Add several new record features
[ghc-hetmet.git] / docs / users_guide / using.xml
index 82d7afe..3ab73a1 100644 (file)
@@ -839,7 +839,17 @@ ghc -c Foo.hs</screen>
        <term><option>-Wall</option>:</term>
        <listitem>
          <indexterm><primary><option>-Wall</option></primary></indexterm>
-         <para>Turns on all warning options.</para>
+         <para>Turns on all warning options that indicate potentially
+         suspicious code.  The warnings that are
+         <emphasis>not</emphasis> enabled by <option>-Wall</option>
+         are:</para>
+          <itemizedlist>
+            <listitem><option>-fwarn-simple-patterns</option></listitem>
+            <listitem><option>-fwarn-tabs</option></listitem>
+            <listitem><option>-fwarn-incomplete-record-updates</option></listitem>
+            <listitem><option>-fwarn-monomorphism-restriction</option></listitem>
+            <listitem><option>-fwarn-implicit-prelude</option></listitem>
+          </itemizedlist>
        </listitem>
       </varlistentry>
 
@@ -905,6 +915,31 @@ ghc -c Foo.hs</screen>
       </varlistentry>
 
       <varlistentry>
+        <term><option>-fwarn-implicit-prelude</option>:</term>
+        <listitem>
+          <indexterm><primary><option>-fwarn-implicit-prelude</option></primary></indexterm>
+          <indexterm><primary>implicit prelude, warning</primary></indexterm>
+          <para>Have the compiler warn if the Prelude is implicitly
+          imported.  This happens unless either the Prelude module is
+          explicitly imported with an <literal>import ... Prelude ...</literal>
+          line, or this implicit import is disabled (either by
+          <option>-fno-implicit-prelude</option> or a
+          <literal>LANGUAGE NoImplicitPrelude</literal> pragma).</para>
+
+          <para>Note that no warning is given for syntax that implicitly
+          refers to the Prelude, even if <option>-fno-implicit-prelude</option>
+          would change whether it refers to the Prelude.
+          For example, no warning is given when
+          <literal>368</literal> means
+          <literal>Prelude.fromInteger (368::Prelude.Integer)</literal>
+          (where <literal>Prelude</literal> refers to the actual Prelude module,
+          regardless of the imports of the module being compiled).</para>
+
+          <para>This warning is off by default.</para>
+        </listitem>
+      </varlistentry>
+
+      <varlistentry>
        <term><option>-fwarn-incomplete-patterns</option>:</term>
        <listitem>
          <indexterm><primary><option>-fwarn-incomplete-patterns</option></primary></indexterm>
@@ -1025,7 +1060,7 @@ f foo = foo { x = 6 }
           in the inadvertent cyclic definition <literal>let x = ... x
           ... in</literal>.</para>
 
-         <para>Consequently, this option does
+         <para>Consequently, this option
           <emphasis>will</emphasis> complain about cyclic recursive
           definitions.</para>
        </listitem>
@@ -1136,6 +1171,20 @@ f "2"    = 2
       </varlistentry>
 
       <varlistentry>
+       <term><option>-fwarn-monomorphism-restriction</option>:</term>
+       <listitem>
+         <indexterm><primary><option>-fwarn-monomorphism-restriction</option></primary></indexterm>
+         <indexterm><primary>monomorphism restriction, warning</primary></indexterm>
+         <para>Have the compiler warn/inform you where in your source
+          the Haskell Monomorphism Restriction is applied.  If applied silently
+         the MR can give rise to unexpected behaviour, so it can be helpful
+         to have an explicit warning that it is being applied.</para>
+
+         <para>This warning is off by default.</para>
+       </listitem>
+      </varlistentry>
+
+      <varlistentry>
        <term><option>-fwarn-unused-binds</option>:</term>
        <listitem>
          <indexterm><primary><option>-fwarn-unused-binds</option></primary></indexterm>
@@ -1227,7 +1276,7 @@ f "2"    = 2
       <para>Note that higher optimisation levels cause more
       cross-module optimisation to be performed, which can have an
       impact on how much of your program needs to be recompiled when
-      you change something.  This is one reaosn to stick to
+      you change something.  This is one reason to stick to
       no-optimisation when developing code.</para>
 
       <variablelist>
@@ -1436,6 +1485,50 @@ f "2"    = 2
 
        <varlistentry>
          <term>
+            <option>-fno-state-hack</option>
+            <indexterm><primary><option>-fno-state-hack</option></primary></indexterm>
+          </term>
+         <listitem>
+           <para>Turn off the "state hack" whereby any lambda with a
+             <literal>State#</literal> token as argument is considered to be
+             single-entry, hence it is considered OK to inline things inside
+             it.  This can improve performance of IO and ST monad code, but it
+           runs the risk of reducing sharing.</para> 
+         </listitem>
+       </varlistentry>
+
+       <varlistentry>
+         <term>
+            <option>-fomit-interface-pragmas</option>
+           <indexterm><primary><option>-fomit-interface-pragmas</option></primary></indexterm>
+          </term>
+         <listitem>
+           <para>Tells GHC to omit all inessential information from the interface file
+             generated for the module being compiled (say M).  This means that a module
+             importing M will see only the <emphasis>types</emphasis> of the functions that M exports, but not
+             their unfoldings, strictness info, etc.  Hence, for example,
+             no function exported by M will be inlined
+             into an importing module.  The benefit is that modules that import M will
+             need to be recompiled less often (only when M's exports change their type,
+             not when they change their implementation).
+             </para>
+         </listitem>
+       </varlistentry>
+
+       <varlistentry>
+         <term>
+            <option>-fignore-interface-pragmas</option>
+           <indexterm><primary><option>-fignore-interface-pragmas</option></primary></indexterm>
+          </term>
+         <listitem>
+           <para>Tells GHC to ignore all inessential information when reading interface files.
+           That is, even if <filename>M.hi</filename> contains unfolding or strictness information
+           for a function, GHC will ignore that information.</para>
+         </listitem>
+       </varlistentry>
+
+       <varlistentry>
+         <term>
             <option>-funbox-strict-fields</option>:
            <indexterm><primary><option>-funbox-strict-fields</option></primary></indexterm>
            <indexterm><primary>strict constructor fields</primary></indexterm>
@@ -1458,7 +1551,7 @@ f "2"    = 2
 
        <varlistentry>
          <term>
-            <option>-funfolding-update-in-place&lt;n&gt;</option>
+            <option>-funfolding-update-in-place=<replaceable>n</replaceable></option>
             <indexterm><primary><option>-funfolding-update-in-place</option></primary></indexterm>
           </term>
          <listitem>
@@ -1476,7 +1569,7 @@ f "2"    = 2
 
        <varlistentry>
          <term>
-            <option>-funfolding-creation-threshold&lt;n&gt;</option>:
+            <option>-funfolding-creation-threshold=<replaceable>n</replaceable></option>:
            <indexterm><primary><option>-funfolding-creation-threshold</option></primary></indexterm>
            <indexterm><primary>inlining, controlling</primary></indexterm>
            <indexterm><primary>unfolding, controlling</primary></indexterm>
@@ -1502,7 +1595,7 @@ f "2"    = 2
        </varlistentry>
 
        <varlistentry>
-         <term><option>-funfolding-use-threshold&lt;n&gt;</option>:</term>
+         <term><option>-funfolding-use-threshold=<replaceable>n</replaceable></option></term>
          <listitem>
            <indexterm><primary><option>-funfolding-use-threshold</option></primary></indexterm>
            <indexterm><primary>inlining, controlling</primary></indexterm>