SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY

[smlnj] / sml / trunk / HISTORY

Diff of /sml/trunk/HISTORY

-revision 572, Thu Mar  9 02:43:06 2000 UTC
+revision 815, Fri May  4 05:09:10 2001 UTC
 Line 11
  Date:
  Tag: <post-commit CVS tag>
  Description:
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2001/05/04 01:57:00 EDT
+ Tag: leunga-20010504-sync
+ Description:
+   MLRISC features.
+. Fix to CMPXCHG instructions.
+. Changed RA interface to allow annotations in callbacks.
+. Added a new method to the stream interface to allow annotations updates.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/05/01 11:45:00 EDT
+ Tag: blume-20010501-pcedittmp
+ Description:
+ Changed install.sh to use the current working directory instead of
+ /usr/tmp for a temporary file (pcedittmp).  The previous choice
+ of /usr/tmp caused trouble with MacOS X because of file premission
+ problems.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/04/20 11:10:00 EDT
+ Tag: blume-20010420-inMLflag
+ Description:
+  - added vp_limitPtrMask to vproc-state.h
+    (for use by the raw-C-calls mechanism to implement proper interrupt
+     handling)
+  - made the ML compiler aware of various data-structure offsets so it
+    can generate code for accessing the vp_inML flag and vp_limitPtrMask
+  - tweaked mlriscGen.sml to have it emit interrupt-handling code for
+    raw C-calls
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2001/04/20 09:15:28 EDT
+ Tag: george-20010420-macosX
+ Description:
+  - Changes to port to Mac OS X; Darwin.
+  - In the process I found that sqrt was broken on the PPC, because the
+    fsqrt instruction is not implemented.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/04/18 12:45:00 EDT
+ Tag: blume-20010418-ccalls
+ Description:
+  - fixed two off-by-4 errors in the x86-specific c-calls implementation
+    (this bug prevented structure arguments containing pointers from being
+     passed correctly)
+  - changed the raw-C-call code in mlriscGen.sml in such a way that
+    structure arguments are represented as a pointer to the beginning
+    of the structure (instead of having a series of synthesized arguments,
+    one for each structure member)
+  - made makeml script's verbosity level configurable via environment
+    variable (MAKEML_VERBOSITY)
+  - eliminated placeholder implementations for f32l, w16s, i16s, and f32s
+    in rawmem-x86.sml; we are now using the real thing
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/03/22 16:25:00 EST
+ Tag: blume-20010322-bootfiles
+ Description:
+ Created a new set of bootfiles (for your automatic installation convenience).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/03/22 15:10:00 EST
+ Tag: blume-20010322-rawmem-parcm
+ Description:
+. All "raw memory access" primitives for the new FFI are implemented now
+    (at least on the x86).
+. Some further cleanup of CM's parallel make mechanism.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/03/19 17:53:00 EST
+ Tag: blume-20010319-parallel
+ Description:
+ Parallel make (using compile servers) now works again.
+ To this end, CM.stabilize and CMB.make have been modified to work in
+ two passes when compile servers are attached:
+. Compile everything, do not perform stabilization; this pass
+       uses compile servers
+. Stabilize everything; this pass does not use compile servers
+ If there are no compile servers, the two passes are combined into one
+ (as before).  Splitting the passes increases the inherent parallelism
+ in the dependency graph because the entire graph including all
+ libraries is available at the same time.  This, in turn, improves
+ server utilization.  The downside is that the master process will
+ have to do some extra work after compilation is done (because for
+ technical reasons it must re-read all the binfiles during stabilization).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/03/16 12:22:00 EST
+ Tag: blume-20010316-bootfiles
+ Description:
+ Created a new set of bootfiles (for your automatic installation convenience).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/03/16 11:00:00 EST
+ Tag: blume-20010316-MLTREE-fixup
+ Description:
+ This is a minor fixup for an (untagged) earlier commit by Allen.
+ (A file was missing).
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: Mon Mar  5 18:54:57 EST 2001
+ Tag: leunga-20010305-cut-support
+. New support for alternative control-flow in MLTREE.
+    Currently we support
+       FLOW_TO(CALL ...., [k1,...,kn])
+    This is needed for 'cuts to' in C-- and try/handle-like constructs
+    in Moby
+    New assembler flag "asm-show-cutsto" to turn on control-flow debugging.
+. Register Allocator
+    Changes in interface [from Fermin, John]
+. Alpha 8-bit SLL support [Fermin]
+. All architectures
+    A new module (ClusterExpandCopies) for expanding parallel copies.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2001/02/27 23:07:00 EST
+ Tag: leunga-20010227-minor-stuff
+. Alpha bug fix for CMOVNE
+. Handle mltree COND(..,FCMP ...,...)
+. Bug fix in simplifier
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/30 17:50:00 EST
+ Tag: blume-20010130-sync
+ Description:
+ This is just a minor update to sync my devel branch with the main brach.
+ The only visible change is the addition of some README files.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/12 23:30:00 JST
+ Tag: blume-20010112-bootfiles
+ Description:
+ Made a new set of bootfiles that goes with the current state of the
+ repository.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/12 21:20:00 JST
+ Tag: blume-20010112-sync
+ Description:
+ I am just flushing out some minor changes that had accumulated in
+ my private branch in order to sync with the main tree.  (This is
+ mainly because I had CVS trouble when trying to merge _into_ my
+ private branch.)
+ Most people should be completely unaffected by this.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: Thu Jan 11 21:03:00 EST 2001
+ Tag: leunga-20010111-labexp=mltree
+ Description:
+.  Removed the type LabelExp and replace it by MLTree.
+.  Rewritten mltree-simplify with the pattern matcher tool.
+.  There were some bugs in alpha code generator which would break
+-bit code generation.
+.  Redo the tools to generate code with the
+.  The CM files in MLRISC (and in src/system/smlnj/MLRISC)
+     are now generated by perl scripts.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/10 21:55:00 JST
+ Tag: blume-20010110-rcc
+ Description:
+ The RCC stuff now seems to work (but only on the x86).
+ This required hacking of the c-calls interface (and -implementation) in
+ MLRISC.
+ Normal compiler users should be unaffected.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/09 01:20:00 JST
+ Tag: blume-20010109-rcc
+ Description:
+ This is a fairly big patch, flushing out a large number of pending
+ changes that I made to my development copy over the last couple of days.
+ Of practical relevance at this moment is a workaround for a pickling
+ bug that Allen ran into the other day.  The cause of the bug itself is
+ still unknown and it might be hard to fix it properly, but the
+ workaround has some merits of its own (namely somewhat reducing pickling
+ overhead for certain libraries).  Therefore, I think this solution should
+ be satisfactory at this time.
+ The rest of the changes (i.e., the vast majority) has to do with my
+ ongoing efforts of providing direct support for C function calls from
+ ML.  At the moment there is a new primop "RAW_CCALL", typing magic
+ in types/cproto.sml (invoked from FLINT/trans/translate.sml), a new
+ case in the FLINT CPS datatype (RCC), changes to cps/convert.sml to
+ translate uses of RAW_CCALL into RCC, and changes to mlriscGen.sml to
+ handle RCC.
+ The last part (the changes to mlriscGen.sml) are still known to be
+ wrong on the x86 and not implemented on all other architectures.  But
+ the infrastructure is in place. I had to change a few functor
+ signatures in the backend to be able to route the CCalls interface
+ from MLRISC there, and I had to specialize the mltree type (on the
+ x86) to include the necessary extensions. (The extensions themselves
+ were already there and redy to go in MLRISC/x86).
+ Everything should be very happy as soon as someone helps me with
+ mlriscGen.sml...
+ In any case, nothing of this should matter to anyone as long as the
+ new primop is not being used (which is going to be the case unless you
+ find it where I hid it :). The rest of the compiler is completely
+ unaffected.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2001/01/05 00:30:00 JST
+ Tag: blume-20010105-primops
+ Description:
+ Added some experimental support for work that I am doing right now.
+ These changes mostly concern added primops, but there is also a new
+ experimental C library in the runtime system (but currently not enabled
+ anywhere except on Linux/X86).
+ In the course of adding primops (and playing with them), I discovered that
+ Zhong's INL_PRIM hack (no type info for certain primops) was, in fact, badly
+ broken.  (Zhong was very right he labeled this stuff as "major gross hack".)
+ To recover, I made type information in INL_PRIM mandatory and changed
+ prim.sml as well as built-in.sml accordingly.  The InLine structure now
+ has complete, correct type information (i.e., no bottom types).
+ Since all these changes mean that we need new binfiles, I also bumped the
+ version number to 110.32.1.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/30 22:10:00 JST
+ Tag: blume-20001230-various
+ Description:
+ Added proxy libraries for MLRISC and let MLRISC libraries refer
+ to each other using path anchors.  (See CM manual for explanation.)
+ Updated CM documentation.
+ Fixed some bugs in CM.
+ Implemented "proxy" libraries (= syntactic sugar for CM).
+ Added "-quiet" option to makeml and changed runtime system accordingly.
+ Added cleanup handler for exportML to reset timers and compiler stats.
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2000/12/22 22:22:58 EST 2000
+ Tag: Release_110_32
+ Description:
+         Infinite precision used throughout MLRISC.
+         see MLRISC/mltree/machine-int.sig
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/22 23:16:00 JST
+ Tag: blume-20001222-warn
+ Description:
+ Corrected wording and formatting of some CM warning message which I
+ broke in my previous patch.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/22 21:20:00 JST
+ Tag: blume-20001222-anchorenv
+ Description:
+ Fixed CM's handling of anchor environments in connection with CMB.make.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/22 13:15:00 JST
+ Tag: blume-20001222-cleanup
+ Description:
+ Removed src/cm/ffi which does not (and did not) belong here.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/21 23:55:00 JST
+ Tag: blume-20001221-exn
+ Description:
+ Probably most important: CM no longer silently swallows all exceptions
+ in the compiler.
+ Plus: some other minor CM changes.  For example, CM now reports some
+ sizes for generated binfiles (code, data, envpickle, lambdapickle).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/12/15 00:01:05 JST
+ Tag: blume-20001215-dirtool
+ Description:
+ - "dir" tool added.
+ - improvements and cleanup to Tools structure
+ - documentation updates
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: Thu Dec 14 03:45:24 EST 2000
+ Description:
+ Tag:  leunga-20001214-int-inf
+ Description:
+    In IntInf, added these standard functions, which are missing from our
+ implementation:
+     andb : int * int -> int
+     xorb : int * int -> int
+     orb  : int * int -> int
+     notb : int -> int
+      <<   : int * word -> int
+     ~>>  : int * word -> int
+    Not tested, I hope they are correct.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: Fri Dec  8 19:23:26 EST 2000
+ Description:
+ Tag:  leunga-20001208-nowhere
+ Description:
+   Slight improvements to the 'nowhere' tool to handle OR-patterns,
+ to generate better error messages etc.  Plus a brief manual.
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2000/12/08 09:54:02 EST 2000
+ Tag: Release_110_31
+ Description:
+ - Version 110.31
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: Thu Dec  7 22:01:04 EST 2000
+ Tag:  leunga-20001207-cell-monster-hack
+ Description:
+ Major MLRISC internal changes.  Affect all clients.
+ Summary:
+.  Type CELLS.cell = int is now replaced by a datatype.
+     As a result, the old regmap is now gone.  Almost all interfaces
+     in MLRISC change as a consequence.
+.  A new brand version of machine description tool (v3.0) that generates
+     modules expecting the new interface.  The old version is removed.
+.  The RA interface has been further abstracted into two new functors.
+     RISC_RA and X86RA.  These functors have much simpler interfaces.
+     [See also directory MLRISC/demo.]
+.  Some other new source->source code generation tools are available:
+     a. MLRISC/Tools/RewriteGen -- generate rewriters from rules.
+     b. MLRISC/Tools/WhereGen -- expands conditional pattern matching rules.
+        I use this tool to generate the peephole optimizers---with the new
+        cell type changes, peephole rules are becoming difficult to write
+        without conditional pattern matching.
+.  More Intmap -> IntHashTable change.  Previous changes by Matthias didn't
+     cover the entire MLRISC source tree so many things broke.
+.  CM files have been moved to the subdirectory MLRISC/cm.
+     They are moved because there are a lot of them and they clutter up the
+     root dir.
+.  More detailed documentation to come...
+     NOTE: To rebuild from 110.30 (ftp distribution), you'll have to do
+     a makeml -rebuild first.  This is because of other other
+     changes that Matthias has made (see below).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/30 23:12:00 JST
+ Tag: blume-20001130-filereorg
+ Description:
+ Some manual updates and some file reorganizations in CM.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/24 17:45:00 JST
+ Tag: blume-20001124-link
+ Description:
+ Drastically improved link traversal code for the case that the dynamic
+ value was already loaded at bootstrap time.  As a result, CM and CMB
+ now both load blazingly fast -- even on a very slow machine.  Also,
+ memory consumption has been further reduced by this.
+ Warning: The format of the PIDMAP file has changed.  THerefore, to
+ bootstrap you have to do this:
+. Run CMB.make
+. Make a symbolic link for the boot directory:
+      ln -s sml.boot.ARCH-OS xxx
+. "Rebuild" the boot directory:
+      ./makeml -boot xxx -rebuild sml ; rm xxx
+. Boot normally:
+       ./makeml
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/21 21:20:00 JST
+ Tag: blume-20001121-tools
+ Description:
+ Continued hacking on autoloading problem -- with success this time.
+ Also changed tool-plugin mechanism.  See new CM manual.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/19 14:30:00 JST
+ Tag:  blume-20001119-autoload
+ Description:
+ Some hacking to make autoloading faster.  Success for CMB, no success
+ so far for CM.  There is a reduced structure CM' that autoloads faster.
+ (This is a temporary, non-documented hack to be eliminated again when
+ the general problem is solved.)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/17 14:10:00 JST
+ Tag: blume-20001117-pickle-lib
+ Description:
+. Eliminated comp-lib.cm
+. Made pickle-lib.cm
+. Eliminated all uses of intset.sml (from comp-lib.cm)
+. Replaced all uses of intmap.{sig,sml} (from comp-lib.cm) with
+    equivalent constructs from smlnj-lib.cm (INtHashTable).
+. Point 4. also goes for those uses of intmap.* in MLRISC.
+    Duplicated intmap modules thrown out.
+. Hunted down all duplicated SCC code and replaced it with
+    equivalent stuff (GraphSCCFn from smlnj-lib.cm).
+. Rewrote Feedback module.
+. Moved sortedlist.sml into viscomp-lib.cm.  Eventually it
+    should be thrown out and equivalent modules from smlnj-lib.cm
+    should be used (IntRedBlackSet, IntListSet, ...).
+ Confirmed that compiler compiles to fixpoint.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/11/10 18:00:00
+ Tag: leunga-20001110-new-x86-fp
+ A new x86 floating point code generator has been added.
+ By default this is turned off.  To turn this on, do:
+     CM.autoload "$smlnj/compiler.cm";
+     Compiler.Control.MLRISC.getFlag "x86-fast-fp" := true;
+ Changes:
+.  Changed FTAN to FPTAN so that the assembly output is correct.
+.  Changed the extension callback for FTANGENT to generate:
+           fptan
+           fstp  %st(0)
+     instead of
+           fptan
+           fstpl ftempmem
+.  Numerous assembly fixes for x86.
+.  Cleaned up the machine code output module x86/x86MC.sml and added
+     support for a whole bunch of instructions and addressing modes:
+       fadd/fsub/fsubr/fmul/fdiv/fdivr  %st, %st(n)
+       faddp/fsubp/fsubrp/fmulp/fdivp/fdivrp  %st, %st(n)
+       fadd/fsub/fsubr/fmul/fdiv/fdivr  %st(n), %st
+       fiadd/fisub/fisubr/fimul/fidiv/fidivr mem
+       fxch %st(n)
+       fld %st(n)
+       fst %st(n)
+       fst mem
+       fstp %st(n)
+       fucom %st(n)
+       fucomp %st(n)
+     All these are now generated when the fast fp mode is turned on.
+.  Removed the dedicated registers %st(0), ..., %st(7) from X86CpsRegs
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/09 11:20:00 JST
+ Tag: blume-20001109-scc
+ Description:
+ Eliminated some code duplication:
+. Added "where" clause to GraphSCCFn in SML/NJ Library.
+    (Otherwise the functor is useless.)
+. Used GraphSCCFn where SCCUtilFun was used previously.
+. Got rid of SCCUtilFun (in comp-lib.cm).
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2000/11/06 09:02:21 EST 2000
+ Tag: Release_110_30
+ Description:
+ - Version 110.30
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/04 14:45:00
+ Tag: blume-20001104-mlbuild
+ Description:
+ - Made ml-build faster on startup.
+ - Documentation fixes.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/11/02 17:00:00 JST
+ Tag: blume-20001102-condcomp
+ Description:
+ - Small tweaks to pickler -- new BOOTFILES!
+ - Version bumped to 110.29.2.
+ - Added conditional compilation facility to init.cmi (see comment there).
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/10/23 19:31:00
+ Tag: leunga-20001023-demo-ra
+. Minor RA changes that improves spilling on x86 (affects Moby and C-- only)
+. Test programs for the graph library updated
+. Some new MLRISC demo programs added
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/08/31 22:15:00 JST
+ Tag: blume-20001017-errmsg
+ Description:
+ More error message grief: Where there used to be no messages, there
+ now were some that had bogus error regions.  Fixed.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/08/31 17:30:00 JST
+ Tag: blume-20001017-v110p29p1
+ Description:
+ I made a version 110.29.1 with new bootfiles.
+ Changes:  Modified pickler/unpickler for faster and leaner unpickling.
+           CM documentation changes and a small bugfix in CM's error reporting.
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2000/09/27 14:42:35 EDT
+ Tag: george-20000927-nodestatus
+ Description:
+ Changed the type of the nodestatus, so that:
+         SPILLED(~1)             is now SPILLED
+         SPILLED(m) where m>=0   is now MEMREG(m)
+         SPILLED(s) where s<~1   is now SPILL_LOC(~s)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/09/07 14:45:00 JST
+ Tag: blume-20000907-cmerrmsg
+ Description:
+ Small tweak to CM to avoid getting ML syntax error messages twice.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/08/31 18:00:00 JST
+ Tag: blume-20000831-cvsbootfiles
+ Description:
+ New URL for boot files (because the 110.29 files on the BL server do
+ now work correctly with my updated install scripts for yacc and lex).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/08/08 12:33:00 JST
+ Tag: blume-20000808-manual
+ Description:
+ Tiny update to CM manual.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/08/7 19:31:00
+ Tag: leunga-20000807-a-whole-bunch-of-stuff
+   Moby, C--, SSA, x86, machine descriptions etc.  Should only affect C--
+ and Mobdy.
+.  x86
+    a.  Fixes to peephole module by John and Dan.
+    b.  Assembly fix to SETcc by Allen.
+    c.  Fix to c-call by John.
+    d.  Fix to spilling by John.  (This one deals with the missing FSTPT case)
+    e.  Instruction selection optimization to SETcc as suggested by John.
+        For example,
+         MV(32, x, COND(32, CMP(32, LT, a, b), LI 1, LI 0))
+        should generate:
+         MOVL a, x
+         SUBL b, x
+         SHRL 31, x
+.  IR stuff
+      A bunch of new DJ-graph related algorithms added.  These
+      speed up SSA construction.
+.  SSA + Scheduling
+      Added code for SSA and scheduling to the repository
+ ----------------------------------------------------------------------
+ Name: Lal George
+ Date: 2000/07/27 11:53:14 EDT
+ Tag: lal-20000727-linux-ppc
+ Description:
+  Made changes to support Linux PPC.
+  p.s. I have confirmation that the 110.29 boot files work fine.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/27 17:40:00 JST
+ Tag: blume-20000727-scripts
+ Description:
+ !!!! WARNING !!!!
+ You must recompile the runtime system!
+ !!!! WARNING !!!!
+ This is basically another round of script-enhancements:
+. sml, ml-build, and ml-makedepend accept options -D and -U to define
+    and undefine CM preprocessor symbols.
+. ml-build avoids generating a new heap image if it finds that the
+    existing one is still ok.  (The condition is that no ML file had to
+    be recompiled and all ML files are found to be older that the heap
+    file.)
+    To make this work smoothly, I also hacked the runtime system as
+    well as SMLofNJ.SysInfo to get access to the heap image suffix
+    (.sparc-solaris, ...) that is currently being used.
+    Moreover, the signature of CM.mk_standalone has changed.  See the
+    CM manual.
+. ml-makedepend accepts additional options -n, -a, and -o.  (See the
+    CM manual for details.)
+. More CM manual updates:
+     - all of the above has been documented.
+     - there is now a section describing the (CM-related) command line
+       arguments that are accepted by the "sml" command
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/25 16:20:00 JST
+ Tag: blume-20000725-makedepend
+ Description:
+ Added a script called ml-makedepend.  This can be used in makefiles
+ for Unix' make in a way very similar to the "makedepend" command for
+ C.
+ The script internally uses function CM.sources.
+ Synopsis:
+     ml-makedepend [-f makefile] cmfile targetname
+ The default for the makefile is "makefile" (or "Makefile" should
+ "makefile" not exist).
+ ml-makedepend adds a cmfile/targetname-specific section to this
+ makefile (after removing the previous version of this section).  The
+ section contains a single dependency specification with targetname on
+ the LHS (targetname is an arbitrary name), and a list of files derived
+ from the cmfile on the RHS.  Some of the files on the RHS are
+ ARCH/OPSYS-specific.  Therefore, ml-makedepend inserts references to
+ "make" variables $(ARCH) and $(OPSYS) in place of the corresponding
+ path names.  The makefile writer is responsible for making sure that
+ these variables have correct at the time "make" is invoked.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/22 23:30:00 JST
+ Tag: blume-20000722-urlupdate
+ Description:
+ Changed BOOT and config/srcarchiveurl to point to BL server:
+     ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.29/
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/18 18:00:00 JST
+ Tag: blume-20000718-Version_110_29
+ Description:
+. Updated src/compiler/TopLevel/main/version.sml to version 110.29
+. Updated config/version to 110.29
+. Updated config/srcarchiveurl
+. New boot files!
+    ftp://ftp.cs.princeton.edu/pub/people/blume/sml/110.29-autofetch
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/11 13:58:00 JST
+ Tag: blume-20000711-doctypo
+ Description:
+ Fixed a few typos in CM manual.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/06/15 00:38:00
+ Tag: leunga-20000704-sparc-x86
+. x86 peephole improvement sp += k; sp -= k => nop  [from John]
+. fix to x86 RET bug [found by Dan Grossman]
+. sparc assembly bug fix for ticc instructions [found by Fermin]
+    Affects c-- and moby only
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/04 15:26:00
+ Tag: blume-20000704-trigger
+ Description:
+. Improvements to CM manual.
+. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
+    of getting a back-trace.  The function, when called, raises an
+    internal exception which explicitly carries the full back-trace history,
+    so it is unaffected by any intervening handle-raise pairs ("trivial"
+    or not).  The interactive loop will print that history once it arrives
+    at top level.
+    Short of having all exceptions implicitly carry the full history, the
+    recommended way of using this facility is:
+      - compile your program with instrumentation "on"
+      - run it, when it raises an exception, look at the history
+      - if the history is "cut off" because of some handler, go and modify
+        your program so that it explicitly calls BTrace.trigger
+      - recompile (still instrumented), and rerun; look at the full history
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/07/03 15:36:00 JST
+ Tag: blume-20000702-manual
+ Description:
+ Small corrections and updates to CM manual.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/29 16:04:00 JST
+ Tag: blume-20000629-yacctool
+ Description:
+ Changes:
+. Class "mlyacc" now takes separate arguments to pass options to
+    generated .sml- and .sig-files independently.
+. Corresponding CM manual updates.
+. BTrace module now also reports call sites.  (However, for loop clusters
+    it only shows from where the cluster was entered.)  There are associated
+    modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/27 16:51:00 JST
+ Tag: blume-20000627-noweb
+ Description:
+ Changes:
+. Implemented "subdir" and "witness" options for noweb tool.
+     This caused some slight internal changes in CM's tool implementation.
+. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
+     some remaining issues from earlier path anchor changes.
+. Updated CM manual accordingly.
+. Changed implementation of back-tracing so that I now consider it
+     ready for prime-time.
+     In particular, you don't have to explicitly trigger the back-trace
+     anymore.  Instead, if you are running BTrace-instrumented code and
+     there is an uncaught exception (regardless of whether or not it was
+     raised in instrumented code), the top-level evalloop will print
+     the back-trace.
+     Features:
+       - Instrumented and uninstrumented code work together seemlessly.
+         (Of course, uninstrumented code is never mentioned in actual
+          back-traces.)
+       - Asymptotic time- and space-complexity of instrumented code is
+         equal to that of uninstrumented code.  (This means that
+         tail-recursion is preserved by the instrumentation phase.)
+       - Modules whose code has been instrumented in different sessions
+         work together without problem.
+       - There is no penalty whatsoever on uninstrumented code.
+       - There is no penalty on "raise" expressions, even in
+         instrumented code.
+     A potential bug (or perhaps it is a feature, too):
+       A back-trace reaches no further than the outermost instrumented
+       non-trivial "raise".  Here, a "trivial" raise is one that is the
+       sole RHS of a "handle" rule.  Thus, back-traces reach trough
+            <exp> handle e => raise e
+       and even
+            <exp> handle Foo => raise Bar
+       and, of course, through
+            <exp> handle Foo => ...
+      if the exception was not Foo.
+      Back-traces always reach right through any un-instrumented code
+      including any of its "handle" expressions, trivial or not.
+    To try this out, do the following:
+      - Erase all existing binfiles for your program.
+        (You may keep binfiles for those modules where you think you
+         definitely don't need back-tracing.)
+      - Turn on back-trace instrumentation:
+           SMLofNJ.Internals.BTrace.mode (SOME true);
+      - Recompile your program.  (I.e., run "CM.make" or "use".)
+      - You may now turn instrumentation off again (if you want):
+           SMLofNJ.Internals.BTrace.mode (SOME false);
+      - Run your program as usual.  If it raises an exception that
+        reaches the interactive toplevel, then a back-trace will
+        automatically be printed.  After that, the toplevel loop
+        will print the exception history as usual.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/26 09:56:46 JST
+ Tag: blume-20000626-setup
+ Description:
+ CM: - setup-parameter to "sml" added; this can be used to run arbitrary
+       ML code before and after compiling a file (e.g., to set compiler
+       flags)
+ Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
+           - associated changes to btrace.sml (BTrace instrumentation pass)
+           - cleaner implementation of btimp.sml (BTrace tracing and report
+             module)
+ CM manual: * new path encoding documented
+            * description of setup-parameter to "sml" added
+ The biggest user-visible change to back-tracing is that it is no
+ longer necessary to compile all traced modules within the same
+ session.  (This was a real limitation.)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/24 12:40:00 JST
+ Tag: blume-20000624-startup
+ Description:
+ Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
+ bit too often -- to put it mildly. :)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/23 18:20:00 JST
+ Tag: blume-20000623-btrace
+ Description:
+ This updates adds a backtrace facility to aid programmers in debugging
+ their programs.  This involves the following changes:
+. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
+    keeping track of the current call stack.  When programs are compiled
+    in a special mode, the compiler will insert calls to these hooks
+    into the user program.
+    "Hook" means that it is possible for different implementations of
+    back-tracing to register themselves (at different times).
+. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
+    as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
+    It can be turned on using this call:
+      SMLofNJ.Internals.BTrace.mode (SOME true);
+    Turning it off again:
+      SMLofNJ.Internals.BTrace.mode (SOME false);
+    Querying the current status:
+      SMLofNJ.Internals.BTrace.mode NONE;
+    Annotated programs are about twice as big as normal ones, and they
+    run a factor of 2 to 4 slower with a dummy back-trace plugin (one
+    where all hooks do nothing).  The slowdown with a plugin that is
+    actually useful (such as the one supplied by default) is even greater,
+    but in the case of the default plugin it is still only an constant
+    factor (amortized).
+. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
+    with a sub-structure BTrace for controlling back-tracing.  In particular,
+    the above-mentioned function "mode" controls whether the annotation
+    phase is invoked by the compiler.  Another important function is
+    "trigger": when called it aborts the current execution and causes
+    the top-level loop to print a full back-trace.
+. compiler/MiscUtil/profile/btimp.sml is the current default plugin
+    for back-tracing.  It keeps track of the dynamic call stack and in
+    addition to that it keeps a partial history at each "level" of that
+    stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
+    d and b (at separate times, dynamically), then the report will show:
+    GOTO   d
+          /c
+    GOTO  \b
+    CALL   a
+    This shows that there was an initial non-tail call of a, then a
+    tail-call to b or c, looping behavior in a cluster of functions that
+    consist of b and c, and then a goto from that cluster (i.e., either from
+    b or from c) to d.
+    Note that (depending on the user program) the amount of information
+    that the back-trace module has to keep track of at each level is bounded
+    by a constant.  Thus, the whole implementation has the same asymptotical
+    complexity as the original program (both in space and in time).
+. compiler/TopLevel/interact/evalloop.sml has been modified to
+    handle the special exception SMLofNJ.Internals.BTrace.BTrace
+    which is raised by the "trigger" function mentioned above.
+ Notes on usage:
+ - Annotated code works well together with unannotated code:
+ Unannotated calls simply do not show up at all in the backtrace.
+ - It is not a good idea to let modules that were annotated during
+ different sessions run at the same time.  This is because the compiler
+ chooses small integers to identify individual functions, and there
+ will be clashes if different modules were compiled in separate sessions.
+ (Nothing will crash, and you will even be told about the clashes, but
+ back-trace information will in general not be useful.)
+ - Back-tracing can be confused by callcc and capture.
+ - The only way of getting a back-trace right now is to explicitly
+ invoke the "trigger" function from your user program.  Eventually, we
+ should make every exception carry back-trace information (if
+ available).  But since this creates more overhead at "raise"-time
+ (similar to the current exnHistory overhead), I have not yet
+ implemented this.  (The implementation will be rather easy.)  With
+ exceptions carrying back-trace information, this facility will be even
+ more useful because users don't need to modify their programs...
+ - While it is possible to compile the compiler with back-trace
+ annotations turned on (I did it to get some confidence in
+ correctness), you must make absolutely sure that core.sml and
+ btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
+ be compiled with annotation because there is no core access yet, but
+ if you compile btimp.sml with annotation, then the system will go into
+ an infinite recursion and crash.)
+ Since CM currently does not know about BTrace, the only way to turn
+ annotations on and off for different modules of the compiler is to
+ interrupt CMB.make, change the settings, and re-invoke it.  Of course,
+ this is awkward and clumsy.
+ Sample sessions:
+ Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
+ - SMLofNJ.Internals.BTrace.mode (SOME true);
+ [autoloading]
+ [autoloading done]
+ val it = false : bool
+ - structure X = struct
+ -     fun main n = let
+ -         fun a (x, 0) = d x
+ -           | a (x, n) = b (x, n - 1)
+ -         and b (x, n) = c (x, n)
+ -         and c (x, n) = a (x, n)
+ -         and d x = e (x, 3)
+ -         and e (x, 0) = f x
+ -           | e (x, n) = e (x, n - 1)
+ -         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
+ -           | f n = n * g (n - 1)
+ -         and g n = a (n, 3)
+ -     in
+ -         f n
+ -     end
+ - end;
+ structure X : sig val main : int -> int end
+ - X.main 3;
+ *** BACK-TRACE ***
+ GOTO   stdIn:4.2-13.20: X.main[2].f
+ GOTO-( stdIn:4.2-13.20: X.main[2].e
+ GOTO   stdIn:4.2-13.20: X.main[2].d
+      / stdIn:4.2-13.20: X.main[2].a
+      | stdIn:4.2-13.20: X.main[2].b
+ GOTO-\ stdIn:4.2-13.20: X.main[2].c
+ CALL   stdIn:4.2-13.20: X.main[2].g
+ GOTO   stdIn:4.2-13.20: X.main[2].f
+ GOTO-( stdIn:4.2-13.20: X.main[2].e
+ GOTO   stdIn:4.2-13.20: X.main[2].d
+      / stdIn:4.2-13.20: X.main[2].a
+      | stdIn:4.2-13.20: X.main[2].b
+ GOTO-\ stdIn:4.2-13.20: X.main[2].c
+ CALL   stdIn:4.2-13.20: X.main[2].g
+ GOTO   stdIn:4.2-13.20: X.main[2].f
+ GOTO-( stdIn:4.2-13.20: X.main[2].e
+ GOTO   stdIn:4.2-13.20: X.main[2].d
+      / stdIn:4.2-13.20: X.main[2].a
+      | stdIn:4.2-13.20: X.main[2].b
+ GOTO-\ stdIn:4.2-13.20: X.main[2].c
+ CALL   stdIn:4.2-13.20: X.main[2].g
+ GOTO   stdIn:4.2-13.20: X.main[2].f
+ CALL   stdIn:2.15-17.4: X.main[2]
+ -
+ (Note that because of a FLINt bug the above code currently does not
+ compile without BTrace turned on.)
+ Here is another example, using my modified Tiger compiler:
+ Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
+ - SMLofNJ.Internals.BTrace.mode (SOME true);
+ [autoloading]
+ [autoloading done]
+ val it = false : bool
+ - CM.make "sources.cm";
+ [autoloading]
+ ...
+ [autoloading done]
+ [scanning sources.cm]
+ [parsing (sources.cm):parse.sml]
+ [creating directory CM/SKEL ...]
+ [parsing (sources.cm):tiger.lex.sml]
+ ...
+ [wrote CM/sparc-unix/semant.sml]
+ [compiling (sources.cm):main.sml]
+ [wrote CM/sparc-unix/main.sml]
+ [New bindings added.]
+ val it = true : bool
+ - Main.compile ("../testcases/merge.tig", "foo.out");
+ *** BACK-TRACE ***
+ CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
+ CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
+ CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
+ GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
+ CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
+ CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
+ CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
+      / lib/semant.sml:411.65-543.8: SemantFun[2].transDec
+ CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
+ CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
+ CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
+ CALL   main.sml:1.18-118.4: Main.compile[2]
+ -
+ ----------------------------------------------------------------------
+ Name: Matthias Blumen
+ Date: 2000/06/21 18:00:00 JST
+ Tag: blume-20000621-manual
+ Description:
+ CM manual update: Path environments documented.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/19 13:40:00
+ Tag: blume-20000619-manual
+ Description:
+ CM manual and system/README update.  This only covers the fact that
+ there are no more implicit anchors.  (Path environments and the "bind"
+ option to "cm" have yet to be documented.)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/19 11:05:00 JST
+ Tag: blume-20000619-chdir-bugfix
+ Description:
+ Fixed a bug in new SrcPath module that sometimes led to a bad chDir call.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/18 22:00:10 JST
+ Tag: blume-20000618-implicit-anchors-really-gone
+ Description:
+ I updates the previous HISTORY entry where I forgot to mention that
+ implicit anchors are no longer with us.
+ The current update also gets rid of the (now useless) controller
+ CM.Control.implicit_anchors.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/16 17:30:00 JST
+ Tag: blume-20000616-anchorenv
+ Description:
+ This patch implements the long anticipated (just kidding :) "anchor
+ environment" mechanism.  In the course of doing this, I also
+ re-implemented CM's internal "SrcPath" module from scratch.  The new
+ one should be more robust in certain boundary cases.  In any case, it
+ is a lot cleaner than its predecessor (IMHO).
+ This time, although there is yet another boot file format change, I
+ kept the unpickler backward-compatible.  As a result, no new bootfiles
+ are necessary and bootstrapping is straightforward.  (You cannot read
+ new bootfiles into an old system, but the other way around is no
+ problem.)
+ Visible changes:
+ ** 0. Implicit path anchors (without the leading $-symbol) are no
+ longer recognized at all. This means that such path names are not
+ illegal either.  For example, the name basis.cm simply refers to a
+ local file called "basis.cm" (i.e, the name is an ordinary path
+ relative to .cm-files directory).  Or, to put it differently, only
+ names that start with $ are anchored paths.
+ ** 1. The $<singlearc> abbreviation for $/<singlearc> has finally
+ vanished.
+ John (Reppy) had critizised this as soon as I originally proposed and
+ implemented it, but at that time I did not really deeply believe
+ him. :) Now I came full-circle because I need the $<singlearc> syntax
+ in another place where it cannot be seen as an abbreviation for
+ $/<singlearc>.  To avoid the confusion, $<singlearc> now means what it
+ seems to mean (i.e., it "expands" into the corresponding anchor
+ value).
+ However, when paths are used as members in CM description files, it
+ continues to be true that there must be at least another arc after the
+ anchor.  This is now enforced separately during semantic analysis
+ (i.e., from a lexical/syntactical point of view, the notation is ok.)
+ ** 2. The "cm" class now accepts an option "bind".  The option's value
+ is a sub-option list of precisely two items -- one labeled "anchor"
+ and the other one labeled "value".  As you might expect, "anchor" is
+ used to specify an anchor name to be bound, and "value" specifies what
+ the anchor is being bound to.
+ The value must be a directory name and can be given in either standard
+ syntax (including the possibility that it is itself an anchored path)
+ or native syntax.
+ Examples:
+    foo.cm (bind:(anchor:bar value:$mystuff/bar))
+    lib.cm (bind:(anchor:a value:"H:\\x\\y\\z"))  (* only works under windows *)
+ and so on.
+ The meaning of this is that the .cm-file will be processed with an
+ augmented anchor environment where the given anchor(s) is/are bound to
+ the given values(s).
+ The rationale for having this feature is this: Suppose you are trying
+ to use two different (already stable) libraries a.cm and b.cm (that
+ you perhaps didn't write yourself).  Further, suppose each of these
+ two libraries internally uses its own auxiliary library $aux/lib.cm.
+ Normally you would now have a problem because the anchor "lib" can not
+ be bound to more than one value globally.  Therefore, the project that
+ uses both a.cm and b.cm must locally redirect the anchor to some other
+ place:
+    a.cm (bind:(anchor:lib value:/usr/lib/smlnj/a-stuff))
+    b.cm (bind:(anchor:lib value:/usr/lib/smlnj/b-stuff))
+ This hard-wires $lib/aux.cm to /usr/lib/smlnj/a-stuff/aux.cm or
+ /usr/lib/smlnj/b-stuff/aux.cm, respectively.
+ Hard-wiring path names is a bit inflexible (and CM will verbosely warn
+ you when you do so at the time of CM.stabilize).  Therefore, you can
+ also use an anchored path as the value:
+   a.cm (bind:(anchor:lib value:$a-lib))
+   b.cm (bind:(anchor:lib value:$b-lib))
+ Now you can globally configure (using the usual CM.Anchor.anchor or
+ pathconfig machinery) bindings for "a-lib" and "b-lib".  Since "lib"
+ itself is always locally bound, setting it globally is no longer
+ meaningful or necessary (but it does not hurt either).  In fact, "lib"
+ can still be used as a global anchor for separate purposes.  As a
+ matter of fact, one can locally define "lib" in terms of a global
+ "lib":
+   a.cm (bind:(anchor:lib value:$lib/a))
+   b.cm (bind:(anchor:lib value:$lib/b))
+ ** 3: The encoding of path names has changed.  This affects the way
+ path names are shown in CM's progress report and also the internal
+ protocol encoding used for parallel make.
+ The encoding now uses one or more ':'-separated segments.  Each
+ segments corresponds to a file that has been specified relative to the
+ file given by its preceding segment.  The first segment is either
+ relative to the CWD, absolute, or anchored.  Each segment itself is
+ basically a Unix pathname; all segments but the first are relative.
+ Example:
+    $foo/bar/baz.cm:a/b/c.sml
+ This path denotes the file bar/a/b/c.sml relative to the directory
+ denoted by anchor "foo".  Notice that the encoding also includes
+ baz.cm which is the .cm-file that listed a/b/c.sml.  As usual, such
+ paths are resolved relative to the .cm-files directory, so baz.cm must
+ be ignored to get the "real" pathname.
+ To make this fact more obvious, CM puts the names of such "virtual
+ arcs" into parentheses when they appear in progress reports. (No
+ parentheses will appear in the internal protocol encoding.)  Thus,
+ what you really see is:
+   $foo/bar/(baz.cm):a/b/c.sml
+ I find this notation to be much more informative than before.
+ Another new feature of the encoding is that special characters
+ including parentheses, colons, (back)slashes, and white space are
+ written as \ddd (where ddd is the decimal encoding of the character).
+ *** The CM manual still needs to be updated.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/06/15 00:38:00
+ Tag: leunga-20000615-x86-peephole
+ x86 Peephole fix by Fermin.  Affects c-- and moby only.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/12 11:40:00
+ Tag: blume-20000612-parmakefix
+ Description:
+ More cleanup after changing the file naming scheme: This time I
+ repaired the parallel make mechanism for CMB.make which I broke earlier.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/06/09 01:25:00
+ Tag: leunga-20000609-various
+ None of these things should affect normal SML/NJ operations
+. Peephole improvements provided by Fermin (c--)
+. New annotation DEFUSE for adding extra dependence (moby)
+. New X86 LOCK instructions (moby)
+. New machine description language for reservation tables (scheduling)
+. Fixes to various optimization/analysis modules (branch chaining, dominator
+    trees etc.)
+. I've changed the CM files so that they can work with versions
+.0.6, 110.25 and 110.28
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/09 12:40:00
+ Tag: blume-20000609-log
+ Description:
+ - Removed all(?) remaining RCS Log entries from sources.
+ - Fixed bug in ml-yacc and ml-lex sources (use explicit anchors for
+   anchored paths).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/07 17:00:00 JST
+ Tag: blume-20000607-no-implicit-anchors
+ Description:
+. This update changes the default setting for
+ CM.Control.implicit_anchors from true to false.  This means that
+ implicit anchors are no longer permitted by default.  I also tried to
+ make sure that nothing else still relies on implicit anchors.
+ (This is the next step on the schedule towards a CM that does not even
+ have the notion of implicit anchors anymore.)
+. More CM manual updates.
+. I managed to track down and fix the pickling bug I mentioned last
+ time.  Because of the previously existing workaround, this entails no
+ immediate practical changes.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/06/06 11:15:00 JST
+ Tag: blume-20000606-lazierpickle
+ Description:
+ !!!! NEW BOOT FILES !!!!
+ * The main purpose of this update is to make library pickles lazier in
+ order to reduce the initial space penalty for autoloading a library.
+ As a result, it is now possible to have $smlnj/compiler.cm
+ pre-registered.  This should take care of the many complaints or
+ inquiries about missing structure Compiler.  This required changes to
+ CM's internal data structures and small tweaks to some algorithms.
+ As a neat additional effect, it is no longer necessary (for the sake
+ of lean heap image files) to distinguish between a "minimal" CM and a
+ "full" CM.  Now, there is only one CM (i.e., the "full" version:
+ $smlnj/cm.cm aka $smlnj/cm/full.cm), and it is always available at the
+ interactive top level. ($smlnj/cm/minimal.cm is gone.)
+ To make the life of compiler-hackers easier, "makeml" now also
+ pre-registers $smlnj/cmb.cm (aka $smlnj/cmb/current.cm).  In other
+ words, after you bootstrap a new sml for the first time, you will not
+ have to autoload $smlnj/cmb.cm again afterwards.  (The first time
+ around you will still have to do it, though.)
+ * A second change consists of major updates to the CM manual.  There
+ are now several appendices with summary information and also a full
+ specification of the CM description file syntax.
+ * In directory src/system I added the script "allcross".  This script
+ invokes sml and cross-compiles the compiler for all supported
+ architectures.  (Useful when providing a new set of boot files.)
+ * There seems to be a latent bug in my "lazy pickles" mechanism.  I
+ added a small tweak to pickle-util.sml to work around this problem,
+ but it is not a proper fix yet.  I will investigate further.  (The
+ effect of the bug was an inflation of library pickle size.)
+ * Version number increased to 110.28.1 (to avoid compatibility problems).
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/05/25 17:28 EDT
+ Tag: leunga-20000525-ra
+ Description:
+   Fixed a bug in freezing phase of the register allocator.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/05/15 22:53 EDT
+ Tag: leunga-20000515-alpha-x86-ra
+ Description:
+. Alpha
+       Slight cleanup.  Removed the instruction SGNXL
+. X86
+       Added the following instructions to the instruction set:
+         ROLx, RORx,
+         BTx, BTSx, BTLx, BTRx,
+         XCHGx, and variants with the LOCK prefix
+. Register Allocation
+       The module ra-rewrite-with-renaming has been improved.
+   These have no effect on SML/NJ.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/05/15 16:20:00 JST
+ Tag: blume-20000515-lightrebuild
+ Description:
+. I added an alternative to "-rebuild" to "makeml".  The difference is
+    that prior to calling CMB.make' the CM-variable "LIGHT" will be
+    defined.  In effect, the command will not build any cross-compiler
+    backends and therefore finish more quickly.
+    The "fixpt" script also takes a "-light" switch to be able to use
+    this new facility while compiling for a fixpoint.
+. I replaced all mentions of anchored paths in group owner specifications
+    with simple relative paths (usually starting with "..").
+    The rationale is that a library's internal workings should not be
+    compromised by the lack of some anchor.  (An anchor is necessary
+    for someone who wants to refer to the library by an anchored path,
+    but it should not be necessary to build the same library in the first
+    place.)
+. I changed the way CM's tool mechanism determines the shell command
+    string used for things like ml-yacc etc. so that it does not break
+    when CM.Control.implicit_anchors is turned off.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/05/12 18:20:00 JST
+ Tag: blume-20000512-ml-build
+ Description:
+ Fixed a bug in config/_ml-build that prevented ml-yacc and ml-lex from
+ getting installed properly (by config/install.sh).
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/05/12 17:30:00 JST
+ Tag: blume-20000512-anchors
+ Description:
+ !!! NEW BOOT FILES !!!
+ This change is in preparation of fading out support for "implicitly
+ anchored path names".  I went through all sources and used the
+ explicit (and relatively new) $-notation.  See system/README and the
+ CM manual for more info on this.
+ I also modified the anchoring scheme for some things such as "smlnj",
+ "MLRISC", "cm", etc. to take advantage of the fact that explicit
+ anchors are more expressive: anchor name and first arc do not have to
+ coincide.  This entails the following user-visible change:
+ You have to write $smlnj/foo/bar instead of smlnj/foo/bar.  In
+ particular, when you fire up sml with a command-line argument, say,
+ e.g.:
+    sml '$smlnj/cmb.cm'
+ At the ML toplevel prompt:
+    CM.autoload "$smlnj/cmb.cm";
+ There is also a new controller in CM.Control that can be used to turn
+ off all remaining support for implicit anchors by saying:
+     CM.autoload "$smlnj/
+     #set CM.Control.implicit_anchors false;
+ This causes CM to reject implicitly anchored paths.  This is (for the
+ time being) less permissive than the "final" version where there will
+ be no more such implicit anchors and relative paths will be just that:
+ relative.
+ The next step (version after next version?) will be to make the
+ default for CM.Control.implicit_anchors false.  After the dust has
+ settled, I can then produce the "final" version of this...
+ Note: Since bootstrapping is a bit tricky, I provided new boot files.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/05/11 16:30:00 JST
+ Tag: blume-20000511-sources
+ Description:
+ The main change is that I added function CM.sources as a generalized
+ version of the earlier CM.makedepend.  This entails the following
+ additional changes:
+   - CM.makedepend has been dropped.
+   - CM manual has been updated.
+   - TOOLS signature and API have been changed.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/05/10 21:17 EDT
+ Tag: leunga-20000510-moby-c--ssa
+ Description:
+   Various bug fixes and new features for C--, Moby and MLRISC optimizations.
+ None of these affect SML/NJ.
+. Register Allocation
+     a. A new ra spilling module (ra/ra-spill-with-renaming) is implemented.
+        This module tries to remove local (i.e. basic block level) redundancies
+        during spilling.
+     b. A new framework for performing region based register allocation.
+        Not yet entirely functional.
+. X86
+    a. DefUse for POP was missing the stack pointer [found by Lal]
+    b. Reload for CALL was incorrect in X86Spill [found by John]
+    c. Various fixes in X86Spill so that it can be used correctly for
+       the new spilling module.
+. SSA/IR
+    a. New module ir/dj-dataflow.sml implements elimination based
+       data flow analysis.
+. MLRiscGen
+    a. Fix for gc type annotation
+. MDGen
+    Various fixes for machine description -> ml code translation.  For ssa
+    only.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/05/08 22:17 EDT
+ Tag: leunga-20000508-labexp
+ Description:
+   Fermin has found a few assembly problems with constant expressions
+   generated in LabelExp.  Mostly, the problems involve extra parentheses,
+   which choke on dumb assemblers.  This is his fix.
+ ----------------------------------------------------------------------
+ Name: Dave MacQueen
+ Date: 2000/04/09 14:00 EDT
+ Tag: dbm-20000502-Version_110_28
+ Description:
+. Updated src/compiler/TopLevel/main/version.sml to version 110.28
+. Updated config/version to 110.28
+. Updated config/srcarchiveurl
+. New boot files!
+    ftp://ftp.research.bell-labs.com/dist/smlnj/working/110.28/
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/05/01 19:05:00 JST
+ Tag: blume-20000501-noweb
+ Description:
+ A new noweb tool has been added.  The existing system is entirely
+ unaffected by this, but some CM users have asked for renewed noweb
+ support.  Everything is documented in the CM manual.
+ New (plugin) libraries:
+    noweb-tool.cm
+    nw-ext.cm
+ ----------------------------------------------------------------------
+ Name: Dave MacQueen
+ Date: 2000/04/30 12:40PM EDT
+ Tag: dbm-20000430-bug_fixes
+ Description:
+. Fix for bug 1498
+    smlnj/src/system/Basis/Implementation/Unsafe/object.sig
+    smlnj/src/system/Basis/Implementation/Unsafe/object.sml
+      added toRealArray function
+    smlnj/src/compiler/MiscUtil/print/ppobj.sml
+      added check for tag Obj.RealArray to array printing case in ppObj
+. Fix for bug 1510
+    smlnj/src/compiler/Semant/types/typesutil.sml
+      fixed definition of dummyargs (used by equalTycon) so that
+      dummy args are distinct types
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/30 01:00:00 JST
+ Tag: blume-20000430-versions
+ Description:
+. CM version numbering added.  This is an implementation of Lal's
+    proposal for adding version numbers and version checking to .cm
+    files.  Lal said that his proposal was just that -- a proposal.
+    For the time being I went ahead and implemented it so that people
+    can comment on it.  Everything is completely backward-compatible
+    (except for the stable library format, i.e., new bootfiles!).
+    As usual, see the CM manual for details.
+. An alternative syntax for anchored paths has been implemented.
+    Dave has recently voiced the same concerns that I had when I did
+    this, so there should be some support.  My take is that eventually
+    I will let support for the current syntax (where anchors are
+    "implicit") fade out in favor of the new, explicit syntax.
+    In order to be backward-compatible, both old and new syntax are
+    currently supported.
+    Again, see the CM manual for details.
+. Parallel make is trying to be slightly smarter:  When the master
+    process finds a "bottleneck", i.e., when there is only one
+    compilation unit that can be compiled and everybody else is
+    waiting on it, then it will simply compile it directly instead
+    of clumsily telling one of the slaves to do it.
+. Support for "unsharing" added.  This is necessary in order to be
+    able to have two different versions of the same library running
+    at the same time (e.g., for trying out a new MLRISC while still
+    having the old MLRISC linked into the current compiler, etc.)
+    See the CM manual.
+. Simple "makedepend" functionality added for generating Makefile
+    dependency information.  (This is rather crude at the moment.
+    Expect some changes here in the future.)
+. ".fun" added as a recognized suffix for ML files. Also documented
+    explicitly in the manual that the fallback behavior (unknown suffix
+    -> ML file) is not an official feature!
+. Small changes to the pickler for stable libraries.
+. Several internal changes to CM (for cleanup/improvement).
+ !!!! NEW BINFILES !!!!
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/28 17:30:00 JST
+ Tag: blume-20000428-pathconfig
+ Description:
+. I changed config/install.sh to remove duplicate entries from the
+    lib/pathconfig file at the end.  Moreover, the final version of
+    lib/pathconfig is sorted alphabetically.  The same (sorting) is done
+    in src/system/installml.
+. The config/install.sh script now consistently uses relative
+    pathnames in lib/pathconfig whenever the anchor is in the lib
+    directory.  (So far this was true for the libraries that come
+    pre-compiled and bundled as part of the bootfiles but not for
+    libraries that are compiled by the script itself.)
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/26 13:10:00 JST
+ Tag: blume-20000426-fun_suffix
+ Description:
+ Added ".fun" as a recognized file name suffix (for ML code).
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/25 17:00:00 EST
+ Tag: leunga-20000425-alpha-ra
+ Description:
+. Alpha
+     PSEUDOARITH was missing in AlphaRewrite.  This causes an endless loop
+ in C--.
+. RA
+    Added a flag "ra-dump-size" to print out the size of the flowgraph
+    and the interference graph.
+ ----------------------------------------------------------------------
+ Name: Dave MacQueen
+ Date: 2000/04/25/
+ Tag: dbm-20000425-mlyacc_doc_examples
+ Description:
+   Updated mlyacc.tex sections 5 and 7 for SML '97 and CM.
+   Updated all three examples in src/ml-yacc/examples to run
+   under 110.* using CM.make.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/20 23:04:00 EST
+ Tag: leunga-20000420-ssa-c---stuff
+ Description:
+   This update synchronizes my repository with Yale's.  Most of these
+ changes, however, do not affect SML/NJ at all (the RA is an exception).
+. Register Allocator
+    a. An improvement in the interference graph construction:
+       Given a copy
+             s <- t
+       no interference edge between s and t is added for this definition of s.
+    b. I've added two new spill heuristic modules that Fermin and I developed
+       (in the new library RA.cm). These are unused in SML/NJ but maybe
+       useful for others (Moby?)
+. X86
+    a. Various fixes in the backend provided by Fermin [C--] and Lal.
+. Alpha
+    a. Added the BSR instruction and code generation that goes with it [C--]
+    b. Other fixes too numerous to recount provided by Fermin [C--]
+. Regmaps
+    a. The regmaps are not initialized with the identity physical bindings
+       at creation time.  This is unneeded.
+. MLRISC Optimizations
+    a. The DJ-Graph module can now compute the iterated dominance frontiers
+       intersects with liveness incrementally in linear time! Woohoo!
+       This is now used in my new SSA construction algorithm.
+    b. THe branch reorganization module is now smarter about linear chains of
+       basic blocks.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/12 13:52:00 JST
+ Tag: blume_main_v110p27_1
+ Description:
+ Changed install.sh script to handle archive files without version number
+ and to use "boot.<arch>-<os>" instead of "sml.boot.<arch>-<os>" for the
+ name of the boot file archive.
+ ----------------------------------------------------------------------
+ Name: Dave MacQueen
+ Date: 2000/04/09 14:00 EDT
+ Tag: dbm-20000410-Version_110_27
+ Description:
+. Updated src/compiler/TopLevel/main/version.sml to version 110.27
+. Updated src/config/version to 110.27
+. New boot files!
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/09 19:09:00 EST
+ Tag: leunga-20000409-misc
+ Description:
+.  Yet another fix for x86 assembly for idivl, imull, mull and friends.
+.  Miscellaneous improvements to MLRISC (unused in sml/nj)
+ ----------------------------------------------------------------------
+ Name: Stefan
+ Date: 2000/04/07 10:00:00 EDT
+ Tag: monnier-20000406-branch-handling
+ Description:
+ Improved handling of branches (mostly those generated from
+ polymorphic equality), removed switchoff and changed the
+ default optimization settings (more cpsopt and less flintopt).
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/06 01:30:00 EST
+ Tag: leunga-20000406-peephole-x86-SSA-2
+ Description:
+    Forgot a few files.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/06 00:36:00 EST
+ Tag: leunga-20000406-peephole-x86-SSA
+ Description:
+.  New Peephole code
+.  Minor improvement to X86 instruction selection
+.  Various fixes to SSA and machine description -> code translator
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/05 12:30:00 JST
+ Tag: blume_main_v110p26p2_3
+ Description:
+ This update just merges three minor cosmetic updates to CM's sources
+ to get ready for the 110.27 code freeze on Friday.  No functionality
+ has changed.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/04 19:39:00 EST
+ Tag: leunga-20000404-x86-asm
+ Description:
+.  Fixed a problem in X86 assembly.
+     Things like
+        jmp %eax
+        jmp (%eax)
+     should be output as
+        jmp *%eax
+        jmp *(%eax)
+.  Assembly output
+       Added a new flag
+           "asm-indent-copies" (default to false)
+       When this flag is on, parallel copies will be indented an extra level.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/04 03:18:00 EST
+ Tag: leunga-20000404-C--Moby
+ Description:
+     All of these fixes are related to C--, Moby, and my own optimization
+     stuff; so they shouldn't affect SML/NJ.
+.  X86
+     Various fixes related floating point, and extensions.
+.  Alpha
+     Some extra patterns related to loads with signed/zero extension
+     provided by Fermin.
+.  Assembly
+     When generating assembly, resolve the value of client defined constants,
+     instead of generating symbolic values.  This is controlled by the
+     new flag "asm-resolve-constants", which is default to true.
+.  Machine Descriptions
+     a. The precedence parser was slightly broken when parsing infixr symbols.
+     b. The type generalizing code had the bound variables reversed, resulting
+        in a problem during arity raising.
+     c. Various fixes in machine descriptions.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/04/03 16:05:00 JST
+ Tag: blume_main_v110p26p2_2
+ Description:
+ I eliminated coreEnv from compInfo.  Access to the "Core" structure is
+ now done via the ordinary static environment that is context to each
+ compilation unit.
+ To this end, I arranged that instead of "structure Core" as "structure
+ _Core" is bound in the pervasive environment.  Core access is done via
+ _Core (which can never be accidentally rebound because _Core is not a
+ legal surface-syntax symbol).
+ The current solution is much cleaner because the core environment is
+ now simply part of the pervasive environment which is part of every
+ compilation unit's context anyway.  In particular, this eliminates all
+ special-case handling that was necessary until now in order to deal
+ with dynamic and symbolic parts of the core environment.
+ Remaining hackery (to bind the "magic" symbol _Core) is localized in the
+ compilation manager's bootstrap compiler (actually: in the "init group"
+ handling).  See the comments in src/system/smlnj/init/init.cmi for
+ more details.
+ I also tried to track down all mentions of "Core" (as string argument
+ to Symbol.strSymbol) in the compiler and replaced them with a
+ reference to the new CoreSym.coreSym.  Seems cleaner since the actual
+ name appears in one place only.
+ Binfile and bootfile format have not changed, but the switchover from
+ the old "init.cmi" to the new one is a bit tricky, so I supplied new
+ bootfiles anyway.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/04/02 21:17:00 EST
+ Tag: leunga-20000402-mltree
+ Description:
+. Renamed the constructor CALL in MLTREE by popular demand.
+. Added a bunch of files from my repository.  These are currently
+       used by other non-SMLNJ backends.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/03/31 21:15:00 EST
+ Tag: leunga-20000331-aliasing
+ Description:
+ This update contains a rewritten (and hopefully more correct) module
+ for extracting aliasing information from CPS.
+    To turn on this feature:
+         Compiler.Control.CG.memDisambiguate := true
+    To pretty print the region information with assembly
+        Compiler.Control.MLRISC.getFlag "asm-show-region" := true;
+    To control how many levels of aliasing information are printed, use:
+        Compiler.Control.MLRISC.getInt "points-to-show-level" := n
+    The default of n is 3.
+ ----------------------------------------------------------------------
+ Name: David MacQueen
+ Date: 2000/03/31 11:15:00 EST
+ Tag: dbm-20000331-runtime_fix
+ Description:
+ This update contains:
+. runtime/c-lib/c-libraries.c
+    includes added in revision 1.2 caused compilation errors on hppa-hpux
+. fix for bug 1556
+    system/Basis/Implementation/NJ/internal-signals.sml
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/03/31 18:00:00 JST
+ Tag: blume_main_v110p26p2_1
+ Description:
+ This update contains:
+. A small change to CM's handling of stable libraries:
+    CM now maintains one "global" modmap that is used for all stable
+    libraries.  The use of such a global modmap maximizes sharing and
+    minimizes the need for re-traversing parts of environments during
+    modmap construction.  (However, this has minor impact since modmap
+    construction seems to account for just one percent or less of total
+    compile time.)
+. I added a "genmap" phase to the statistics.  This is where I got the
+    "one percent" number (see above).
+. CM's new tool parameter mechanism just became _even_ better. :)
+    - The parser understands named parameters and recursive options.
+    - The "make" and "shell" tools use these new features.
+      (This makes it a lot easier to cascade these tools.)
+    - There is a small syntax change: named parameters use a
+        <name> : ( <option> ... )            or
+        <name> : <string>
+      syntax.  Previously, named parameters were implemented in an
+      ad-hoc fashion by each tool individually (by parsing strings)
+      and had the form
+        <name>=<string>
+    See the CM manual for a full description of these issues.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/03/30 18:00:00 JST
+ Tag: blume_main_v110p26p2_0
+ Description:
+ !!!!! WARNING !!!!!!
+ !!  New binfiles  !!
+ !!!!!!!!!!!!!!!!!!!!
+ This update contains:
+. Moderate changes to CM:
+    - Changes to CM's tools mechanism.  In particular, it is now possible
+    to have tools that accept additional "command line" parameters
+    (specified in the .cm file at each instance where the tool's class is
+    used).
+    This was done to accommodate the new "make" and "shell" tools which
+    facilitate fairly seamless hookup to portions of code managed using
+    Makefiles or Shell scripts.
+    There are no classes "shared" or "private" anymore.  Instead, the
+    sharing annotation is now a parameter to the "sml" class.
+    There is a bit of generic machinery for implementing one's own
+    tools that accept command-line parameters.  However, I am not yet fully
+    satisfied with that part, so expect changes here in the future.
+    All existing tools are described in the CM manual.
+    - Slightly better error handling.  (CM now suppresses many followup
+    error messages that tended to be more annoying than helpful.)
+. Major changes to the compiler's static environment data structures.
+    - no CMStaticEnv anymore.
+         - no CMEnv, no "BareEnvironment" (actually, _only_ BareEnvironment,
+           but it is called Environment), no conversions between different
+           kinds of static environments
+    - There is still a notion of a "modmap", but such modmaps are generated
+      on demand at the time when they are needed.  This sounds slow, but I
+      sped up the code that generates modmaps enough for this not to lead to
+      a slowdown of the compiler (at least I didn't detect any).
+    - To facilitate rapid modmap generation, static environments now
+      contain an (optional) "modtree" structure.  Modtree annotations are
+      constructed by the unpickler during unpickling.  (This means that
+      the elaborator does not have to worry about modtrees at all.)
+      Modtrees have the advantage that they are compositional in the same
+      way as the environment data structure itself is compositional.
+      As a result, modtrees never hang on to parts of an environment that
+      has already been rendered "stale" by filtering or rebinding.
+    - I went through many, many trials and errors before arriving at the
+      current solution.  (The initial idea of "linkpaths" did not work.)
+      But the result of all this is that I have touched a lot of files that
+      depend on the "modules" and "types" data structures (most of the
+      elaborator). There were a lot of changes during my "linkpath" trials
+      that could have been reverted to their original state but weren't.
+      Please, don't be too harsh on me for messing with this code a bit more
+      than what was strictly necessary...  (I _did_ resist the tempation
+      of doing any "global reformatting" to avoid an untimely death at
+      Dave's hands. :)
+    - One positive aspect of the previous point:  At least I made sure that
+      all files that I touched now compile without warnings (other than
+      "polyEqual").
+    - compiler now tends to run "leaner" (i.e., ties up less memory in
+      redundant modmaps)
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/03/29 18:00:00
+ Tag: leunga-20000327-mlriscGen_hppa_alpha_x86
+ Boot files (optional): ftp://react-ilp.cs.nyu.edu/leunga/110.26.1-sml.boot.x86-unix-20000330.tar.gz
+ Description:
+    This update contains *MAJOR* changes to the way code is generated from CPS
+ in the module mlriscGen, and in various backend modules.
+ CHANGES
+ =======
+. MLRiscGen: forward propagation fix.
+    There was a bug in forward propagation introduced at about the same time
+    as the MLRISC x86 backend, which prohibits coalescing to be
+    performed effectively in loops.
+    Effect: speed up of loops in RISC architectures.
+            By itself, this actually slowed down certain benchmarks on the x86.
+. MLRiscGen:  forward propagating addresses from consing.
+    I've changed the way consing code is generated.  Basically I separated
+    out the initialization part:
+         store tag,   offset(allocptr)
+         store elem1, offset+4(allocptr)
+         store elem2, offset+8(allocptr)
+         ...
+         store elemn, offset+4n(allocptr)
+    and the address computation part:
+         celladdr <- offset+4+alloctpr
+    and move the address computation part
+    Effect:  register pressure is generally lower as a result.  This
+             makes compilation of certain expressions much faster, such as
+             long lists with non-trivial elements.
+              [(0,0), (0,0), .... (0,0)]
+. MLRiscGen: base pointer elimination.
+     As part of the linkage mechanism, we generate the sequence:
+      L:  ...  <- start of the code fragment
+      L1:
+          base pointer <- linkreg - L1 + L
+      The base pointer was then used for computing relocatable addresses
+    in the code fragment.  Frequently (such as in lots of continuations)
+    this is not needed.  We now eliminate this sequence whenever possible.
+      For compile time efficiency, I'm using a very stupid local heuristic.
+    But in general, this should be done as a control flow analysis.
+    Effect:  Smaller code size.  Speed up of most programs.
+. Hppa back end
+      Long jumps in span dependence resolution used to depend on the existence
+   of the base pointer.
+      A jump to a long label L was expanded into the following sequence:
+       LDIL %hi(L-8192), %r29
+       LDO  %lo(L-8192)(%r29), %r29
+       ADD  %r29, baseptr, %r29
+       BV,n %r0(%r29)
+      In the presence of change (3) above, this will not work.  I've changed
+    it so that the following sequence of instructions are generated, which
+    doesn't mention the base pointer at all:
+          BL,n  L', %r29           /* branch and link, L' + 4 -> %r29 */
+     L':  ADDIL L-(L'+4), %r29     /* Compute address of L */
+          BV,n  %r0(%r29)          /* Jump */
+. Alpha back end
+       New alpha instructions LDB/LDW have been added, as per Fermin's
+    suggestions.   This is unrelated to all other changes.
+. X86 back end
+      I've changed andl to testl in the floating point test sequence
+      whenever appropriate.  The Intel optimization guide states that
+      testl is preferable to andl.
+. RA (x86 only)
+      I've improved the spill propagation algorithm, using an approximation
+    of maximal weighted independent sets.   This seems to be necessary to
+    alleviate the negative effect in light of the slow down in (1).
+      I'll write down the algorithm one of these days.
+. MLRiscGen: frequencies
+      I've added an annotation that states that all call gc blocks have zero
+    execution frequencies.  This improves register allocation on the x86.
+ BENCHMARKS
+ ==========
+    I've only perform the comparison on 110.25.
+    The platforms are:
+     HPPA  A four processor HP machine (E9000) with 5G of memory.
+     X86   A 300Hhz Pentium II with 128M of memory, and
+     SPARC An Ultra sparc 2 with 512M of memory.
+    I used the following parameters for the SML benchmarks:
+              @SMLalloc
+      HPPA    256k
+      SPARC   512k
+      X86     256k
+ COMPILATION TIME
+ ----------------
+    Here are the numbers comparing the compilation times of the compilers.
+    I've only compared 110.25 compiling the new sources versus
+    a fixpoint version of the new compiler compiling the same.
+.25                                  New
+            Total  Time in RA  Spill+Reload   Total  Time In RA Spill+Reload
+      HPPA   627s    116s        2684+3584     599s    95s       1003+1879
+      SPARC  892s    173s        2891+3870     708s    116s      1004+1880
+      X86    999s    315s       94006+130691   987s    296s    108877+141957
+.25         New
+             Code Size      Code Size
+      HPPA   8596736         8561421
+      SPARC  8974299         8785143
+      X86    9029180         8716783
+    So in summary, things are at least as good as before.   Dramatic
+    reduction in compilation is obtained on the Sparc; I can't explain it,
+    but it is reproducible.  Perhaps someone should try to reproduce this
+    on their own machines.
+ SML BENCHMARKS
+ --------------
+     On the average, all benchmarks perform at least as well as before.
+       HPPA         Compilation Time     Spill+Reload      Run Time
+.25  New            110.25    New   110.25  New
+       barnesHut  3.158  3.015  4.75%    1+1       0+0   2.980  2.922   2.00%
+           boyer  6.152  5.708  7.77%    0+0       0+0   0.218  0.213   2.34%
+    count-graphs  1.168  1.120  4.32%    0+0       0+0  22.705 23.073  -1.60%
+             fft  0.877  0.792 10.74%    1+3       1+3   0.602  0.587   2.56%
+     knuthBendix  3.180  2.857 11.32%    0+0       0+0   0.675  0.662   2.02%
+          lexgen  6.190  5.290 17.01%    0+0       0+0   0.913  0.788  15.86%
+            life  0.803  0.703 14.22%   25+25      0+0   0.153  0.140   9.52%
+           logic  2.048  2.007  2.08%    6+6       1+1   4.133  4.008   3.12%
+      mandelbrot  0.077  0.080 -4.17%    0+0       0+0   0.765  0.712   7.49%
+          mlyacc 22.932 20.937  9.53%  154+181    32+57  0.468  0.430   8.91%
+         nucleic  5.183  5.060  2.44%    2+2       0+0   0.125  0.120   4.17%
+   ratio-regions  3.357  3.142  6.84%    0+0       0+0  116.225 113.173 2.70%
+             ray  1.283  1.290 -0.52%    0+0       0+0   2.887  2.855   1.11%
+          simple  6.307  6.032  4.56%   28+30      5+7   3.705  3.658   1.28%
+             tsp  0.888  0.862  3.09%    0+0       0+0   7.040  6.893   2.13%
+            vliw 24.378 23.455  3.94%  106+127    25+45  2.758  2.707   1.91%
+   --------------------------------------------------------------------------
+    Average                     6.12%                                   4.09%
+       SPARC        Compilation Time     Spill+Reload      Run Time
+.25  New            110.25    New   110.25  New
+       barnesHut  3.778  3.592  5.20%    2+2       0+0   3.648  3.453    5.65%
+           boyer  6.632  6.110  8.54%    0+0       0+0   0.258  0.242    6.90%
+    count-graphs  1.435  1.325  8.30%    0+0       0+0  33.672 34.737   -3.07%
+             fft  0.980  0.940  4.26%    3+9       2+6   0.838  0.827    1.41%
+     knuthBendix  3.590  3.138 14.39%    0+0       0+0   0.962  0.967   -0.52%
+          lexgen  6.593  6.072  8.59%    1+1       0+0   1.077  1.078   -0.15%
+            life  0.972  0.868 11.90%   26+26      0+0   0.143  0.140    2.38%
+           logic  2.525  2.387  5.80%    7+7       1+1   5.625  5.158    9.05%
+      mandelbrot  0.090  0.093 -3.57%    0+0       0+0   0.855  0.728   17.39%
+          mlyacc 26.732 23.827 12.19%  162+189    32+57  0.550  0.560   -1.79%
+         nucleic  6.233  6.197  0.59%    3+3       0+0   0.163  0.173   -5.77%
+   ratio-regions  3.780  3.507  7.79%    0+0       0+0 133.993 131.035   2.26%
+             ray  1.595  1.550  2.90%    1+1       0+0   3.440  3.418    0.63%
+          simple  6.972  6.487  7.48%   29+32      5+7   3.523  3.525   -0.05%
+             tsp  1.115  1.063  4.86%    0+0       0+0   7.393  7.265    1.77%
+            vliw 27.765 24.818 11.87%  110+135    25+45  2.265  2.135    6.09%
+   ----------------------------------------------------------------------------
+    Average                     6.94%                                    2.64%
+       X86          Compilation Time     Spill+Reload      Run Time
+.25  New            110.25    New   110.25  New
+       barnesHut  5.530  5.420  2.03%  593+893   597+915   3.532  3.440   2.66%
+           boyer  8.768  7.747 13.19%  493+199   301+289   0.327  0.297  10.11%
+    count-graphs  2.040  2.010  1.49%  298+394   315+457  26.578 28.660  -7.26%
+             fft  1.327  1.302  1.92%  112+209   115+210   1.055  0.962   9.71%
+     knuthBendix  5.218  5.475 -4.69%  451+598   510+650   0.928  0.932  -0.36%
+          lexgen  9.970  9.623  3.60% 1014+841  1157+885   0.947  0.928   1.97%
+            life  1.183  1.183  0.00%  162+182   145+148   0.127  0.103  22.58%
+           logic  3.285  3.512 -6.45%  514+684   591+836   5.682  5.577   1.88%
+      mandelbrot  0.147  0.143  2.33%   38+41     33+54    0.703  0.690   1.93%
+          mlyacc 35.457 32.763  8.22% 3496+4564 3611+4860  0.552  0.550   0.30%
+         nucleic  7.100  6.888  3.07%  239+168   201+158   0.175  0.173   0.96%
+   ratio-regions  6.388  6.843 -6.65% 1182+257   981+300  120.142 120.345 -0.17%
+             ray  2.332  2.338 -0.29%  346+398   402+494   3.593  3.540   1.51%
+          simple  9.912  9.903  0.08% 1475+941  1579+1168  3.057  3.178  -3.83%
+             tsp  1.623  1.532  5.98%  266+200   250+211   8.045  7.878   2.12%
+            vliw 33.947 35.470 -4.29% 2629+2774 2877+3171  2.072  1.890   9.61%
+   ----------------------------------------------------------------------------
+    Average                     1.22%                                     3.36%
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/03/23 16:25:00
+ Tag: leunga-20000323-fix_x86_alpha
+ Description:
+. X86 fixes/changes
+    a.  The old code generated for SETcc was completely wrong.
+        The Intel optimization guide is VERY misleading.
+. ALPHA fixes/changes
+    a.  Added the instructions LDBU, LDWU, STB, STW as per Fermin's suggestion.
+    b.  Added a new mode byteWordLoadStores to the functor parameter to Alpha()
+    c.  Added reassociation code for address computation.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 2000/03/22 01:23:00
+ Tag: leunga-20000322-fix_x86_hppa_ra
+ Description:
+. X86 fixes/changes
+    a.  x86Rewrite bug with MUL3 (found by Lal)
+    b.  Added the instructions FSTS, FSTL
+. PA-RISC fixes/changes
+    a.  B label should not be a delay slot candidate!  Why did this work?
+    b.  ADDT(32, REG(32, r), LI n) now generates one instruction instead of two,
+        as it should be.
+    c.  The assembly syntax for fstds and fstdd was wrong.
+    d.  Added the composite instruction COMICLR/LDO, which is the immediate
+        operand variant of COMCLR/LDO.
+. Generic MLRISC
+    a.  shuffle.sml rewritten to be slightly more efficient
+    b.  DIV bug in mltree-simplify fixed (found by Fermin)
+. Register Allocator
+    a.  I now release the interference graph earlier during spilling.
+        May improve memory usage.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/03/14 14:15:32
+ Tag: blume_main_v110p26p1_2
+ Description:
+. Tools.registerStdShellCmdTool (from smlnj/cm/tool.cm) takes an
+ additional argument called "template" which is an optional string that
+ specifies the layout of the tool command line.  See the CM manual for
+ explanation.
+. A special-purpose tool can be "registered" by simply dropping the
+ corresponding <...>-tool.cm (and/or <...>-ext.cm) into the same
+ directory where the .cm file lives that uses this tool.  (The
+ behavior/misfeature until now was to look for the tool description
+ files in the current working directory.)  As before, tool description
+ files could also be anchored -- in which case they can live anywhere
+ they like.  Following the recent e-mail discussion, this change should
+ make it easier to have special-purpose tools that are shipped together
+ with the sources of the program that uses them.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 2000/03/10 07:48:34
+ Tag: blume_main_v110p26p1_1
+ Description:
+ I added a re-written version of Dave's fixpt script to src/system.
+ Changes relative to the original version:
+   - sh-ified (not everybody has ksh)
+   - automatically figures out which architecture it runs on
+   - uses ./makeml a bit more cleverly
+   - never invokes ./installml (and, thus, does not clobber your
+     good and working installation of sml in case something goes wrong)
+   - accepts max iteration count using option "-iter <n>"
+   - accepts a "base" name using option "-base <base>"
+ It does not build any extraneous heap images but directly rebuilds
+ bin- and boot-hierarchies using makeml's "-rebuild" switch. Finally,
+ it can incorporate existing bin- and boot- hierarchies.  For example,
+ suppose the base is set to "sml" (which is the default).  Then it
+ successively builds
+         sml.bin.<arch>-unix and sml.boot.<arch>-unix
+ then    sml1.bin.<arch>-unix and sml1.boot.<arch>-unix
+ then    sml2.bin.<arch>-unix and sml2.boot.<arch>-unix
+ ...
+ then    sml<n>.bin.<arch>-unix and sml<n>.boot.<arch>-unix
+ and so on.  If any of these already exist, it will just use what's
+ there.  In particular, many people will have the initial set of bin
+ and boot files around, so this saves time for at least one full
+ rebuild.  Having sets of the form <base><k>.{bin,boot}.<arch>-unix for
+ <k>=1,2,... is normally not a good idea when invoking fixpt.  However,
+ they might be the result of an earlier partial run of fixpt (which
+ perhaps got accidentally killed).  In this case, fixpt will quickly
+ move through what exists before continuing where it left off earlier,
+ and, thus, saves a lot of time.
+ ----------------------------------------------------------------------
+ Name: Allen Leung
+ Date: 00/03/10 02:20:00
+ Tag: leunga-20000310-fix_x86_asm_ra
+ Description:
+ More assembly output problems involving the indexed addressing mode
+ on the x86 have been found and corrected. Thanks to Fermin Reig for the
+ fix.
+ The interface and implementation of the register allocator have been changed
+ slightly to accommodate the possibility to skip the register allocation
+ phases completely and go directly to memory allocation.  This is needed
+ for C-- use.
+ ----------------------------------------------------------------------
+ Name: Matthias Blume
+ Date: 00/03/09 10:23:53
+ Tag: blume_main_v110p26p1_0
+ Description:
+ * Complete re-organization of library names.  Many libraries have been
+ consolidated so that they share the same path anchor.  For example,
+ all MLRISC-related libraries are anchored at MLRISC, most libraries that
+ are SML/NJ-specific are under "smlnj".  Notice that names like
+ host-cmb.cm or host-compiler.cm no longer exist.  See system/README
+ for a complete description of the new naming scheme.  Quick reference:
+    host-cmb.cm        -> smlnj/cmb.cm
+    host-compiler.cm   -> smlnj/compiler.cm
+    full-cm.cm         -> smlnj/cm.cm
+    <arch>-<os>.cm     -> smlnj/cmb/<arch>-<os>.cm
+    <arch>-compiler.cm -> smlnj/compiler/<arch>.cm
+ * Bug fixes in CM.
+     - exceptions in user code are being passed through (i.e., reach top level)
+     - more bugs in paranoia mode fixed
+     - bug related to checking group owners fixed
+ * New install.sh script that automagically fetches archive files:
+   The new file config/srcarchiveurl must contain the URL of the
+   (remote) directory that contains bin files (or other source archives).
+   If install.sh does not find the archive locally, it tries to get
+   it from that remote directory.
+   This should simplify installation further:  For machines that have
+   access to the internet, just fetch <version>-config.tgz, unpack it,
+   edit config/targets, and go (run config/install.sh).  The script will
+   fetch everything else that it might need all by itself.
+   For CVS users, this mechanism is not relevant for source archives, but
+   it is convenient for getting new sets of binfiles.
+   Archives should be tar files compressed with either gzip, compress, or
+   bzip2.  The script recognizes .tgz, .tar, tar.gz, tz, .tar.Z, and .tar.bz2.
  ----------------------------------------------------------------------
  Name: Matthias Blume

 Legend:



Removed from v.572
 


changed lines


 
Added in v.815
 Legend:



Removed from v.572
 


changed lines


 
Added in v.815
-Removed from v.572
+Added in v.815

root@smlnj-gforge.cs.uchicago.edu	ViewVC Help
Powered by ViewVC 1.0.0