<para>Sometimes it is useful to make the connection between a
source file and the command-line options it requires quite
- tight. For instance, if a Haskell source file uses GHC
- extensions, it will always need to be compiled with the
- <option>-fglasgow-exts</option> option. Rather than maintaining
+ tight. For instance, if a Haskell source file deliberately
+ uses name shadowing, it should be compiled with the
+ <option>-fno-warn-name-shadowing</option> option. Rather than maintaining
the list of per-file options in a <filename>Makefile</filename>,
it is possible to do this directly in the source file using the
<literal>OPTIONS_GHC</literal> pragma <indexterm><primary>OPTIONS_GHC
pragma</primary></indexterm>:</para>
<programlisting>
-{-# OPTIONS_GHC -fglasgow-exts #-}
+{-# OPTIONS_GHC -fno-warn-name-shadowing #-}
module X where
...
</programlisting>
- <para><literal>OPTIONS_GHC</literal> pragmas are only looked for at
- the top of your source files, upto the first
- (non-literate,non-empty) line not containing
- <literal>OPTIONS_GHC</literal>. Multiple <literal>OPTIONS_GHC</literal>
- pragmas are recognised. Do not put comments before, or on the same line
- as, the <literal>OPTIONS_GHC</literal> pragma.</para>
+ <para><literal>OPTIONS_GHC</literal> is a <emphasis>file-header pragma</emphasis>
+ (see <xref linkend="pragmas"/>).</para>
+
+ <para>Only <emphasis>dynamic</emphasis> flags can be used in an <literal>OPTIONS_GHC</literal> pragma
+ (see <xref linkend="static-dynamic-flags"/>).</para>
<para>Note that your command shell does not
get to the source file options, they are just included literally
<listitem>
<para>Most non-mode flags fall into this category. A dynamic flag
may be used on the command line, in a
- <literal>GHC_OPTIONS</literal> pragma in a source file, or set
+ <literal>OPTIONS_GHC</literal> pragma in a source file, or set
using <literal>:set</literal> in GHCi.</para>
</listitem>
</varlistentry>
of warnings which are generally likely to indicate bugs in your
program. These are:
<option>-fwarn-overlapping-patterns</option>,
- <option>-fwarn-deprecations</option>,
+ <option>-fwarn-warnings-deprecations</option>,
<option>-fwarn-deprecated-flags</option>,
<option>-fwarn-duplicate-exports</option>,
- <option>-fwarn-missing-fields</option>, and
- <option>-fwarn-missing-methods</option>. The following flags are
+ <option>-fwarn-missing-fields</option>,
+ <option>-fwarn-missing-methods</option>,
+ <option>-fwarn-lazy-unlifted-bindings</option>,
+ <option>-fwarn-wrong-do-bind</option>, and
+ <option>-fwarn-dodgy-foreign-imports</option>. The following
+ flags are
simple ways to select standard “packages” of warnings:
</para>
<option>-fwarn-simple-patterns</option>,
<option>-fwarn-tabs</option>,
<option>-fwarn-incomplete-record-updates</option>,
- <option>-fwarn-monomorphism-restriction</option>, and
+ <option>-fwarn-monomorphism-restriction</option>,
+ <option>-fwarn-unused-do-bind</option>, and
<option>-fwarn-implicit-prelude</option>.</para>
</listitem>
</varlistentry>
<variablelist>
<varlistentry>
- <term><option>-fwarn-deprecations</option>:</term>
+ <term><option>-fwarn-unrecognised-pragmas</option>:</term>
+ <listitem>
+ <indexterm><primary><option>-fwarn-unrecognised-pragmas</option></primary>
+ </indexterm>
+ <indexterm><primary>warnings</primary></indexterm>
+ <indexterm><primary>pragmas</primary></indexterm>
+ <para>Causes a warning to be emitted when a
+ pragma that GHC doesn't recognise is used. As well as pragmas
+ that GHC itself uses, GHC also recognises pragmas known to be used
+ by other tools, e.g. <literal>OPTIONS_HUGS</literal> and
+ <literal>DERIVE</literal>.</para>
+
+ <para>This option is on by default.</para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
+ <term><option>-fwarn-warnings-deprecations</option>:</term>
<listitem>
- <indexterm><primary><option>-fwarn-deprecations</option></primary>
+ <indexterm><primary><option>-fwarn-warnings-deprecations</option></primary>
</indexterm>
+ <indexterm><primary>warnings</primary></indexterm>
<indexterm><primary>deprecations</primary></indexterm>
- <para>Causes a warning to be emitted when a deprecated
- function or type is used. Entities can be marked as
- deprecated using a pragma, see <xref
- linkend="deprecated-pragma"/>.</para>
+ <para>Causes a warning to be emitted when a
+ module, function or type with a WARNING or DEPRECATED pragma
+ is used. See <xref linkend="warning-deprecated-pragma"/> for more
+ details on the pragmas.</para>
<para>This option is on by default.</para>
</listitem>
</varlistentry>
<varlistentry>
+ <term><option>-fwarn-dodgy-foreign-imports</option>:</term>
+ <listitem>
+ <indexterm><primary><option>-fwarn-dodgy-foreign-imports</option></primary>
+ </indexterm>
+ <para>Causes a warning to be emitted for foreign imports of
+ the following form:</para>
+<programlisting>
+foreign import "f" f :: FunPtr t
+</programlisting>
+ <para>on the grounds that it probably should be</para>
+<programlisting>
+foreign import "&f" f :: FunPtr t
+</programlisting>
+ <para>The first form declares that `f` is a (pure) C
+ function that takes no arguments and returns a pointer to a
+ C function with type `t`, whereas the second form declares
+ that `f` itself is a C function with type `t`. The first
+ declaration is usually a mistake, and one that is hard to
+ debug because it results in a crash, hence this
+ warning.</para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><option>-fwarn-dodgy-imports</option>:</term>
<listitem>
<indexterm><primary><option>-fwarn-dodgy-imports</option></primary>
</indexterm>
- <para>Causes a warning to be emitted when a a datatype
+ <para>Causes a warning to be emitted when a datatype
<literal>T</literal> is imported
with all constructors, i.e. <literal>T(..)</literal>, but has been
exported abstractly, i.e. <literal>T</literal>.</para>
</varlistentry>
<varlistentry>
+ <term><option>-fwarn-lazy-unlifted-bindings</option>:</term>
+ <listitem>
+ <indexterm><primary><option>-fwarn-lazy-unlifted-bindings</option></primary>
+ </indexterm>
+ <para>Causes a warning to be emitted when an unlifted type
+ is bound in a way that looks lazy, e.g.
+ <literal>where (I# x) = ...</literal>. Use
+ <literal>where !(I# x) = ...</literal> instead. This will be an
+ error, rather than a warning, in GHC 6.14.
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><option>-fwarn-duplicate-exports</option>:</term>
<listitem>
<indexterm><primary><option>-fwarn-duplicate-exports</option></primary></indexterm>
typographical errors that turn into hard-to-find bugs, e.g.,
in the inadvertent capture of what would be a recursive call in
<literal>f = ... let f = id in ... f ...</literal>.</para>
+ <para>The warning is suppressed for names beginning with an underscore. For example
+ <programlisting>
+ f x = do { _ignore <- this; _ignore <- that; return (the other) }
+ </programlisting>
+ </para>
</listitem>
</varlistentry>
<para>The trouble with orphans is that GHC must pro-actively read the interface
files for all orphan modules, just in case their instances or rules
play a role, whether or not the module's interface would otherwise
- be of any use. Other things being equal, avoid orphan modules.</para>
+ be of any use. See <xref linkend="orphan-modules"/> for details.
+ </para>
</listitem>
</varlistentry>
</listitem>
</varlistentry>
+ <varlistentry>
+ <term><option>-fwarn-unused-do-bind</option>:</term>
+ <listitem>
+ <indexterm><primary><option>-fwarn-unused-do-bind</option></primary></indexterm>
+ <indexterm><primary>unused do binding, warning</primary></indexterm>
+ <indexterm><primary>do binding, unused</primary></indexterm>
+
+ <para>Report expressions occuring in <literal>do</literal> and <literal>mdo</literal> blocks
+ that appear to silently throw information away.
+ For instance <literal>do { mapM popInt xs ; return 10 }</literal> would report
+ the first statement in the <literal>do</literal> block as suspicious,
+ as it has the type <literal>StackM [Int]</literal> and not <literal>StackM ()</literal>, but that
+ <literal>[Int]</literal> value is not bound to anything. The warning is suppressed by
+ explicitly mentioning in the source code that your program is throwing something away:
+ <programlisting>
+ do { _ <- mapM popInt xs ; return 10 }
+ </programlisting>
+ Of course, in this particular situation you can do even better:
+ <programlisting>
+ do { mapM_ popInt xs ; return 10 }
+ </programlisting>
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
+ <term><option>-fwarn-wrong-do-bind</option>:</term>
+ <listitem>
+ <indexterm><primary><option>-fwarn-wrong-do-bind</option></primary></indexterm>
+ <indexterm><primary>apparently erroneous do binding, warning</primary></indexterm>
+ <indexterm><primary>do binding, apparently erroneous</primary></indexterm>
+
+ <para>Report expressions occuring in <literal>do</literal> and <literal>mdo</literal> blocks
+ that appear to lack a binding.
+ For instance <literal>do { return (popInt 10) ; return 10 }</literal> would report
+ the first statement in the <literal>do</literal> block as suspicious,
+ as it has the type <literal>StackM (StackM Int)</literal> (which consists of two nested applications
+ of the same monad constructor), but which is not then "unpacked" by binding the result.
+ The warning is suppressed by explicitly mentioning in the source code that your program is throwing something away:
+ <programlisting>
+ do { _ <- return (popInt 10) ; return 10 }
+ </programlisting>
+ For almost all sensible programs this will indicate a bug, and you probably intended to write:
+ <programlisting>
+ do { popInt 10 ; return 10 }
+ </programlisting>
+ </para>
+ </listitem>
+ </varlistentry>
+
</variablelist>
<para>If you're feeling really paranoid, the
linkend="lang-parallel" /> we describe the language features that affect
parallelism.</para>
- <sect2 id="parallel-options">
- <title>Options to enable SMP parallelism</title>
+ <sect2 id="parallel-compile-options">
+ <title>Compile-time options for SMP parallelism</title>
<para>In order to make use of multiple CPUs, your program must be
linked with the <option>-threaded</option> option (see <xref
- linkend="options-linker" />). Then, to run a program on multiple
- CPUs, use the RTS <option>-N</option> option:</para>
+ linkend="options-linker" />). Additionally, the following
+ compiler options affect parallelism:</para>
<variablelist>
+ <varlistentry>
+ <term><option>-feager-blackholing</option></term>
+ <indexterm><primary><option>-feager-blackholing</option></primary></indexterm>
+ <listitem>
+ <para>
+ Blackholing is the act of marking a thunk (lazy
+ computuation) as being under evaluation. It is useful for
+ three reasons: firstly it lets us detect certain kinds of
+ infinite loop (the <literal>NonTermination</literal>
+ exception), secondly it avoids certain kinds of space
+ leak, and thirdly it avoids repeating a computation in a
+ parallel program, because we can tell when a computation
+ is already in progress.</para>
+
+ <para>
+ The option <option>-feager-blackholing</option> causes
+ each thunk to be blackholed as soon as evaluation begins.
+ The default is "lazy blackholing", whereby thunks are only
+ marked as being under evaluation when a thread is paused
+ for some reason. Lazy blackholing is typically more
+ efficient (by 1-2% or so), because most thunks don't
+ need to be blackholed. However, eager blackholing can
+ avoid more repeated computation in a parallel program, and
+ this often turns out to be important for parallelism.
+ </para>
+
+ <para>
+ We recommend compiling any code that is intended to be run
+ in parallel with the <option>-feager-blackholing</option>
+ flag.
+ </para>
+ </listitem>
+ </varlistentry>
+ </variablelist>
+ </sect2>
+
+ <sect2 id="parallel-options">
+ <title>RTS options for SMP parallelism</title>
+
+ <para>To run a program on multiple CPUs, use the
+ RTS <option>-N</option> option:</para>
+
+ <variablelist>
<varlistentry>
- <term><option>-N<replaceable>x</replaceable></option></term>
+ <term><option>-N<optional><replaceable>x</replaceable></optional></option></term>
<listitem>
<para><indexterm><primary><option>-N<replaceable>x</replaceable></option></primary><secondary>RTS option</secondary></indexterm>
Use <replaceable>x</replaceable> simultaneous threads when
running the program. Normally <replaceable>x</replaceable>
- should be chosen to match the number of CPU cores on the machine.
- There is no means (currently) by which this value may vary after
- the program has started.</para>
-
- <para>For example, on a dual-core machine we would probably use
+ should be chosen to match the number of CPU cores on the
+ machine<footnote><para>Whether hyperthreading cores should be counted or not is an
+ open question; please feel free to experiment and let us know what
+ results you find.</para></footnote>. For example,
+ on a dual-core machine we would probably use
<literal>+RTS -N2 -RTS</literal>.</para>
- <para>Whether hyperthreading cores should be counted or not is an
- open question; please feel free to experiment and let us know what
- results you find.</para>
+ <para>Omitting <replaceable>x</replaceable>,
+ i.e. <literal>+RTS -N -RTS</literal>, lets the runtime
+ choose the value of <replaceable>x</replaceable> itself
+ based on how many processors are in your machine.</para>
+
+ <para>Be careful when using all the processors in your
+ machine: if some of your processors are in use by other
+ programs, this can actually harm performance rather than
+ improve it.</para>
+
+ <para>Setting <option>-N</option> also has the effect of
+ enabling the parallel garbage collector (see
+ <xref linkend="rts-options-gc" />).</para>
+
+ <para>There is no means (currently) by which this value
+ may vary after the program has started.</para>
</listitem>
</varlistentry>
</variablelist>
+
+ <para>The following options affect the way the runtime schedules
+ threads on CPUs:</para>
+
+ <variablelist>
+ <varlistentry>
+ <term><option>-qm</option></term>
+ <indexterm><primary><option>-qm</option></primary><secondary>RTS
+ option</secondary></indexterm>
+ <listitem>
+ <para>Disable automatic migration for load balancing.
+ Normally the runtime will automatically try to schedule
+ threads across the available CPUs to make use of idle
+ CPUs; this option disables that behaviour. It is probably
+ only of use if you are explicitly scheduling threads onto
+ CPUs with <literal>GHC.Conc.forkOnIO</literal>.</para>
+ </listitem>
+ </varlistentry>
+ <varlistentry>
+ <term><option>-qw</option></term>
+ <indexterm><primary><option>-qw</option></primary><secondary>RTS
+ option</secondary></indexterm>
+ <listitem>
+ <para>Migrate a thread to the current CPU when it is woken
+ up. Normally when a thread is woken up after being
+ blocked it will be scheduled on the CPU it was running on
+ last; this option allows the thread to immediately migrate
+ to the CPU that unblocked it.</para>
+
+ <para>The rationale for allowing this eager migration is
+ that it tends to move threads that are communicating with
+ each other onto the same CPU; however there are
+ pathalogical situations where it turns out to be a poor
+ strategy. Depending on the communication pattern in your
+ program, it may or may not be a good idea.</para>
+ </listitem>
+ </varlistentry>
+ </variablelist>
</sect2>
<sect2>
<title>Hints for using SMP parallelism</title>
- <para>Add the <literal>-sstderr</literal> RTS option when
+ <para>Add the <literal>-s</literal> RTS option when
running the program to see timing stats, which will help to tell you
whether your program got faster by using more CPUs or not. If the user
time is greater than
<para>GHC can dump its optimized intermediate code (said to be in “Core” format)
to a file as a side-effect of compilation. Non-GHC back-end tools can read and process Core files; these files have the suffix
- <filename>.hcr</filename>. The Core format is described in <ulink url="http://www.haskell.org/ghc/docs/papers/core.ps.gz">
+ <filename>.hcr</filename>. The Core format is described in <ulink url="../ext-core/core.pdf">
<citetitle>An External Representation for the GHC Core Language</citetitle></ulink>,
and sample tools
for manipulating Core files (in Haskell) are in the GHC source distribution
projects / ghc-hetmet.git / blobdiff