para>


<section><title>A simple Nix expression</title>

<para>This section shows how to add and test the <link
xlink:href='http://www.gnu.org/software/hello/hello.html'>GNU Hello
package</link> to the Nix Packages collection.  Hello is a program
that prints out the text <quote>Hello, world!</quote>.</para>

<para>To add a package to the Nix Packages collection, you generally
need to do three things:

<orderedlist>

  <listitem><para>Write a Nix expression for the package.  This is a
  file that describes all the inputs involved in building the package,
  such as dependencies, sources, and so on.</para></listitem>

  <listitem><para>Write a <emphasis>builder</emphasis>.  This is a
  shell script<footnote><para>In fact, it can be written in any
  language, but typically it's a <command>bash</command> shell
  script.</para></footnote> that actually builds the package from
  the inputs.</para></listitem>

  <listitem><para>Add the package to the file
  <filename>pkgs/top-level/all-packages.nix</filename>.  The Nix
  expression written in the first step is a
  <emphasis>function</emphasis>; it requires other packages in order
  to build it.  In this step you put it all together, i.e., you call
  the function with the right arguments to build the actual
  package.</para></listitem>

</orderedlist>

</para>


<section><title>The Nix expression</title>

<example xml:id='ex-hello-nix'><title>Nix expression for GNU Hello
(<filename>default.nix</filename>)</title>
<programlisting>
{stdenv, fetchurl, perl}: <co xml:id='ex-hello-nix-co-1' />

stdenv.mkDerivation { <co xml:id='ex-hello-nix-co-2' />
  name = "hello-2.1.1"; <co xml:id='ex-hello-nix-co-3' />
  builder = ./builder.sh; <co xml:id='ex-hello-nix-co-4' />
  src = fetchurl { <co xml:id='ex-hello-nix-co-5' />
    url = ftp://ftp.nluug.nl/pub/gnu/hello/hello-2.1.1.tar.gz;
    md5 = "70c9ccf9fac07f762c24f2df2290784d";
  };
  inherit perl; <co xml:id='ex-hello-nix-co-6' />
}</programlisting>
</example>

<para><xref linkend='ex-hello-nix' /> shows a Nix expression for GNU
Hello.  It's actually already in the Nix Packages collection in
<filename>pkgs/applications/misc/hello/ex-1/default.nix</filename>.
It is customary to place each package in a separate directory and call
the single Nix expression in that directory
<filename>default.nix</filename>.  The file has the following elements
(referenced from the figure by number):

<calloutlist>

  <callout arearefs='ex-hello-nix-co-1'>

    <para>This states that the expression is a
    <emphasis>function</emphasis> that expects to be called with three
    arguments: <varname>stdenv</varname>, <varname>fetchurl</varname>,
    and <varname>perl</varname>.  They are needed to build Hello, but
    we don't know how to build them here; that's why they are function
    arguments.  <varname>stdenv</varname> is a package that is used
    by almost all Nix Packages packages; it provides a
    <quote>standard</quote> environment consisting of the things you
    would expect in a basic Unix environment: a C/C++ compiler (GCC,
    to be precise), the Bash shell, fundamental Unix tools such as
    <command>cp</command>, <command>grep</command>,
    <command>tar</command>, etc.  <varname>fetchurl</varname> is a
    function that downloads files.  <varname>perl</varname> is the
    Perl interpreter.</para>

    <para>Nix functions generally have the form <literal>{x, y, ...,
    z}: e</literal> where <varname>x</varname>, <varname>y</varname>,
    etc. are the names of the expected arguments, and where
    <replaceable>e</replaceable> is the body of the function.  So
    here, the entire remainder of the file is the body of the
    function; when given the required arguments, the body should
    describe how to build an instance of the Hello package.</para>
    
  </callout>

  <callout arearefs='ex-hello-nix-co-2'>

    <para>So we have to build a package.  Building something from
    other stuff is called a <emphasis>derivation</emphasis> in Nix (as
    opposed to sources, which are built by humans instead of
    computers).  We perform a derivation by calling
    <varname>stdenv.mkDerivation</varname>.
    <varname>mkDerivation</varname> is a function provided by
    <varname>stdenv</varname> that builds a package from a set of
    <emphasis>attributes</emphasis>.  An attribute set is just a list
    of key/value pairs where each value is an arbitrary Nix
    expression.  They take the general form
    <literal>{<replaceable>name1</replaceable> =
    <replaceable>expr1</replaceable>; <replaceable>...</replaceable>
    <replaceable>nameN</replaceable> =
    <replaceable>exprN</replaceable>;}</literal>.</para>

  </callout>

  <callout arearefs='ex-hello-nix-co-3'>

    <para>The attribute <varname>name</varname> specifies the symbolic
    name and version of the package.  Nix doesn't really care about
    these things, but they are used by for instance <command>nix-env
    -q</command> to show a <quote>human-readable</quote> name for
    packages.  This attribute is required by
    <varname>mkDerivation</varname>.</para>

  </callout>

  <callout arearefs='ex-hello-nix-co-4'>

    <para>The attribute <varname>builder</varname> specifies the
    builder.  This attribute can sometimes be omitted, in which case
    <varname>mkDerivation</varname> will fill in a default builder
    (which does a <literal>configure; make; make install</literal>, in
    essence).  Hello is sufficiently simple that the default builder
    would suffice, but in this case, we will show an actual builder
    for educational purposes.  The value
    <command>./builder.sh</command> refers to the shell script shown
    in <xref linkend='ex-hello-builder' />, discussed below.</para>

  </callout>

  <callout arearefs='ex-hello-nix-co-5'>

    <para>The builder has to know what the sources of the package
    are.  Here, the attribute <varname>src</varname> is bound to the
    result of a call to the <command>fetchurl</command> function.
    Given a URL and an MD5 hash of the expected contents of the file
    at that URL, this function builds a derivation that downloads the
    file and checks its hash.  So the sources are a dependency that
    like all other dependencies is built before Hello itself is
    built.</para>

    <para>Instead of <varname>src</varname> any other name could have
    been used, and in fact there can be any number of sources (bound
    to different attributes).  However, <varname>src</varname> is
    customary, and it's also expected by the default builder (which we
    don't use in this example).</para>

  </callout>

  <callout arearefs='ex-hello-nix-co-6'>

    <para>Since the derivation requires Perl, we have to pass the
    value of the <varname>perl</varname> function argument to the
    builder.  All attributes in the set are actually passed as
    environment variables to the builder, so declaring an attribute

    <programlisting>
perl = perl;</programlisting>

    will do the trick: it binds an attribute <varname>perl</varname>
    to the function argument which also happens to be called
    <varname>perl</varname>.  However, it looks a bit silly, so there
    is a shorter syntax.  The <literal>inherit</literal> keyword
    causes the specified attributes to be bound to whatever variables
    with the same name happen to be in scope.</para>

  </callout>
  
</calloutlist>

</para>

</section>


<section><title>The builder</title>

<example xml:id='ex-hello-builder'><title>Build script for GNU Hello
(<filename>builder.sh</filename>)</title>
<programlisting>
source $stdenv/setup <co xml:id='ex-hello-builder-co-1' />

PATH=$perl/bin:$PATH <co xml:id='ex-hello-builder-co-2' />

tar xvfz $src <co xml:id='ex-hello-builder-co-3' />
cd hello-*
./configure --prefix=$out <co xml:id='ex-hello-builder-co-4' />
make <co xml:id='ex-hello-builder-co-5' />
make install</programlisting>
</example>

<para><xref linkend='ex-hello-builder' /> shows the builder referenced
from Hello's Nix expression (stored in
<filename>pkgs/applications/misc/hello/ex-1/builder.sh</filename>).
The builder can actually be made a lot shorter by using the
<emphasis>generic builder</emphasis> functions provided by
<varname>stdenv</varname>, but here we write out the build steps to
elucidate what a builder does.  It performs the following
steps:</para>

<calloutlist>

  <callout arearefs='ex-hello-builder-co-1'>

    <para>When Nix runs a builder, it initially completely clears the
    environment (except for the attributes declared in the
    derivation).  For instance, the <envar>PATH</envar> variable is
    empty<footnote><para>Actually, it's initialised to
    <filename>/path-not-set</filename> to prevent Bash from setting it
    to a default value.</para></footnote>.  This is done to prevent
    undeclared inputs from being used in the build process.  If for
    example the <envar>PATH</envar> contained
    <filename>/usr/bin</filename>, then you might accidentally use
    <filename>/usr/bin/gcc</filename>.</para>

    <para>So the first step is to set up the environment.  This is
    done by calling the <filename>setup</filename> script of the
    standard environment.  The environment variable
    <envar>stdenv</envar> points to the location of the standard
    environment being used.  (It wasn't specified explicitly as an
    attribute in <xref linkend='ex-hello-nix' />, but
    <varname>mkDerivation</varname> adds it automatically.)</para>

  </callout>

  <callout arearefs='ex-hello-builder-co-2'>

    <para>Since Hello needs Perl, we have to make sure that Perl is in
    the <envar>PATH</envar>.  The <envar>perl</envar> environment
    variable points to the location of the Perl package (since it
    was passed in as an attribute to the derivation), so
    <filename><replaceable>$perl</replaceable>/bin</filename> is the
    directory containing the Perl interpreter.</para>

  </callout>

  <callout arearefs='ex-hello-builder-co-3'>

    <para>Now we have to unpack the sources.  The
    <varname>src</varname> attribute was bound to the result of
    fetching the Hello source tarball from the network, so the
    <envar>src</envar> environment variable points to the location in
    the Nix store to which the tarball was downloaded.  After
    unpacking, we <command>cd</command> to the resulting source
    directory.</para>

    <para>The whole build is performed in a temporary directory
    created in <varname>/tmp</varname>, by the way.  This directory is
    removed after the builder finishes, so there is no need to clean
    up the sources afterwards.  Also, the temporary directory is
    always newly created, so you don't have to worry about files from
    previous builds interfering with the current build.</para>

  </callout>

  <callout arearefs='ex-hello-builder-co-4'>

    <para>GNU Hello is a typical Autoconf-based package, so we first
    have to run its <filename>configure</filename> script.  In Nix
    every package is stored in a separate location in the Nix store,
    for instance
    <filename>/nix/store/9a54ba97fb71b65fda531012d0443ce2-hello-2.1.1</filename>.
    Nix computes this path by cryptographically hashing all attributes
    of the derivation.  The path is passed to the builder through the
    <envar>out</envar> environment variable.  So here we give
    <filename>configure</filename> the parameter
    <literal>--prefix=$out</literal> to cause Hello to be installed in
    the expected location.</para>

  </callout>

  <callout arearefs='ex-hello-builder-co-5'>

    <para>Finally we build Hello (<literal>make</literal>) and install
    it into the location specified by <envar>out</envar>
    (<literal>make install</literal>).</para>

  </callout>
  
</calloutlist>

<para>If you are wondering about the absence of error checking on the
result of various commands called in the builder: this is because the
shell script is evaluated with Bash's <option>-e</option> option,
which causes the script to be aborted if any command fails without an
error check.</para>

</section>


<section><title>Composition</title>

<example xml:id='ex-hello-composition'><title>Composing GNU Hello
(<filename>all-packages.nix</filename>)</title>
<programlisting>
...

rec { <co xml:id='ex-hello-composition-co-1' />
  
  hello = (import ../applications/misc/hello/ex-1 <co xml:id='ex-hello-composition-co-2' />) { <co xml:id='ex-hello-composition-co-3' />
    inherit fetchurl stdenv perl;
  };

  perl = (import ../development/interpreters/perl) { <co xml:id='ex-hello-composition-co-4' />
    inherit fetchurl stdenv;
  };

  fetchurl = (import ../build-support/fetchurl) { 
    inherit stdenv; ...
  };
  
  stdenv = ...;

}
</programlisting>
</example>

<para>The Nix expression in <xref linkend='ex-hello-nix' /> is a
function; it is missing some arguments that have to be filled in
somewhere.  In the Nix Packages collection this is done in the file
<filename>pkgs/top-level/all-packages.nix</filename>, where all
Nix expressions for packages are imported and called with the
appropriate arguments.  <xref linkend='ex-hello-composition' /> shows
some fragments of
<filename>all-packages.nix</filename>.</para>

<calloutlist>

  <callout arearefs='ex-hello-composition-co-1'>

    <para>This file defines a set of attributes, all of which are
    concrete derivations (i.e., not functions).  In fact, we define a
    <emphasis>mutually recursive</emphasis> set of attributes.  That
    is, the attributes can refer to each other.  This is precisely
    what we want since we want to <quote>plug</quote> the
    various packages into each other.</para>

  </callout>

  <callout arearefs='ex-hello-composition-co-2'>

    <para>Here we <emphasis>import</emphasis> the Nix expression for
    GNU Hello.  The import operation just loads and returns the
    specified Nix expression. In fact, we could just have put the
    contents of <xref linkend='ex-hello-nix' /> in
    <filename>all-packages.nix</filename> at this point.  That
    would be completely equivalent, but it would make the file rather
    bulky.</para>

    <para>Note that we refer to
    <filename>../applications/misc/hello/ex-1</filename>, not
    <filename>../applications/misc/hello/ex-1/default.nix</filename>.
    When you try to import a directory, Nix automatically appends
    <filename>/default.nix</filename> to the file name.</para>

  </callout>

  <callout arearefs='ex-hello-composition-co-3'>

    <para>This is where the actual composition takes place.  Here we
    <emphasis>call</emphasis> the function imported from
    <filename>../applications/misc/hello/ex-1</filename> with an
    attribute set containing the things that the function expects,
    namely <varname>fetchurl</varname>, <varname>stdenv</varname>, and
    <varname>perl</varname>.  We use inherit again to use the
    attributes defined in the surrounding scope (we could also have
    written <literal>fetchurl = fetchurl;</literal>, etc.).</para>

    <para>The result of this function call is an actual derivation
    that can be built by Nix (since when we fill in the arguments of
    the function, what we get is its body, which is the call to
    <varname>stdenv.mkDerivation</varname> in <xref
    linkend='ex-hello-nix' />).</para>

  </callout>

  <callout arearefs='ex-hello-composition-co-4'>

    <para>Likewise, we have to instantiate Perl,
    <varname>fetchurl</varname>, and the standard environment.</para>

  </callout>

</calloutlist>

</section>


<section><title>Testing</title>

<para>You can now try to build Hello.  Of course, you could do
<literal>nix-env -f pkgs/top-level/all-packages.nix -i hello</literal>,
but you may not want to install a possibly broken package just yet.
The best way to test the package is by using the command <command
linkend="sec-nix-build">nix-build</command>, which builds a Nix
expression and creates a symlink named <filename>result</filename> in
the current directory:

<screen>
$ nix-build pkgs/top-level/all-packages.nix -A hello
building path `/nix/store/632d2b22514d...-hello-2.1.1'
hello-2.1.1/
hello-2.1.1/intl/
hello-2.1.1/intl/ChangeLog
<replaceable>...</replaceable>

$ ls -l result
lrwxrwxrwx ... 2006-09-29 10:43 result -> /nix/store/632d2b22514d...-hello-2.1.1

$ ./result/bin/hello
Hello, world!</screen>

The <link linkend='opt-attr'><option>-A</option></link> option selects
the <literal>hello</literal> attribute from
<filename>all-packages.nix</filename>.  This is faster than using the
symbolic package name specified by the <literal>name</literal>
attribute (which also happens to be <literal>hello</literal>) and is
unambiguous (there can be multiple packages with the symbolic name
<literal>hello</literal>, but there can be only one attribute in a set
named <literal>hello</literal>).</para>

<para><command>nix-build</command> registers the
<filename>./result</filename> symlink as a garbage collection root, so
unless and until you delete the <filename>./result</filename> symlink,
the output of the build will be safely kept on your system.  You can
use <command>nix-build</command>’s <option
linkend='opt-out-link'>-o</option> switch to give the symlink another
name.</para>

<para>Nix has a transactional semantics.  Once a build finishes
successfully, Nix makes a note of this in its database: it registers
that the path denoted by <envar>out</envar> is now
<quote>valid</quote>.  If you try to build the derivation again, Nix
will see that the path is already valid and finish immediately.  If a
build fails, either because it returns a non-zero exit code, because
Nix or the builder are killed, or because the machine crashes, then
the output path will not be registered as valid.  If you try to build
the derivation again, Nix will remove the output path if it exists
(e.g., because the builder died half-way through <literal>make
install</literal>) and try again.  Note that there is no
<quote>negative caching</quote>: Nix doesn't remember that a build
failed, and so a failed build can always be repeated.  This is because
Nix cannot distinguish between permanent failures (e.g., a compiler
error due to a syntax error in the source) and transient failures
(e.g., a disk full condition).</para>

<para>Nix also performs locking.  If you run multiple Nix builds
simultaneously, and they try to build the same derivation, the first
Nix instance that gets there will perform the build, while the others
block (or perform other derivations if available) until the build
finishes:

<screen>
$ nix-build pkgs/top-level/all-packages.nix -A hello
waiting for lock on `/nix/store/0h5b7hp8d4hqfrw8igvx97x1xawrjnac-hello-2.1.1x'</screen>

So it is always safe to run multiple instances of Nix in parallel
(which isn’t the case with, say, <command>make</command>).</para>

<para>If you have a system with multiple CPUs, you may want to have
Nix build different derivations in parallel (insofar as possible).
Just pass the option <link linkend='opt-max-jobs'><option>-j
<replaceable>N</replaceable></option></link>, where
<replaceable>N</replaceable> is the maximum number of jobs to be run
in parallel, or set.  Typically this should be the number of
CPUs.</para>

</section>


<section><title>The generic builder</title>

<para>Recall from <xref linkend='ex-hello-builder' /> that the builder
looked something like this:

<programlisting>
PATH=$perl/bin:$PATH
tar xvfz $src
cd hello-*
./configure --prefix=$out
make
make install</programlisting>

The builders for almost all Unix packages look like this — set up some
environment variables, unpack the sources, configure, build, and
install.  For this reason the standard environment provides some Bash
functions that automate the build process.  A builder using the
generic build facilities in shown in <xref linkend='ex-hello-builder2'
/>.</para>

<example xml:id='ex-hello-builder2'><title>Build script using the generic
build functions</title>
<programlisting>
buildInputs="$perl" <co xml:id='ex-hello-builder2-co-1' />

source $stdenv/setup <co xml:id='ex-hello-builder2-co-2' />

genericBuild <co xml:id='ex-hello-builder2-co-3' /></programlisting>
</example>

<calloutlist>

  <callout arearefs='ex-hello-builder2-co-1'>
    
    <para>The <envar>buildInputs</envar> variable tells
    <filename>setup</filename> to use the indicated packages as
    <quote>inputs</quote>.  This means that if a package provides a
    <filename>bin</filename> subdirectory, it's added to
    <envar>PATH</envar>; if it has a <filename>include</filename>
    subdirectory, it's added to GCC's header search path; and so
    on.<footnote><para>How does it work? <filename>setup</filename>
    tries to source the file
    <filename><replaceable>pkg</replaceable>/nix-support/setup-hook</filename>
    of all dependencies.  These “setup hooks” can then set up whatever
    environment variables they want; for instance, the setup hook for
    Perl sets the <envar>PERL5LIB</envar> environment variable to
    contain the <filename>lib/site_perl</filename> directories of all
    inputs.</para></footnote>
    </para>

  </callout>

  <callout arearefs='ex-hello-builder2-co-2'>

    <para>The function <function>genericBuild</function> is defined in
    the file <literal>$stdenv/setup</literal>.</para>

  </callout>
  
  <callout arearefs='ex-hello-builder2-co-3'>

    <para>The final step calls the shell function
    <function>genericBuild</function>, which performs the steps that
    were done explicitly in <xref linkend='ex-hello-builder' />.  The
    generic builder is smart enough to figure out whether to unpack
    the sources using <command>gzip</command>,
    <command>bzip2</command>, etc.  It can be customised in many ways;
    see <xref linkend='sec-standard-environment' />.</para>

  </callout>
  
</calloutlist>

<para>Discerning readers will note that the
<envar>buildInputs</envar> could just as well have been set in the Nix
expression, like this:

<programlisting>
  buildInputs = [perl];</programlisting>

The <varname>perl</varname> attribute can then be removed, and the
builder becomes even shorter:

<programlisting>
source $stdenv/setup
genericBuild</programlisting>

In fact, <varname>mkDerivation</varname> provides a default builder
that looks exactly like that, so it is actually possible to omit the
builder for Hello entirely.</para>

</section>


</section>



<section><title>The Nix expression language</title>

<para>The Nix expression language is a pure, lazy, functional
language.  Purity means that operations in the language don't have
side-effects (for instance, there is no variable assignment).
Laziness means that arguments to functions are evaluated only when
they are needed.  Functional means that functions are
<quote>normal</quote> values that can be passed around and manipulated
in interesting ways.  The language is not a full-featured, general
purpose language.  It's main job is to describe packages,
compositions of packages, and the variability within
packages.</para>

<para>This section presents the various features of the
language.</para>


<section xml:id='ssec-values'><title>Values</title>


<simplesect><title>Simple values</title>

<para>Nix has the following basic data types:

<itemizedlist>

  <listitem>

    <para><emphasis>Strings</emphasis> can be written in three
    ways.</para>

    <para>The most common way is to enclose the string between double
    quotes, e.g., <literal>"foo bar"</literal>.  Strings can span
    multiple lines.  The special characters <literal>"</literal> and
    <literal>\</literal> and the character sequence
    <literal>${</literal> must be escaped by prefixing them with a
    backslash (<literal>\</literal>).  Newlines, carriage returns and
    tabs can be written as <literal>\n</literal>,
    <literal>\r</literal> and <literal>\t</literal>,
    respectively.</para>

    <para>You can include the result of an expression into a string by
    enclosing it in
    <literal>${<replaceable>...</replaceable>}</literal>, a feature
    known as <emphasis>antiquotation</emphasis>.  The enclosed
    expression must evaluate to something that can be coerced into a
    string (meaning that it must be a string, a path, or a
    derivation).  For instance, rather than writing

<programlisting>
"--with-freetype2-library=" + freetype + "/lib"</programlisting>

    (where <varname>freetype</varname> is a derivation), you can
    instead write the more natural

<programlisting>
"--with-freetype2-library=${freetype}/lib"</programlisting>

    The latter is automatically translated to the former.  A more
    complicated example (from the Nix expression for <link
    xlink:href='http://www.trolltech.com/products/qt'>Qt</link>):

<programlisting>
configureFlags = "
  -system-zlib -system-libpng -system-libjpeg
  ${if openglSupport then "-dlopen-opengl
    -L${mesa}/lib -I${mesa}/include
    -L${libXmu}/lib -I${libXmu}/include" else ""}
  ${if threadSupport then "-thread" else "-no-thread"}
";</programlisting>

    Note that Nix expressions and strings can be arbitrarily nested;
    in this case the outer string contains various antiquotations that
    themselves contain strings (e.g., <literal>"-thread"</literal>),
    some of which in turn contain expressions (e.g.,
    <literal>${mesa}</literal>).</para>

    <para>The second way to write string literals is as an
    <emphasis>indented string</emphasis>, which is enclosed between
    pairs of <emphasis>double single-quotes</emphasis>, like so:

<programlisting>
''
  This is the first line.
  This is the second line.
    This is the third line.
''</programlisting>

    This kind of string literal intelligently strips indentation from
    the start of each line.  To be precise, it strips from each line a
    number of spaces equal to the minimal indentation of the string as
    a whole (disregarding the indentation of empty lines).  For
    instance, the first and second line are indented two space, while
    the third line is indented three spaces.  Thus, two spaces are
    stripped from each line, so the resulting string is

<programlisting>    
"This is the first line.\nThis is the second line.\n  This is the third line.\n"</programlisting>

    </para>

    <para>Note that the whitespace and newline following the opening
    <literal>''</literal> is ignored if there is no non-whitespace
    text on the initial line.</para>

    <para>Antiquotation
    (<literal>${<replaceable>expr</replaceable>}}</literal>) is
    supported in indented strings.</para>

    <para>Since <literal>${</literal> and <literal>''</literal> have
    special meaning in indented strings, you need a way to quote them.
    <literal>${</literal> can be escaped by prefixing it with
    <literal>''</literal>, i.e., <literal>''${</literal>.
    <literal>''</literal> can be escaped by prefixing it with
    <literal>'</literal>, i.e., <literal>'''</literal>.  Finally,
    linefeed, carriage-return and tab characters can be writted as
    <literal>''\n</literal>, <literal>''\r</literal>,
    <literal>''\t</literal>.</para>
    
    <para>Indented strings are primarily useful in that they allow
    multi-line string literals to follow the indentation of the
    enclosing Nix expression, and that less escaping is typically
    necessary for strings representing languages such as shell scripts
    and configuration files because <literal>''</literal> is much less
    common than <literal>"</literal>.  Example:

<programlisting>
stdenv.mkDerivation {
  <replaceable>...</replaceable>
  postInstall =
    ''
      mkdir $out/bin $out/etc
      cp foo $out/bin
      echo "Hello World" > $out/etc/foo.conf
      ${if enableBar then "cp bar $out/bin" else ""}
    '';
  <replaceable>...</replaceable>
}    
</programlisting>

    </para>

    <para>Finally, as a convenience, <emphasis>URIs</emphasis> as
    defined in appendix B of <link
    xlink:href='http://www.ietf.org/rfc/rfc2396.txt'>RFC 2396</link>
    can be written <emphasis>as is</emphasis>, without quotes.  For
    instance, the string
    <literal>"http://example.org/foo.tar.bz2"</literal>
    can also be written as
    <literal>http://example.org/foo.tar.bz2</literal>.</para>

  </listitem>

  <listitem><para><emphasis>Integers</emphasis>, e.g.,
  <literal>123</literal>.</para></listitem>

  <listitem><para><emphasis>Paths</emphasis>, e.g.,
  <filename>/bin/sh</filename> or <filename>./builder.sh</filename>.
  A path must contain at least one slash to be recognised as such; for
  instance, <filename>builder.sh</filename> is not a
  path<footnote><para>It's parsed as an expression that selects the
  attribute <varname>sh</varname> from the variable
  <varname>builder</varname>.</para></footnote>.  If the file name is
  relative, i.e., if it does not begin with a slash, it is made
  absolute at parse time relative to the directory of the Nix
  expression that contained it.  For instance, if a Nix expression in
  <filename>/foo/bar/bla.nix</filename> refers to
  <filename>../xyzzy/fnord.nix</filename>, the absolutised path is
  <filename>/foo/xyzzy/fnord.nix</filename>.</para></listitem>

  <listitem><para><emphasis>Booleans</emphasis> with values
  <literal>true</literal> and
  <literal>false</literal>.</para></listitem>
  
</itemizedlist>

</para>

</simplesect>


<simplesect><title>Lists</title>

<para>Lists are formed by enclosing a whitespace-separated list of
values between square brackets.  For example,

<programlisting>
[ 123 ./foo.nix "abc" (f {x=y;}) ]</programlisting>

defines a list of four elements, the last being the result of a call
to the function <varname>f</varname>.  Note that function calls have
to be enclosed in parentheses.  If they had been omitted, e.g.,

<programlisting>
[ 123 ./foo.nix "abc" f {x=y;} ]</programlisting>

the result would be a list of five elements, the fourth one being a
function and the fifth being an attribute set.</para>

</simplesect>


<simplesect><title>Attribute sets</title>

<para>Attribute sets are really the core of the language, since
ultimately it's all about creating derivations, which are really just
sets of attributes to be passed to build scripts.</para>

<para>Attribute sets are just a list of name/value pairs enclosed in
curly brackets, where each value is an arbitrary expression terminated
by a semicolon.  For example:

<programlisting>
{ x = 123;
  text = "Hello";
  y = f { bla = 456; };
}</programlisting>

This defines an attribute set with attributes named
<varname>x</varname>, <varname>test</varname>, <varname>y</varname>.
The order of the attributes is irrelevant.  An attribute name may only
occur once.</para>

<para>Attributes can be selected from an attribute set using the
<literal>.</literal> operator.  For instance,

<programlisting>
{ a = "Foo"; b = "Bar"; }.a</programlisting>

evaluates to <literal>"Foo"</literal>.</para>

</simplesect>


</section>


<section><title>Language constructs</title>


<simplesect><title>Recursive attribute sets</title>

<para>Recursive attribute sets are just normal attribute sets, but the
attributes can refer to each other.  For example,

<programlisting>
rec {
  x = y;
  y = 123;
}.x
</programlisting>

evaluates to <literal>123</literal>.  Note that without
<literal>rec</literal> the binding <literal>x = y;</literal> would
refer to the variable <varname>y</varname> in the surrounding scope,
if one exists, and would be invalid if no such variable exists.  That
is, in a normal (non-recursive) attribute set, attributes are not
added to the lexical scope; in a recursive set, they are.</para>

<para>Recursive attribute sets of course introduce the danger of
infinite recursion.  For example,

<programlisting>
rec {
  x = y;
  y = x;
}.x</programlisting>

does not terminate<footnote><para>Actually, Nix detects infinite
recursion in this case and aborts (<quote>infinite recursion
encountered</quote>).</para></footnote>.</para>

</simplesect>


<simplesect><title>Let-expressions</title>

<para>A let-expression allows you define local
variables for an expression.  For instance,

<programlisting>
let
  x = "foo";
  y = "bar";
in x + y</programlisting>

evaluates to <literal>"foobar"</literal>.

</para>

<note><para>There is also an obsolete form of let-expression,
<literal>let { <replaceable>attrs</replaceable> }</literal>, which is
translated to <literal>rec { <replaceable>attrs</replaceable>
}.body</literal>.  That is, the body of the let-expression is the
<literal>body</literal> attribute of the attribute set.</para></note>

</simplesect>


<simplesect><title>Inheriting attributes</title>

<para>When defining an attribute set it is often convenient to copy
variables from the surrounding lexical scope (e.g., when you want to
propagate attributes).  This can be shortened using the
<literal>inherit</literal> keyword.  For instance,

<programlisting>
let
  x = 123;
in 
  {
    inherit x;
    y = 456;
  }</programlisting>

evaluates to <literal>{x = 123; y = 456;}</literal>.  (Note that this
works because <varname>x</varname> is added to the lexical scope by
the <literal>let</literal> construct.)  It is also possible to inherit
attributes from another attribute set.  For instance, in this fragment
from <filename>all-packages.nix</filename>,

<programlisting>
  graphviz = (import ../tools/graphics/graphviz) {
    inherit fetchurl stdenv libpng libjpeg expat x11 yacc;
    inherit (xlibs) libXaw;
  };

  xlibs = {
    libX11 = ...;
    libXaw = ...;
    ...
  }

  libpng = ...;
  libjpg = ...;
  ...</programlisting>

the attribute set used in the function call to the function defined in
<filename>../tools/graphics/graphviz</filename> inherits a number of
variables from the surrounding scope (<varname>fetchurl</varname>
... <varname>yacc</varname>), but also inherits
<varname>libXaw</varname> (the X Athena Widgets) from the
<varname>xlibs</varname> (X11 client-side libraries) attribute
set.</para>

</simplesect>


<simplesect xml:id="ss-functions"><title>Functions</title>

<para>Functions have the following form:

<programlisting>
<replaceable>pattern</replaceable>: <replaceable>body</replaceable></programlisting>

The pattern specifies what the argument of the function must look
like, and binds variables in the body to (parts of) the
argument.  There are three kinds of patterns:</para>

<itemizedlist>

  
  <listitem><para>If a pattern is a single identifier, then the
  function matches any argument.  Example:

  <programlisting>
let negate = x: !x;
    concat = x: y: x + y;
in if negate true then concat "foo" "bar" else ""</programlisting>

  Note that <function>concat</function> is a function that takes one
  argument and returns a function that takes another argument.  This
  allows partial parameterisation (i.e., only filling some of the
  arguments of a function); e.g.,

  <programlisting>
map (concat "foo") ["bar" "bla" "abc"]</programlisting>

  evaluates to <literal>["foobar" "foobla"
  "fooabc"]</literal>.</para></listitem>

  
  <listitem><para>An <emphasis>attribute set pattern</emphasis> of the
  form <literal>{name1, name2, …, nameN}</literal>
  matches an attribute set containing the listed attributes, and binds
  the values of those attributes to variables in the function body.
  For example, the function

<programlisting>
{x, y, z}: z + y + x</programlisting>

  can only be called with a set containing exactly the attributes
  <varname>x</varname>, <varname>y</varname> and
  <varname>z</varname>.  No other attributes are allowed.  If you want
  to allow additional arguments, you can use an ellipsis
  (<literal>...</literal>): 

<programlisting>
{x, y, z, ....}: z + y + x</programlisting>

  This works on any set that contains at least the three named
  attributes.</para>

  <para>It is possible to provide <emphasis>default values</emphasis>
  for attributes, in which case they are allowed to be missing.  A
  default value is specified by writing
  <literal><replaceable>name</replaceable> ?
  <replaceable>e</replaceable></literal>, where
  <replaceable>e</replaceable> is an arbitrary expression.  For example,

<programlisting>
{x, y ? "foo", z ? "bar"}: z + y + x</programlisting>

  specifies a function that only requires an attribute named
  <varname>x</varname>, but optionally accepts <varname>y</varname>
  and <varname>z</varname>.</para></listitem>


  <listitem><para>An <literal>@</literal>-pattern requires that the
  argument matches with the patterns on the left- and right-hand side
  of the <literal>@</literal>-sign.  For example:

<programlisting>
args@{x, y, z, ...}: z + y + x + args.a</programlisting>

  Here <varname>args</varname> is bound to the entire argument, which
  is further matches against the pattern <literal>{x, y, z,
  ...}</literal>.</para></listitem>


</itemizedlist>

<para>Note that functions do not have names.  If you want to give them
a name, you can bind them to an attribute, e.g.,

<programlisting>
let concat = {x, y}: x + y;
in concat {x = "foo"; y = "bar";}</programlisting>

</para>

</simplesect>


<simplesect><title>Conditionals</title>

<para>Conditionals look like this:

<programlisting>
if <replaceable>e1</replaceable> then <replaceable>e2</replaceable> else <replaceable>e3</replaceable></programlisting>

where <replaceable>e1</replaceable> is an expression that should
evaluate to a Boolean value (<literal>true</literal> or
<literal>false</literal>).</para>

</simplesect>


<simplesect><title>Assertions</title>

<para>Assertions are generally used to check that certain requirements
on or between features and dependencies hold.  They look like this:

<programlisting>
assert <replaceable>e1</replaceable>; <replaceable>e2</replaceable></programlisting>

where <replaceable>e1</replaceable> is an expression that should
evaluate to a Boolean value.  If it evaluates to
<literal>true</literal>, <replaceable>e2</replaceable> is returned;
otherwise expression evaluation is aborted and a backtrace is printed.</para>

<example xml:id='ex-subversion-nix'><title>Nix expression for Subversion</title>
<programlisting>
{ localServer ? false
, httpServer ? false
, sslSupport ? false
, pythonBindings ? false
, javaSwigBindings ? false
, javahlBindings ? false
, stdenv, fetchurl
, openssl ? null, httpd ? null, db4 ? null, expat, swig ? null, j2sdk ? null
}:

assert localServer -> db4 != null; <co xml:id='ex-subversion-nix-co-1' />
assert httpServer -> httpd != null &amp;&amp; httpd.expat == expat; <co xml:id='ex-subversion-nix-co-2' />
assert sslSupport -> openssl != null &amp;&amp; (httpServer -> httpd.openssl == openssl); <co xml:id='ex-subversion-nix-co-3' />
assert pythonBindings -> swig != null &amp;&amp; swig.pythonSupport;
assert javaSwigBindings -> swig != null &amp;&amp; swig.javaSupport;
assert javahlBindings -> j2sdk != null;

stdenv.mkDerivation {
  name = "subversion-1.1.1";
  ...
  openssl = if sslSupport then openssl else null; <co xml:id='ex-subversion-nix-co-4' />
  ...
}</programlisting>
</example>

<para><xref linkend='ex-subversion-nix' /> show how assertions are
used in the Nix expression for Subversion.</para>

<calloutlist>

  <callout arearefs='ex-subversion-nix-co-1'>
    <para>This assertion states that if Subversion is to have support
    for local repositories, then Berkeley DB is needed.  So if the
    Subversion function is called with the
    <varname>localServer</varname> argument set to
    <literal>true</literal> but the <varname>db4</varname> argument
    set to <literal>null</literal>, then the evaluation fails.</para>
  </callout>

  <callout arearefs='ex-subversion-nix-co-2'>
    <para>This is a more subtle condition: if Subversion is built with
    Apache (<literal>httpServer</literal>) support, then the Expat
    library (an XML library) used by Subversion should be same as the
    one used by Apache.  This is because in this configuration
    Subversion code ends up being linked with Apache code, and if the
    Expat libraries do not match, a build- or runtime link error or
    incompatibility might occur.</para>
  </callout>

  <callout arearefs='ex-subversion-nix-co-2'>
    <para>This assertion says that in order for Subversion to have SSL
    support (so that it can access <literal>https</literal> URLs), an
    OpenSSL library must be passed.  Additionally, it says that
    <emphasis>if</emphasis> Apache support is enabled, then Apache's
    OpenSSL should match Subversion's.  (Note that if Apache support
    is not enabled, we don't care about Apache's OpenSSL.)</para>
  </callout>

  <callout arearefs='ex-subversion-nix-co-4'>
    <para>The conditional here is not really related to assertions,
    but is worth pointing out: it ensures that if SSL support is
    disabled, then the Subversion derivation is not dependent on
    OpenSSL, even if a non-<literal>null</literal> value was passed.
    This prevents an unnecessary rebuild of Subversion if OpenSSL
    changes.</para>
  </callout>

</calloutlist>

</simplesect>
  


<simplesect><title>With-expressions</title>

<para>A <emphasis>with-expression</emphasis>,

<programlisting>
with <replaceable>e1</replaceable>; <replaceable>e2</replaceable></programlisting>

introduces the attribute set <replaceable>e1</replaceable> into the
lexical scope of the expression <replaceable>e2</replaceable>.  For
instance,

<programlisting>
let as = {x = "foo"; y = "bar";};
in with as; x + y</programlisting>

evaluates to <literal>"foobar"</literal> since the
<literal>with</literal> adds the <varname>x</varname> and
<varname>y</varname> attributes of <varname>as</varname> to the
lexical scope in the expression <literal>x + y</literal>.  The most
common use of <literal>with</literal> is in conjunction with the
<function>import</function> function.  E.g.,

<programlisting>
with (import ./definitions.nix); ...</programlisting>

makes all attributes defined in the file
<filename>definitions.nix</filename> available as if they were defined
locally in a <literal>rec</literal>-expression.</para>

</simplesect>


<simplesect><title>Comments</title>

<para>Comments can be single-line, started with a <literal>#</literal>
character, or inline/multi-line, enclosed within <literal>/*
... */</literal>.</para>

</simplesect>


</section>


<section><title>Operators</title>

<para><xref linkend='table-operators' /> lists the operators in the
Nix expression language, in order of precedence (from strongest to
weakest binding).</para>

<table xml:id='table-operators'>
  <title>Operators</title>
  <tgroup cols='3'>
    <thead>
      <row>
        <entry>Syntax</entry>
        <entry>Associativity</entry>
        <entry>Description</entry>
      </row>
    </thead>
    <tbody>
      <row>
        <entry><replaceable>e</replaceable> .
        <replaceable>id</replaceable></entry>
        <entry>none</entry>
        <entry>Select attribute named <replaceable>id</replaceable>
        from attribute set <replaceable>e</replaceable>.  Abort
        evaluation if the attribute doesn’t exist.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> <replaceable>e2</replaceable></entry>
        <entry>left</entry>
        <entry>Call function <replaceable>e1</replaceable> with
        argument <replaceable>e2</replaceable>.</entry>
      </row>
      <row>
        <entry><replaceable>e</replaceable> ?
        <replaceable>id</replaceable></entry>
        <entry>none</entry>
        <entry>Test whether attribute set <replaceable>e</replaceable>
        contains an attribute named <replaceable>id</replaceable>;
        return <literal>true</literal> or
        <literal>false</literal>.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> ++ <replaceable>e2</replaceable></entry>
        <entry>right</entry>
        <entry>List concatenation.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> + <replaceable>e2</replaceable></entry>
        <entry>left</entry>
        <entry>String or path concatenation.</entry>
      </row>
      <row>
        <entry>! <replaceable>e</replaceable></entry>
        <entry>left</entry>
        <entry>Boolean negation.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> //
        <replaceable>e2</replaceable></entry>
        <entry>right</entry>
        <entry>Return an attribute set consisting of the attributes in
        <replaceable>e1</replaceable> and
        <replaceable>e2</replaceable> (with the latter taking
        precedence over the former in case of equally named attributes).</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> ==
        <replaceable>e2</replaceable></entry>
        <entry>none</entry>
        <entry>Equality.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> !=
        <replaceable>e2</replaceable></entry>
        <entry>none</entry>
        <entry>Inequality.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> &amp;&amp;
        <replaceable>e2</replaceable></entry>
        <entry>left</entry>
        <entry>Logical AND.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> ||
        <replaceable>e2</replaceable></entry>
        <entry>left</entry>
        <entry>Logical OR.</entry>
      </row>
      <row>
        <entry><replaceable>e1</replaceable> ->
        <replaceable>e2</replaceable></entry>
        <entry>none</entry>
        <entry>Logical implication (equivalent to
        <literal>!<replaceable>e1</replaceable> ||
        <replaceable>e2</replaceable></literal>).</entry>
      </row>
    </tbody>
  </tgroup>
</table>

</section>


<section xml:id="ssec-derivation"><title>Derivations</title>

<para>The most important built-in function is
<function>derivation</function>, which is used to describe a
single derivation (a build action).  It takes as input an attribute
set, the attributes of which specify the inputs of the build.</para>

<itemizedlist>

  <listitem xml:id="attr-system"><para>There must be an attribute named
  <varname>system</varname> whose value must be a string specifying a
  Nix platform identifier, such as <literal>"i686-linux"</literal> or
  <literal>"powerpc-darwin"</literal><footnote><para>To figure out
  your platform identifier, look at the line <quote>Checking for the
  canonical Nix system name</quote> in the output of Nix's
  <filename>configure</filename> script.</para></footnote> The build
  can only be performed on a machine and operating system matching the
  platform identifier.  (Nix can automatically forward builds for
  other platforms by forwarding them to other machines; see <xref
  linkend='sec-distributed-builds' />.)</para></listitem>

  <listitem><para>There must be an attribute named
  <varname>name</varname> whose value must be a string.  This is used
  as a symbolic name for the package by <command>nix-env</command>,
  and it is appended to the hash in the output path of the
  derivation.</para></listitem>

  <listitem><para>There must be an attribute named
  <varname>builder</varname> that identifies the program that is
  executed to perform the build.  It can be either a derivation or a
  source (a local file reference, e.g.,
  <filename>./builder.sh</filename>).</para></listitem>

  <listitem><para>Every attribute is passed as an environment variable
  to the builder.  Attribute values are translated to environment
  variables as follows:

    <itemizedlist>

      <listitem><para>Strings, URIs, and integers are just passed
      verbatim.</para></listitem>

      <listitem><para>A <emphasis>path</emphasis> (e.g.,
      <filename>../foo/sources.tar</filename>) causes the referenced
      file to be copied to the store; its location in the store is put
      in the environment variable.  The idea is that all sources
      should reside in the Nix store, since all inputs to a derivation
      should reside in the Nix store.</para></listitem>

      <listitem><para>A <emphasis>derivation</emphasis> causes that
      derivation to be built prior to the present derivation; the
      output path is put in the environment
      variable.</para></listitem>

      <listitem><para>Lists of the previous types are also allowed.
      They are simply concatenated, separated by
      spaces.</para></listitem>

      <listitem><para><literal>true</literal> is passed as the string
      <literal>1</literal>, <literal>false</literal> and
      <literal>null</literal> are passed as an empty string.
      </para></listitem>
    </itemizedlist>

  </para></listitem>

  <listitem><para>The optional attribute <varname>args</varname>
  specifies command-line arguments to be passed to the builder.  It
  should be a list.</para></listitem>

</itemizedlist>

<para>(Note that <function>mkDerivation</function> in the standard
environment is a wrapper around <function>derivation</function> that
adds a default value for <varname>system</varname> and always uses
Bash as the builder, to which the supplied builder is passed as a
command-line argument.  See <xref linkend='sec-standard-environment'
/>.)</para>

<para>The builder is executed as follows:

<itemizedlist>

  <listitem><para>A temporary directory is created under the directory
  specified by <envar>TMPDIR</envar> (default
  <filename>/tmp</filename>) where the build will take place.  The
  current directory is changed to this directory.</para></listitem>

  <listitem><para>The environment is cleared and set to the derivation
  attributes, as specified above.</para></listitem>

  <listitem><para>In addition, the following variables are set:

  <itemizedlist>

    <listitem><para><envar>NIX_BUILD_TOP</envar> contains the path of
    the temporary directory for this build.</para></listitem>

    <listitem><para>Also, <envar>TMPDIR</envar>,
    <envar>TEMPDIR</envar>, <envar>TMP</envar>, <envar>TEMP</envar>
    are set to point to the temporary directory.  This is to prevent
    the builder from accidentally writing temporary files anywhere
    else.  Doing so might cause interference by other
    processes.</para></listitem>

    <listitem><para><envar>PATH</envar> is set to
    <filename>/path-not-set</filename> to prevent shells from
    initialising it to their built-in default value.</para></listitem>

    <listitem><para><envar>HOME</envar> is set to
    <filename>/homeless-shelter</filename> to prevent programs from
    using <filename>/etc/passwd</filename> or the like to find the
    user's home directory, which could cause impurity.  Usually, when
    <envar>HOME</envar> is set, it is used as the location of the home
    directory, even if it points to a non-existent
    path.</para></listitem>

    <listitem><para><envar>NIX_STORE</envar> is set to the path of the
    top-level Nix store directory (typically,
    <filename>/nix/store</filename>).</para></listitem>

    <listitem><para><envar>out</envar> is set to point to the output
    path of the derivation, which is a subdirectory of the Nix store.
    The output path is a concatenation of the cryptographic hash of
    all build inputs, and the <varname>name</varname>
    attribute.</para></listitem>
    
  </itemizedlist>

  </para></listitem>

  <listitem><para>If the output path already exists, it is removed.
  Also, locks are acquired to prevent multiple Nix instances from
  performing the same build at the same time.</para></listitem>

  <listitem><para>A log of the combined standard output and error is
  written to <filename>/nix/var/log/nix</filename>.</para></listitem>

  <listitem><para>The builder is executed with the arguments specified
  by the attribute <varname>args</varname>.  If it exits with exit
  code 0, it is considered to have succeeded.</para></listitem>

  <listitem><para>The temporary directory is removed (unless the
  <option>-K</option> option was specified).</para></listitem>

  <listitem><para>If the build was successful, Nix scans the output
  for references to the paths of the inputs.  These so-called
  <emphasis>retained dependencies</emphasis> could be used when the
  output of the derivation is used (e.g., when it's executed or used
  as input to another derivation), so if we deploy the derivation, we
  should copy the retained dependencies as well.  The scan is
  performed by looking for the hash parts of file names of the
  inputs.</para></listitem>

  <listitem><para>After the build, Nix sets the last-modified
  timestamp on all files in the build result to 0 (00:00:00 1/1/1970
  UTC), sets the group to the default group, and sets the mode of the
  file to 0444 or 0555 (i.e., read-only, with execute permission
  enabled if the file was originally executable).  Note that possible
  <literal>setuid</literal> and <literal>setgid</literal> bits are
  cleared.  Setuid and setgid programs are not currently supported by
  Nix.  This is because the Nix archives used in deployment have no
  concept of ownership information, and because it makes the build
  result dependent on the user performing the build.</para></listitem>

</itemizedlist>

</para>


<section><title>Advanced attributes</title>

<para>Derivations can declare some infrequently used optional
attributes.</para>

<variablelist>

  <varlistentry><term><varname>allowedReferences</varname></term>
  
    <listitem><para>The optional attribute
    <varname>allowedReferences</varname> specifies a list of legal
    references (dependencies) of the output of the builder.  For
    example,

<programlisting>
allowedReferences = [];
</programlisting>

    enforces that the output of a derivation cannot have any runtime
    dependencies on its inputs.  This is used in NixOS to check that
    generated files such as initial ramdisks for booting Linux don’t
    have accidental dependencies on other paths in the Nix
    store.</para></listitem>

  </varlistentry>

  
  <varlistentry><term><varname>exportReferencesGraph</varname></term>

    <listitem><para>This attribute allows builders access to the
    references graph of their inputs.  The attribute is a list of
    inputs in the Nix store whose references graph the builder needs
    to know.  The value of this attribute should be a list of pairs
    <literal>[<replaceable>name1</replaceable>
    <replaceable>path1</replaceable> <replaceable>name2</replaceable>
    <replaceable>path2</replaceable>
    <replaceable>...</replaceable>]</literal>.  The references graph
    of each <replaceable>pathN</replaceable> will be stored in a text
    file <replaceable>nameN</replaceable> in the temporary build
    directory.  The text files have the format used by
    <command>nix-store --register-validity</command> (with the deriver
    fields left empty).  For example, when the following derivation is
    built:

<programlisting>
derivation {
  ...
  exportReferencesGraph = ["libfoo-graph" libfoo];
};
</programlisting>

    the references graph of <literal>libfoo</literal> is placed in the
    file <filename>libfoo-graph</filename> in the temporary build
    directory.</para>

    <para><varname>exportReferencesGraph</varname> is useful for
    builders that want to do something with the closure of a store
    path.  Examples include the builders in NixOS that generate the
    initial ramdisk for booting Linux (a <command>cpio</command>
    archive containing the closure of the boot script) and the
    ISO-9660 image for the installation CD (which is populated with a
    Nix store containing the closure of a bootable NixOS
    configuration).</para></listitem>

  </varlistentry>


  <varlistentry xml:id="fixed-output-drvs">
    <term><varname>outputHash</varname></term>
    <term><varname>outputHashAlgo</varname></term>
    <term><varname>outputHashMode</varname></term>

    <listitem><para>These attributes declare that the derivation is a
    so-called <emphasis>fixed-output derivation</emphasis>, which
    means that a cryptographic hash of the output is already known in
    advance.  When the build of a fixed-output derivation finishes,
    Nix computes the cryptographic hash of the output and compares it
    to the hash declared with these attributes.  If there is a
    mismatch, the build fails.</para>

    <para>The rationale for fixed-output derivations is derivations
    such as those produced by the <function>fetchurl</function>
    function.  This function downloads a file from a given URL.  To
    ensure that the downloaded file has not been modified, the caller
    must also specify a cryptographic hash of the file.  For example,

<programlisting>
fetchurl {
  url = http://ftp.gnu.org/pub/gnu/hello/hello-2.1.1.tar.gz;
  md5 = "70c9ccf9fac07f762c24f2df2290784d";
}
</programlisting>

    It sometimes happens that the URL of the file changes, e.g.,
    because servers are reorganised or no longer available.  We then
    must update the call to <function>fetchurl</function>, e.g.,

<programlisting>
fetchurl {
  url = ftp://ftp.nluug.nl/pub/gnu/hello/hello-2.1.1.tar.gz;
  md5 = "70c9ccf9fac07f762c24f2df2290784d";
}
</programlisting>

    If a <function>fetchurl</function> derivation was treated like a
    normal derivation, the output paths of the derivation and
    <emphasis>all derivations depending on it</emphasis> would change.
    For instance, if we were to change the URL of the Glibc source
    distribution in Nixpkgs (a package on which almost all other
    packages depend) massive rebuilds would be needed.  This is
    unfortunate for a change which we know cannot have a real effect
    as it propagates upwards through the dependency graph.</para>

    <para>For fixed-output derivations, on the other hand, the name of
    the output path only depends on the <varname>outputHash*</varname>
    and <varname>name</varname> attributes, while all other attributes
    are ignored for the purpose of computing the output path.  (The
    <varname>name</varname> attribute is included because it is part
    of the path.)</para>

    <para>As an example, here is the (simplified) Nix expression for
    <varname>fetchurl</varname>:

<programlisting>
{stdenv, curl}: # The <command>curl</command> program is used for downloading.

{url, md5}:

stdenv.mkDerivation {
  name = baseNameOf (toString url);
  builder = ./builder.sh;
  buildInputs = [curl];

  # This is a fixed-output derivation; the output must be a regular
  # file with MD5 hash <varname>md5</varname>.
  outputHashMode = "flat";
  outputHashAlgo = "md5";
  outputHash = md5;
  
  inherit url;
}
</programlisting>

    </para>

    <para>The <varname>outputHashAlgo</varname> attribute specifies
    the hash algorithm used to compute the hash.  It can currently be
    <literal>"md5"</literal>, <literal>"sha1"</literal> or
    <literal>"sha256"</literal>.</para>

    <para>The <varname>outputHashMode</varname> attribute determines
    how the hash is computed.  It must be one of the following two
    values:

    <variablelist>

      <varlistentry><term><literal>"flat"</literal></term>

        <listitem><para>The output must be a non-executable regular
        file.  If it isn’t, the build fails.  The hash is simply
        computed over the contents of that file (so it’s equal to what
        Unix commands like <command>md5sum</command> or
        <command>sha1sum</command> produce).</para>

        <para>This is the default.</para></listitem>

      </varlistentry>
    
      <varlistentry><term><literal>"recursive"</literal></term>

        <listitem><para>The hash is computed over the NAR archive dump
        of the output (i.e., the result of <link
        linkend="refsec-nix-store-dump"><command>nix-store
        --dump</command></link>).  In this case, the output can be
        anything, including a directory tree.</para></listitem>

      </varlistentry>

    </variablelist>

    </para>

    <para>The <varname>outputHash</varname> attribute, finally, must
    be a string containing the hash in either hexadecimal or base-32
    notation.  (See the <link
    linkend="sec-nix-hash"><command>nix-hash</command> command</link>
    for information about converting to and from base-32
    notation.)</para></listitem>
    
  </varlistentry>

  
  <varlistentry><term><varname>impureEnvVars</varname></term>

    <listitem><para>This attribute allows you to specify a list of
    environment variables that should be passed from the environment
    of the calling user to the builder.  Usually, the environment is
    cleared completely when the builder is executed, but with this
    attribute you can allow specific environment variables to be
    passed unmodified.  For example, <function>fetchurl</function> in
    Nixpkgs has the line

<programlisting>
impureEnvVars = ["http_proxy" "https_proxy" <replaceable>...</replaceable>];
</programlisting>

    to make it use the proxy server configuration specified by the
    user in the environment variables <envar>http_proxy</envar> and
    friends.</para>

    <para>This attribute is only allowed in <link
    linkend="fixed-output-drvs">fixed-output derivations</link>, where
    impurities such as these are okay since (the hash of) the output
    is known in advance.  It is ignored for all other
    derivations.</para></listitem>

  </varlistentry>

  
  

</variablelist>

</section>


</section>



<xi:include href="builtins.xml" />


</section>



<section xml:id='sec-standard-environment'><title>The standard environment</title>


<para>The standard environment is used by passing it as an input
called <envar>stdenv</envar> to the derivation, and then doing

<programlisting>
source $stdenv/setup</programlisting>

at the top of the builder.</para>

<para>Apart from adding the aforementioned commands to the
<envar>PATH</envar>, <filename>setup</filename> also does the
following:

<itemizedlist>

  <listitem><para>All input packages specified in the
  <envar>buildInputs</envar> environment variable have their
  <filename>/bin</filename> subdirectory added to <envar>PATH</envar>,
  their <filename>/include</filename> subdirectory added to the C/C++
  header file search path, and their <filename>/lib</filename>
  subdirectory added to the linker search path.  This can be extended.
  For instance, when the <command>pkgconfig</command> package is
  used, the subdirectory <filename>/lib/pkgconfig</filename> of each
  input is added to the <envar>PKG_CONFIG_PATH</envar> environment
  variable.</para></listitem>

  <listitem><para>The environment variable
  <envar>NIX_CFLAGS_STRIP</envar> is set so that the compiler strips
  debug information from object files.  This can be disabled by
  setting <envar>NIX_STRIP_DEBUG</envar> to
  <literal>0</literal>.</para></listitem>

</itemizedlist>

</para>

<para>The <filename>setup</filename> script also exports a function
called <function>genericBuild</function> that knows how to build
typical Autoconf-style packages.  It can be customised to perform
builds for any type of package.  It is advisable to use
<function>genericBuild</function> since it provides facilities that
are almost always useful such as unpacking of sources, patching of
sources, nested logging, etc.</para>

<para>The definitive, up-to-date documentation of the generic builder
is the source itself, which resides in
<filename>pkgs/stdenv/generic/setup.sh</filename>.</para>


<section><title>Customising the generic builder</title>

<para>The operation of the generic builder can be modified in many
places by setting certain variables.  These <emphasis>hook
variables</emphasis> are typically set to the name of some shell
function defined by you.  For instance, to perform some additional
steps after <command>make install</command> you would set the
<varname>postInstall</varname> variable:

<programlisting>
postInstall=myPostInstall

myPostInstall() {
    mkdir $out/share/extra
    cp extrafiles/* $out/share/extra
}</programlisting>

</para>


</section>


<section><title>Debugging failed builds</title>

<para>At the beginning of each phase, the set of all shell variables
is written to the file <filename>env-vars</filename> at the top-level
build directory.  This is useful for debugging: it allows you to
recreate the environment in which a build was performed.  For
instance, if a build fails, then assuming you used the
<option>-K</option> flag, you can go to the output directory and
<quote>switch</quote> to the environment of the builder:

<screen>
$ nix-build -K ./foo.nix
... fails, keeping build directory `/tmp/nix-1234-0'

$ cd /tmp/nix-1234-0

$ source env-vars

<lineannotation>(edit some files...)</lineannotation>

$ make

<lineannotation>(execution continues with the same GCC, make, etc.)</lineannotation></screen>

</para>

</section>


</section>


</chapter>