$CASE/bdelog2ks/aareadme_bde  (was ~lechner/bdetest/aareadme) 
- RJL 2k0301: Moved from $CASE/2ks523 (now linked to $CASE/bdelog2ks)

This file identifies recent versions of bde and chgen and their 
project version executables, platforms and documentation paths.
It also reports simple tests which I did this week by opening an existing .dat file on each version.

The versions are identified by their major authors in chronologal order:

	Mark Aldred, Fredy Jara, Carl Parisi, 
	Yu Wong, Haiyan Huang, and R Poonsup.
================================================================

Mark Aldred  (95fbde) 

This project was moved to reclaim space and linked to $CASE:
 $CASE/95f523/bdesrc -> /usr/cs/fac1/lechner/factmp/95f523/bdesrc

Aldred's report is in both 
	~lechner/factmp/95f523/bdesrc/maldred/doc/alphaPort_maldred.*
and 	 $CASE/gen/ver_10/chgen/doc/alphaPort_maldred.*

His source code for bde is in bdesrc, but his executable is 
linked to $CASE/95s523/95sbde/bde/executables/decstation/bde,
which was probably over-written by later checkouts.
His version probably ran on a decstation and a vax.
His HN/HA toggle simplification is valuable and perhaps salvageable
(I am guessing that all his mods for this are in his bde/src/*.cc.new,
since they affect the menu callbacks and button options.)

Here is 2k0216 run/abort and 'gdb <bdepath> core' + 'info stack':
after I did FILE/Open and selected Mark's linked test.dat file.
This file is saved locally as bdeabort.dat: I can't see anything funny
There, but coding onconsistencies may be at fault.
-------
...
key encode: invalid character key length: 1; key: \
<<< invalid test.dat file?? - RJL
signal_handler: signal 11
signal_handler: saving diagram to bdeabort.dat
Segmentation fault (core dumped)
remus(63)> gdb /usr/proj3/case/95s523/95sbde/bde/executables/decstation/bde core 
GDB 4.14 (mips-dec-ultrix), Copyright 1995 Free Software Foundation, Inc...
Core was generated by `bde'.
Program terminated with signal 11, Segmentation fault.
#0  0x41f4a8 in graph::graph (this=0x100cb260, passHG=0x0)
    at graphobject.cc:1473
graphobject.cc:1473: No such file or directory. <<<not local or linked - RJL
(gdb) info stack
#0  0x41f4a8 in graph::graph (this=0x100cb260, passHG=0x0)
    at graphobject.cc:1473
#1  0x418bc8 in DisplayGraph (w=0x100ca000, c_data=0x0, cbs=0x7fffb46c)
    at fileio.cc:1142
#2  0x55e688 in XtCallCallbackList ()
#3  0x4957f4 in _XmSelectionBoxNoGeoRequest ()
#4  0x55e6bc in XtCallCallbackList ()
#5  0x4a298c in XmCreateToggleButtonGadget ()
...
---------------

Mark tried hard to port both chgen and bde to alpha CPUs but finished 
only the chgen port. His report has one valuable improvement to bde
that got lost in the latter parallel version shuffle. It simplifies
the switch from node label input to HA-text-line additions which are 
now almost hidden in the top EDIT/HN.or.HAtext pop-up menu option.
This feature  is described in his report as follows:
--------------
1.3. BDE User Guide Release Notes

This new version of bde includes a new toggle switch that replaces a popup
menu in the options menubar.  This toggle switch is located in the lower left
corner of the bde window.  It allows the user to easily switch between
modifying node attributes and node names.  This changes is not complex but
makes bde much easier to use.
-------------

TBD: re-introduce this feature into bde.

============================================================


Fredy Jara (96subdelog):
        $CASE/96su523/bdelog/fjara/bde/executables/bde (decstation)
        $CASE/96su523/bdelog/fjara/bde/doc/96fbdecode_report.tex

This version was built for remus/decstations.
It contains line restyle options on-screen but not in
the fprint.cc routine, which needs repair to save-as .ps
It will not compile on alpha machines.

=========================================================

Carl Parisi (96fbdecode):
        $CASE/96f523/bdecode/cparisi/bde/executables/bde (decstation)
        $CASE/96f523/bdecode/cparisi/bde/doc/bdecode.*

cparisi added table XB so tables HA and GX can share its text display 
code. This version predated the ghostvars concept, so it was more disruptive 
to maintenance. It changed bde's schema.sch, which was not under
CVS revision control, so different schema names were invented or mapped
to a constant name. This complicated Imake and configuration files,
as described in HHuang's report below.

XB is a new 'superclass' of tables HA and HI,
and a candidate for refactoring, because it provides common text rendering
methods that were previously duplicated in HA and GX (and HI).

Parisi did not retain ghost copies of XB's text line back in HI or HA.
He did retain the line-relative semi-fixed offset of each HA or HI text line
in real [0..1] interval (world) coordinates, and he also converted them
into an absolute window-relative integer pixel location in table XB.

Unfortunately he did not implement the symmetric transformations
or maintain the ghost variable correspondence that incremental
refactoring requires.
                              
==========================================================

YuWong (97sbde2gif)
	($CASE/97s523/bde2gif/yuwong/

Bde2gif includes changes to bde that add a FILE/Print menu save-as-gif  
option that converts one or all diagrams to .xbm and then to .gif format.
The 97s523/bde2ht1 project used these gif files instead of the old method 
of saving to .ps via fprint() and invoking ps2gif 
(a 2-step process via ppm bitmaps).  Bde2ht1 generated
html versions of bde diagrams with clickable hot-spots
(to linked URL's) in each node.

YuWong's source code changes were tested and worked on a vax CPU only; 
His report states that his changes were also included in the bde version
under the $CASE/97s523/bde2ht1 project and their demos use his options.     

YuWong imported freeware utilities to save-as bitmap (.xbm) and 
convert to .gif.  These are in two directories lateral to bde: 

	gifutil contains xpaint and xfonts contains 3MB of font files. 

From these he derived the directory bde/extern_code and compiled it 
via xmkmf -a (imake):
-------
jupiter.cs.uml.edu(73)> pwd
/usr/proj3/case/97s523/bde2gif/yuwong/bde
------
jupiter.cs.uml.edu(74)> grep extern_code Imakefile
104:SUBDIRS = pr_util extern_code src
jupiter.cs.uml.edu(75)>               
-------
jupiter.cs.uml.edu(60)> ls extern_code
Imakefile    hash.c       image.c      imageComp.c  palette.c    protocol.c
Makefile     hash.o       image.o      imageComp.o  palette.o    writeGIF.c
jupiter.cs.uml.edu(61)>
--------

TBD: reintroduce this option in later vesions of bde.




















=============================================================

Haiyan Huang (98sbdesym):
	$CASE/97f522/bde2java/hhuang/bde/executables/bde (for decstations)

	$CASE/97f522/bde2java/hhuang/doc/98sbdesym_report.txt 

This version requires schema tables XB, SY,TK and SD.
It displays two .dat files in ~/bdetest named bdeHhuang*.dat.
Two hhuang schemas with all four tables are
        bde_schema.sch 98sbde_schema.sch

CParisi used XB but NOT TK, SY and SD.
Hhuang's schema dir'y also includes 3 schema
in which even XB is NOT used:
	oldbde.sch 95sbde_schema.sch 96subde_schema.sch
	(CParisi used XB but NOT the others.)
	
Note that doc is lateral to, not under bde, because it deals with 
chgen and symtxt not just bde. This version does not build on alpha CPUs.

Huang wrote this  20+ page report on bdesym. Her report
details changes to add TK and SY and SD tables as ghost var for the 
text string field in table XB. She built upon excellent work by 
rcostell/rconnell in 96s523/bdesym. 

What is more significant, Huang also details the configuration
problems she faced with varying schema and her not-so-robust fixes
that may be compounding the error. Note also that she used chgen 
ver_10.2 that contained some changes (TT and TA in chgen)  but
may not have been integrated with bdelog by KSpinney/JKarner.

Her version should be rebuilt using gen/ver_11 (genmerge). 
Then her code (if stable) should be integrated with bdealpha below.

hhuang's version does not fix fjara's fprint.cc nor does it include 
97sbde2gif/yuwong's extern_code directory and its save-as .gif FILE option. 

HHuang's XB schema extension is not very disruptive
since TK is a new child table and XB is unchanged as a ghost copy. 
Table XB's text field and its child-set of TK tokens are (in principle 
at least) a ghost-variable 'pair' with complex forward and 
backward transformations.













====================================================================

RPoonsup/VDhamoda (99sbde2alpha)

	$CASE/99s523/bde2alpha/vdhamoda/bde/executables/bde (for alpha CPUs)

	$CASE/99s523/bde2alpha/vdhamoda/bde/doc/bde2alpha99sReport.* 

	(see $CASE/99s523/bde2alpha/vdhamoda/bde/aareadme for make problems.)
	(see $CASE/gen/ver_10/chgen/doc/alphaPort* for Mark Aldred's earlier 
	 95s523 critique of alpha-porting problems for chgen and bde.
	No vdhamoda/bde/schema/*.sch file contains table XB.


I tested this version from the repository checkout tree,
	/usr/proj3/case/95s523/95sbde/bde/executables
which has bde.<author> scripts.
I opened bdetest.dat there and it aborted with this stack:
#0  0x12003ea98 in gen_pr_add (viewname=0x1 <Address 0x1 out of bounds>, 
    tbl_abbrv=0x1 <Address 0x1 out of bounds>, tbl_ptr=0x11ffff6a0 "\004")
    at pr_load.c:2246


This is the first successful port of bde to dec/alpha CPUs. It would
probably also build for a decstation but that is untested. It benefited
from earlier advice by Mark Aldred above, who ported chgen to alpha.

99sbdealpha has some peculiarities - [particularly hand-modified pr_load 
and pr_dump sources (backed-up in vdhamoda/apr). These replace  chgen-produced 
sources in vdhamoda/bde/pr_util]. These are for use with chgen ver_10.2; 
they are supposedly not to be needed with chgen ver_11 (gen_merge).

TBD: Fix the peculiarities. See what prior contributions are missing 
and if possible re-integrate them to this alpha version. 
Note that fjara's version was the starting point for this port. 

The bdesym revision sequence was not used in this bde2alpha port 
since the XB table and TK/SY/SD tables changed the schema 
and the executable was not as robust as fjara's version.
Therefore .dat files with tables TK/SY/SD will crash this executable.
There 3 bde/schema/*.sch  files, none of which contain table XB:
	94sbde_schema.sch   95sbde_schema.sch   96subde_schema.sch

See $CASE/bdealpha/bde/executables/alpha/aareadme  for a discussion 
of bdealpha builds and tests - its build is sensitive to 
g++ vs dec's cxx compiler #defines in Imakefiles.