[xemacs-hg @ 2002-04-14 12:41:59 by ben] latest changes TODO.ben-mule-21-5: Update. make-docfile.c: Add basic support for handling ISO 2022 doc strings -- we parse the basic charset designation sequences so we know whether we're in ASCII and have to pay attention to end quotes and such. Reformat code according to coding standards. abbrev.el: Add `global-abbrev-mode', which turns on or off abbrev-mode in all buffers. Added `defining-abbrev-turns-on-abbrev-mode' -- if non-nil, defining an abbrev through an interactive function will automatically turn on abbrev-mode, either globally or locally depending on the command. This is the "what you'd expect" behavior. indent.el: general function for indenting a balanced expression in a mode-correct way. Works similar to indent-region in that a mode can specify a specific command to do the whole operation; if not, figure out the region using forward-sexp and indent each line using indent-according-to-mode. keydefs.el: Removed. Modify M-C-backslash to do indent-region-or-balanced-expression. Make S-Tab just insert a TAB char, like it's meant to do. make-docfile.el: Now that we're using the call-process-in-lisp, we need to load an extra file win32-native.el because we're running a bare temacs. menubar-items.el: Totally redo the Cmds menu so that most used commands appear directly on the menu and less used commands appear in submenus. The old way may have been very pretty, but rather impractical. process.el: Under Windows, don't ever use old-call-process-internal, even in batch mode. We can do processes in batch mode. subr.el: Someone recoded truncate-string-to-width, saying "the FSF version is too complicated and does lots of hard-to-understand stuff" but the resulting recoded version was *totally* wrong! it misunderstood the basic point of this function, which is work in *columns* not chars. i dumped ours and copied the version from FSF 21.1. Also added truncate-string-with-continuation-dots, since this idiom is used often. config.inc.samp, xemacs.mak: Separate out debug and optimize flags. Remove all vestiges of USE_MINIMAL_TAGBITS, USE_INDEXED_LRECORD_IMPLEMENTATION, and GUNG_HO, since those ifdefs have long been removed. Make error-checking support actually work. Some rearrangement of config.inc.samp to make it more logical. Remove callproc.c and ntproc.c from xemacs.mak, no longer used. Make pdump the default. lisp.h: Add support for strong type-checking of Bytecount, Bytebpos, Charcount, Charbpos, and others, by making them classes, overloading the operators to provide integer-like operation and carefully controlling what operations are allowed. Not currently enabled in C++ builds because there are still a number of compile errors, and it won't really work till we merge in my "8-bit-Mule" workspace, in which I make use of the new types Charxpos, Bytexpos, Memxpos, representing a "position" either in a buffer or a string. (This is especially important in the extent code.) abbrev.c, alloc.c, eval.c, buffer.c, buffer.h, editfns.c, fns.c, text.h: Warning fixes, some of them related to new C++ strict type checking of Bytecount, Charbpos, etc. dired.c: Caught an actual error due to strong type checking -- char len being passed when should be byte len. alloc.c, backtrace.h, bytecode.c, bytecode.h, eval.c, sysdep.c: Further optimize Ffuncall: -- process arg list at compiled-function creation time, converting into an array for extra-quick access at funcall time. -- rewrite funcall_compiled_function to use it, and inline this function. -- change the order of check for magic stuff in SPECBIND_FAST_UNSAFE to be faster. -- move the check for need to garbage collect into the allocation code, so only a single flag needs to be checked in funcall. buffer.c, symbols.c: add debug funs to check on mule optimization info in buffers and strings. eval.c, emacs.c, text.c, regex.c, scrollbar-msw.c, search.c: Fix evil crashes due to eistrings not properly reinitialized under pdump. Redo a bit some of the init routines; convert some complex_vars_of() into simple vars_of(), because they didn't need complex processing. callproc.c, emacs.c, event-stream.c, nt.c, process.c, process.h, sysdep.c, sysdep.h, syssignal.h, syswindows.h, ntproc.c: Delete. Hallelujah, praise the Lord, there is no god but Allah!!! fix so that processes can be invoked in bare temacs -- thereby eliminating any need for callproc.c. (currently only eliminated under NT.) remove all crufty and unnecessary old process code in ntproc.c and elsewhere. move non-callproc-specific stuff (mostly environment) into process.c, so callproc.c can be left out under NT. console-tty.c, doc.c, file-coding.c, file-coding.h, lstream.c, lstream.h: fix doc string handling so it works with Japanese, etc docs. change handling of "character mode" so callers don't have to manually set it (quite error-prone). event-msw.c: spacing fixes. lread.c: eliminate unused crufty vintage-19 "FSF defun hack" code. lrecord.h: improve pdump description docs. buffer.c, ntheap.c, unexnt.c, win32.c, emacs.c: Mule-ize some unexec and startup code. It was pseudo-Mule-ized before by simply always calling the ...A versions of functions, but that won't cut it -- eventually we want to be able to run properly even if XEmacs has been installed in a Japanese directory. (The current problem is the timing of the loading of the Unicode tables; this will eventually be fixed.) Go through and fix various other places where the code was not Mule-clean. Provide a function mswindows_get_module_file_name() to get our own name without resort to PATH_MAX and such. Add a big comment in main() about the problem with Unicode table load timing that I just alluded to. emacs.c: When error-checking is enabled (interpreted as "user is developing XEmacs"), don't ask user to "pause to read messages" when a fatal error has occurred, because it will wedge if we are in an inner modal loop (typically when a menu is popped up) and make us unable to get a useful stack trace in the debugger. text.c: Correct update_entirely_ascii_p_flag to actually work. lisp.h, symsinit.h: declarations for above changes.

Debugging GNU Emacs
Copyright (c) 1985 Richard M. Stallman.

   Permission is granted to anyone to make or distribute verbatim copies
   of this document as received, in any medium, provided that the
   copyright notice and permission notice are preserved,
   and that the distributor grants the recipient permission
   for further redistribution as permitted by this notice.

   Permission is granted to distribute modified versions
   of this document, or of portions of it,
   under the above conditions, provided also that they
   carry prominent notices stating who last changed them.

On 4.2 you will probably find that dbx does not work for
debugging GNU Emacs.  For one thing, dbx does not keep the
inferior process's terminal modes separate from its own.
For another, dbx does not put the inferior in a separate
process group, which makes trouble when an inferior uses
interrupt input, which GNU Emacs must do on 4.2.

dbx has also been observed to have other problems,
such as getting incorrect values for register variables
in stack frames other than the innermost one.

The Emacs distribution now contains GDB, the new source-level
debugger for the GNU system.  GDB works for debugging Emacs.
GDB currently runs on vaxes under 4.2 and on Sun 2 and Sun 3
systems.


** Some useful techniques

`Fsignal' is a very useful place to stop in.
All Lisp errors go through there.

It is useful, when debugging, to have a guaranteed way
to return to the debugger at any time.  If you are using
interrupt-driven input, which is the default, then Emacs is using
RAW mode and the only way you can do it is to store
the code for some character into the variable stop_character:

    set stop_character = 29

makes Control-] (decimal code 29) the stop character.
Typing Control-] will cause immediate stop.  You cannot
use the set command until the inferior process has been started.
Put a breakpoint early in `main', or suspend the Emacs,
to get an opportunity to do the set command.

If you are using cbreak input (see the Lisp function set-input-mode),
then typing Control-g will cause a SIGINT, which will return control
to the debugger immediately unless you have done

    ignore 3  (in dbx)
or  handle 3 nostop noprint  (in gdb)

You will note that most of GNU Emacs is written to avoid
declaring a local variable in an inner block, even in
cases where using one would be the cleanest thing to do.
This is because dbx cannot access any of the variables
in a function which has even one variable defined in an
inner block.  A few functions in GNU Emacs do have variables
in inner blocks, only because I wrote them before realizing
that dbx had this problem and never rewrote them to avoid it.

I believe that GDB does not have such a problem.


** Examining Lisp object values.

When you have a live process to debug, and it has not encountered a
fatal error, you can use the GDB command `pr'.  First print the value
in the ordinary way, with the `p' command.  Then type `pr' with no
arguments.  This calls a subroutine which uses the Lisp printer.

If you can't use this command, either because the process can't run
a subroutine or because the data is invalid, you can fall back on
lower-level commands.

Use the `xtype' command to print out the data type of the last data
value.  Once you know the data type, use the command that corresponds
to that type.  Here are these commands:

    xint xptr xwindow xmarker xoverlay xmiscfree xintfwd xboolfwd xobjfwd
    xbufobjfwd xkbobjfwd xbuflocal xbuffer xsymbol xstring xvector xframe
    xwinconfig xcompiled xcons xcar xcdr xsubr xprocess xfloat xscrollbar

Each one of them applies to a certain type or class of types.
(Some of these types are not visible in Lisp, because they exist only
internally.)

Each x... command prints some information about the value, and
produces a GDB value (subsequently available in $) through which you
can get at the rest of the contents.

In general, most of the rest of the contents will be addition Lisp
objects which you can examine in turn with the x... commands.

** If GDB does not run and your debuggers can't load Emacs.

On some systems, no debugger can load Emacs with a symbol table,
perhaps because they all have fixed limits on the number of symbols
and Emacs exceeds the limits.  Here is a method that can be used
in such an extremity.  Do

    nm -n temacs > nmout
    strip temacs
    adb temacs
    0xd:i
    0xe:i
    14:i
    17:i
    :r -l loadup   (or whatever)

It is necessary to refer to the file `nmout' to convert
numeric addresses into symbols and vice versa.

It is useful to be running under a window system.
Then, if Emacs becomes hopelessly wedged, you can create
another window to do kill -9 in.  kill -ILL is often
useful too, since that may make Emacs dump core or return
to adb.


** Debugging incorrect screen updating.

To debug Emacs problems that update the screen wrong, it is useful
to have a record of what input you typed and what Emacs sent to the
screen.  To make these records, do

(open-dribble-file "~/.dribble")
(open-termscript "~/.termscript")

The dribble file contains all characters read by Emacs from the
terminal, and the termscript file contains all characters it sent to
the terminal.  The use of the directory `~/' prevents interference
with any other user.

If you have irreproducible display problems, put those two expressions
in your ~/.emacs file.  When the problem happens, exit the Emacs that
you were running, kill it, and rename the two files.  Then you can start
another Emacs without clobbering those files, and use it to examine them.