TODO - libtool - Git at Google

 In the near future:
 ********************

 * Check whether the version of libtool.m4 is compatible with
 ltconfig/ltmain.sh.  Meanwhile, the recommended approach for
 developers using automake is to insert libtool.m4 in acinclude.m4.

 * Inter-library dependencies should be fully tracked by libtool and
 need to work for ltlibraries too. This requires looking up installed
 libtool libraries for transparent support.  Support for this feature
 is under development, and is expected to be available in libtool 1.4.

 * We could have an option to hardcode paths into libraries, as well as
 binaries: `... -Wl,-soname -Wl,/tmp/libtest.so.0 ...'.  This is not
 possible on all platforms, and is in part obviated by the ability of
 linking libtool libraries specified with -lname, but it might still
 be desirable.

 * Lists of exported symbols should be stored in the pseudo library
 so that the size of lt_preloaded_symbols can be reduced.

 In the future:
 **************

 * Godmar Back writes:
   libltdl uses such stdio functions as fopen, fgets, feof, fclose, and others.
   These functions are not async-signal-safe. While this does not make
   libltdl unusable, it restricts its usefulness and puts an
   unnecessary burden on the user.

   As a remedy, I'd recommend to replace those functions with functions
   that POSIX says are async-signal-safe, such as open, read, close.
   This will require you to handle interrupted system calls and implement
   fgets, but the former isn't hard and there's plenty of implementations
   out from which you can steal the latter.

   I believe relying on async-signal-safe functions to the greatest extent
   possible would greatly improve libltdl's ability to be embedded in and
   used by other systems.

 * Arrange that EXEEXT suffixes are stripped from wrapper script names
 only when needed, and that a timestamp file or a wrapper program is
 created with the EXEEXT suffix, so that `make' doesn't build it every
 time.

 * Figure out how to use data items in dlls with win32.
 The difficult part is compiling each object which will be linked with an
 import lib differently than if it will be linked with a static lib.  This will
 almost definitely require that automake pass some hints about linkage in to
 each object compilation line.

 * Resolve the name clash between import libs and static libs on win32.
 Probably the best way to do this is to create lib$name-dll.a for the import
 library, and continue to use lib$name.a for the static lib.  libtool
 --mode=link can then favour -dll.a over .a if there is a choice.  No point in
 doing this until we can export data items (above).

 * If not cross-compiling, have the static flag test run the resulting
 binary to make sure everything works.

 * Implement full multi-language support.  Currently, this is only for
 C++, but there are beginnings of this in the manual (Other Languages).
 This includes writing libtool not to be so dependent on the compiler
 used to configure it.

 We especially need this for C++ linking, for which libtool currently
 does not handle static constructors properly, even on operating
 systems that support them.  ``Don't use static constructors'' is no
 longer a satisfactory answer.

 * We need some mechanism to allow users to pass flags to the linker
 and/or to the compiler, when creating libtool archives.  We could
 recognize linker flags such as `-Wl,flag' and `-Xlinker flag' in
 libtool's command line, and passing them down to the linker, if "$wl"
 is `-Wl,', or stripping the `-Wl,' part if we're calling `ld'
 directly.  We could also introduce `-Wc,flag' and `-Xcompiler flag' to
 allow unrecognized flags to be passed to the compiler, after stripping
 by libtool.  This is already implemented in the CVS tree.

 * Another form of convenience library is to have undocumented utility
 libraries, where only the shared version is installed.

 * We could use libtool object convenience libraries that resolve
 symbols to be included in a libtool archive.  This would require some
 sort of -whole-archive option, as well.

 * Currently, convenience libraries (.al) are built from .lo objects,
 except when --disable-shared.  When we can build both shared and
 static libraries, we should probably create a .al out of .lo objects
 and also a .a out of .o objects.  The .al would only be used to create
 shared libraries, whereas the .a would be used for creating static
 libraries and programs.  We could also explicitly support `empty'
 convenience libraries, that behave as macros that expand to a set of
 -Rs, -Ls and -ls switches.

 * We should include tests with convenience libraries and reloadable
 objects in the testsuite.

 * Try to find a work-around for -[all-]static and libltdl on platforms
 that will fail to find dlopening functions in this case.  Maybe
 creating an alternate libltdl that provides only for dlpreopening, or
 creating an additional static library to provide dummy implementations
 of the functions that can't be linked statically.  This could hardly
 be made completely transparent, though.

 * Need to finalize the documentation, and give a specification of
 `.la' files so that people can depend on their format.  This would be
 a good thing to put before the maintainance notes.

 * Filenames containing shell meta-characters are not properly handled
 by libtool.  Compiling a file named "a;b.c", for example, fails.

 * We could introduce a mechanism to allow for soname rewriting, to
 ease multi-libc support.  Installers could specify a prefix, suffix or
 sed command to modify the soname, and libtool would create the
 corresponding link.  This would allow for rebuilding a library with
 the same version number, but depending on different versions of libc,
 for example.  In the future, we might even have an option to encode
 the sonames of all dependencies of a library into its soname.

 Things to think about:
 **********************

 * Talk with RMS about his so-called `automatic package generation
 tool.'  This is probably what Thomas has been murmuring about for the
 Hurd.  We'll need to integrate package-supplied programs such as
 libtool into that scheme, since it manages some of the preinstall and
 postinstall commands, but isn't installed itself.  Probably, things
 like libtool should be distributed as part of such a binary package.

 * Maybe implement full support for other orthogonal library types
 (libhello_g, libhello_p, 64 vs 32-bit ABI's, etc).  Make these types
 configurable.
	In the near future:
	********************

	* Check whether the version of libtool.m4 is compatible with
	ltconfig/ltmain.sh. Meanwhile, the recommended approach for
	developers using automake is to insert libtool.m4 in acinclude.m4.

	* Inter-library dependencies should be fully tracked by libtool and
	need to work for ltlibraries too. This requires looking up installed
	libtool libraries for transparent support. Support for this feature
	is under development, and is expected to be available in libtool 1.4.

	* We could have an option to hardcode paths into libraries, as well as
	binaries: `... -Wl,-soname -Wl,/tmp/libtest.so.0 ...'. This is not
	possible on all platforms, and is in part obviated by the ability of
	linking libtool libraries specified with -lname, but it might still
	be desirable.

	* Lists of exported symbols should be stored in the pseudo library
	so that the size of lt_preloaded_symbols can be reduced.

	In the future:
	**************

	* Godmar Back writes:
	libltdl uses such stdio functions as fopen, fgets, feof, fclose, and others.
	These functions are not async-signal-safe. While this does not make
	libltdl unusable, it restricts its usefulness and puts an
	unnecessary burden on the user.

	As a remedy, I'd recommend to replace those functions with functions
	that POSIX says are async-signal-safe, such as open, read, close.
	This will require you to handle interrupted system calls and implement
	fgets, but the former isn't hard and there's plenty of implementations
	out from which you can steal the latter.

	I believe relying on async-signal-safe functions to the greatest extent
	possible would greatly improve libltdl's ability to be embedded in and
	used by other systems.

	* Arrange that EXEEXT suffixes are stripped from wrapper script names
	only when needed, and that a timestamp file or a wrapper program is
	created with the EXEEXT suffix, so that `make' doesn't build it every
	time.

	* Figure out how to use data items in dlls with win32.
	The difficult part is compiling each object which will be linked with an
	import lib differently than if it will be linked with a static lib. This will
	almost definitely require that automake pass some hints about linkage in to
	each object compilation line.

	* Resolve the name clash between import libs and static libs on win32.
	Probably the best way to do this is to create lib$name-dll.a for the import
	library, and continue to use lib$name.a for the static lib. libtool
	--mode=link can then favour -dll.a over .a if there is a choice. No point in
	doing this until we can export data items (above).

	* If not cross-compiling, have the static flag test run the resulting
	binary to make sure everything works.

	* Implement full multi-language support. Currently, this is only for
	C++, but there are beginnings of this in the manual (Other Languages).
	This includes writing libtool not to be so dependent on the compiler
	used to configure it.

	We especially need this for C++ linking, for which libtool currently
	does not handle static constructors properly, even on operating
	systems that support them. ``Don't use static constructors'' is no
	longer a satisfactory answer.

	* We need some mechanism to allow users to pass flags to the linker
	and/or to the compiler, when creating libtool archives. We could
	recognize linker flags such as `-Wl,flag' and `-Xlinker flag' in
	libtool's command line, and passing them down to the linker, if "$wl"
	is `-Wl,', or stripping the `-Wl,' part if we're calling `ld'
	directly. We could also introduce `-Wc,flag' and `-Xcompiler flag' to
	allow unrecognized flags to be passed to the compiler, after stripping
	by libtool. This is already implemented in the CVS tree.

	* Another form of convenience library is to have undocumented utility
	libraries, where only the shared version is installed.

	* We could use libtool object convenience libraries that resolve
	symbols to be included in a libtool archive. This would require some
	sort of -whole-archive option, as well.

	* Currently, convenience libraries (.al) are built from .lo objects,
	except when --disable-shared. When we can build both shared and
	static libraries, we should probably create a .al out of .lo objects
	and also a .a out of .o objects. The .al would only be used to create
	shared libraries, whereas the .a would be used for creating static
	libraries and programs. We could also explicitly support `empty'
	convenience libraries, that behave as macros that expand to a set of
	-Rs, -Ls and -ls switches.

	* We should include tests with convenience libraries and reloadable
	objects in the testsuite.

	* Try to find a work-around for -[all-]static and libltdl on platforms
	that will fail to find dlopening functions in this case. Maybe
	creating an alternate libltdl that provides only for dlpreopening, or
	creating an additional static library to provide dummy implementations
	of the functions that can't be linked statically. This could hardly
	be made completely transparent, though.

	* Need to finalize the documentation, and give a specification of
	`.la' files so that people can depend on their format. This would be
	a good thing to put before the maintainance notes.

	* Filenames containing shell meta-characters are not properly handled
	by libtool. Compiling a file named "a;b.c", for example, fails.

	* We could introduce a mechanism to allow for soname rewriting, to
	ease multi-libc support. Installers could specify a prefix, suffix or
	sed command to modify the soname, and libtool would create the
	corresponding link. This would allow for rebuilding a library with
	the same version number, but depending on different versions of libc,
	for example. In the future, we might even have an option to encode
	the sonames of all dependencies of a library into its soname.

	Things to think about:
	**********************

	* Talk with RMS about his so-called `automatic package generation
	tool.' This is probably what Thomas has been murmuring about for the
	Hurd. We'll need to integrate package-supplied programs such as
	libtool into that scheme, since it manages some of the preinstall and
	postinstall commands, but isn't installed itself. Probably, things
	like libtool should be distributed as part of such a binary package.

	* Maybe implement full support for other orthogonal library types
	(libhello_g, libhello_p, 64 vs 32-bit ABI's, etc). Make these types
	configurable.