</para>
<para>
- In GHC version 6.12 building shared libraries is supported for Linux on
- x86 and x86-64 architectures and there is partial support on Windows (see
- <xref linkend="win32-dlls"/>). The crucial difference in support on
- Windows is that it is not currently possible to build each Haskell
- package as a separate DLL, it is only possible to link an entire Haskell
- program as one massive DLL.
+ In GHC version 6.12 building shared libraries is supported for Linux (on
+ x86 and x86-64 architectures). GHC version 7.0 adds support on Windows
+ (see <xref linkend="win32-dlls"/>), FreeBSD and OpenBSD (x86 and x86-64),
+ Solaris (x86) and Mac OS X (x86 and PowerPC).
</para>
<para>
that it can be linked against shared library versions of Haskell
packages (such as base). The second is when linking, to link against
the shared versions of the packages' libraries rather than the static
- versions. Obviously this requires that the packages were build with
+ versions. Obviously this requires that the packages were built with
shared libraries. On supported platforms GHC comes with shared
libraries for all the core packages, but if you install extra packages
(e.g. with Cabal) then they would also have to be built with shared
In particular Haskell shared libraries <emphasis>must</emphasis> be
made into packages. You cannot freely assign which modules go in which
shared libraries. The Haskell shared libraries must match the package
- boundaries. Most of the conventions GHC expects when using packages are
- described in <xref linkend="building-packages"/>.
- </para>
- <para>
+ boundaries. The reason for this is that
GHC handles references to symbols <emphasis>within</emphasis> the same
shared library (or main executable binary) differently from references
to symbols <emphasis>between</emphasis> different shared libraries. GHC
<literal>-dynamic</literal> in the link step. That means to
statically link the rts all the base libraries into your new shared
library. This would make a very big, but standalone shared library.
- Indeed this is exactly what we must currently do on Windows where
- -dynamic is not yet supported (see <xref linkend="win32-dlls"/>).
On most platforms however that would require all the static libraries
to have been built with <literal>-fPIC</literal> so that the code is
suitable to include into a shared library and we do not do that at the
The details of how this works varies between platforms, in particular
the three major systems: Unix ELF platforms, Windows and Mac OS X.
</para>
+ <sect3 id="finding-shared-libs-unix">
+ <title>Unix</title>
<para>
On Unix there are two mechanisms. Shared libraries can be installed
into standard locations that the dynamic linker knows about. For
<para>
GHC has a <literal>-dynload</literal> linking flag to select the method
that is used to find shared libraries at runtime. There are currently
- three modes:
+ two modes:
<variablelist>
<varlistentry>
<term>sysdep</term>
<listitem>
<para>
A system-dependent mode. This is also the default mode. On Unix
- ELF systems this embeds rpaths into the shared library or
- executable. In particular it uses absolute paths to where the
- shared libraries for the rts and each package can be found.
- This means the program can immediately be run and it will be
- able to find the libraries it needs. However it may not be
- suitable for deployment if the libraries are installed in a
- different location on another machine.
+ ELF systems this embeds
+ <literal>RPATH</literal>/<literal>RUNPATH</literal> entries into the
+ shared library or executable. In particular it uses absolute paths to
+ where the shared libraries for the rts and each package can be found.
+ This means the program can immediately be run and it will be able to
+ find the libraries it needs. However it may not be suitable for
+ deployment if the libraries are installed in a different location on
+ another machine.
</para>
</listitem>
</varlistentry>
</varlistentry>
</variablelist>
To use relative paths for dependent libraries on Linux and Solaris you
- can use the <literal>deploy</literal> mode and pass suitable a -rpath
- flag to the linker:
+ can pass a suitable <literal>-rpath</literal> flag to the linker:
<programlisting>
ghc -dynamic Main.hs -o main -lfoo -L. -optl-Wl,-rpath,'$ORIGIN'
</programlisting>
executable e.g. <literal>-optl-Wl,-rpath,'$ORIGIN/lib'</literal>.
</para>
<para>
- The standard assumption on Darwin/MacOS X is that dynamic libraries will
+ This relative path technique can be used with either of the two
+ <literal>-dynload</literal> modes, though it makes most sense with the
+ <literal>deploy</literal> mode. The difference is that with the
+ <literal>deploy</literal> mode, the above example will end up with an ELF
+ <literal>RUNPATH</literal> of just <literal>$ORIGIN</literal> while with
+ the <literal>sysdep</literal> mode the <literal>RUNPATH</literal> will be
+ <literal>$ORIGIN</literal> followed by all the library directories of all
+ the packages that the program depends on (e.g. <literal>base</literal>
+ and <literal>rts</literal> packages etc.) which are typically absolute
+ paths. The unix tool <literal>readelf --dynamic</literal> is handy for
+ inspecting the <literal>RPATH</literal>/<literal>RUNPATH</literal>
+ entries in ELF shared libraries and executables.
+ </para>
+ </sect3>
+ <sect3 id="finding-shared-libs-mac">
+ <title>Mac OS X</title>
+ <para>
+ The standard assumption on Darwin/Mac OS X is that dynamic libraries will
be stamped at build time with an "install name", which is the full
ultimate install path of the library file. Any libraries or executables
that subsequently link against it (even if it hasn't been installed yet)
for you. It automatically sets the install name for dynamic libraries to
the absolute path of the ultimate install location.
</para>
+ </sect3>
</sect2>
</sect1>