[xemacs-hg @ 2002-07-31 07:23:39 by michaels] 2002-07-17 Marcus Crestani <crestani@informatik.uni-tuebingen.de> Markus Kaltenbach <makalten@informatik.uni-tuebingen.de> Mike Sperber <mike@xemacs.org> configure flag to turn these changes on: --use-kkcc First we added a dumpable flag to lrecord_implementation. It shows, if the object is dumpable and should be processed by the dumper. * lrecord.h (struct lrecord_implementation): added dumpable flag (MAKE_LRECORD_IMPLEMENTATION): fitted the different makro definitions to the new lrecord_implementation and their calls. Then we changed mark_object, that it no longer needs a mark method for those types that have pdump descritions. * alloc.c: (mark_object): If the object has a description, the new mark algorithm is called, and the object is marked according to its description. Otherwise it uses the mark method like before. These procedures mark objects according to their descriptions. They are modeled on the corresponding pdumper procedures. (mark_with_description): (get_indirect_count): (structure_size): (mark_struct_contents): These procedures still call mark_object, this is needed while there are Lisp_Objects without descriptions left. We added pdump descriptions for many Lisp_Objects: * extents.c: extent_auxiliary_description * database.c: database_description * gui.c: gui_item_description * scrollbar.c: scrollbar_instance_description * toolbar.c: toolbar_button_description * event-stream.c: command_builder_description * mule-charset.c: charset_description * device-msw.c: devmode_description * dialog-msw.c: mswindows_dialog_id_description * eldap.c: ldap_description * postgresql.c: pgconn_description pgresult_description * tooltalk.c: tooltalk_message_description tooltalk_pattern_description * ui-gtk.c: emacs_ffi_description emacs_gtk_object_description * events.c: * events.h: * event-stream.c: * event-Xt.c: * event-gtk.c: * event-tty.c: To write a pdump description for Lisp_Event, we converted every struct in the union event to a Lisp_Object. So we created nine new Lisp_Objects: Lisp_Key_Data, Lisp_Button_Data, Lisp_Motion_Data, Lisp_Process_Data, Lisp_Timeout_Data, Lisp_Eval_Data, Lisp_Misc_User_Data, Lisp_Magic_Data, Lisp_Magic_Eval_Data. We also wrote makro selectors and mutators for the fields of the new designed Lisp_Event and added everywhere these new abstractions. We implemented XD_UNION support in (mark_with_description), so we can describe exspecially console/device specific data with XD_UNION. To describe with XD_UNION, we added a field to these objects, which holds the variant type of the object. This field is initialized in the appendant constructor. The variant is an integer, it has also to be described in an description, if XD_UNION is used. XD_UNION is used in following descriptions: * console.c: console_description (get_console_variant): returns the variant (create_console): added variant initialization * console.h (console_variant): the different console types * console-impl.h (struct console): added enum console_variant contype * device.c: device_description (Fmake_device): added variant initialization * device-impl.h (struct device): added enum console_variant devtype * objects.c: image_instance_description font_instance_description (Fmake_color_instance): added variant initialization (Fmake_font_instance): added variant initialization * objects-impl.h (struct Lisp_Color_Instance): added color_instance_type * objects-impl.h (struct Lisp_Font_Instance): added font_instance_type * process.c: process_description (make_process_internal): added variant initialization * process.h (process_variant): the different process types

This is a list of the status of XEmacs on various machines and systems.
See PROBLEMS for particular problems and possible workarounds.

Much effort has gone into making XEmacs work on as many different
machines, configurations, and compilers as possible.

Much effort has gone into making XEmacs 64-bit clean.

Much effort has gone into removing system-specific code, and replacing
such code with autodetection at configure time.

The XEmacs core should build "out of the box" on most Unix-like systems.

XEmacs 21.2 was tested and `make check' succeeded on these Unix
configurations as of 2001-02-10:

alphaev56-dec-osf4.0e (both Compaq C and gcc)
i386-unknown-freebsd4.2
i386-unknown-netbsdelf1.5
i586-sco-sysv5uw7.0.1 (both SCO's cc and gcc)
i686-pc-linux-gnu
hppa2.0-hp-hpux10.20 (both HP's ANSI cc and gcc)
mips-sgi-irix6.5 (both MIPSpro cc and gcc)
rs6000-ibm-aix4.3.0.0 (both IBM's xlc and gcc)
sparc-sun-solaris2.6 (both Sun's Forte C and gcc)
sparc-sun-solaris2.7 (both Sun's Forte C and gcc)
sparc-sun-sunos4.1.4 (gcc)

Some systems have a dual mode 32-bit/64-bit compiler.  On most of
these, XEmacs requires the --pdump configure option to build
correctly with the 64-bit version of the compiler.

mips-sgi-irix6.5, CC="gcc -mabi=64"
mips-sgi-irix6.5, CC="cc -64"
rs6000-ibm-aix4.3.0.0, CC="cc -q64"

On most of these systems, XEmacs also builds with a C++ compiler,
but not "out of the box".  This feature is only for use by the
maintainers.

XEmacs 21.2 is known _not_ to work on any machines with m680x0
processors.  Sorry, all you sun3 and Unix PC nostalgia buffs out there.

VMS has never been supported by XEmacs.  In fact, all the old VMS code
inherited from Emacs has been removed.  Sorry, all you VMS fans out there.