RJLRef: $CASE/gen/ver_1l/aareadme.2k0316
 (renamed from aareadme.genv11)
SUBJ: Building genv11 for bde2alpha

==================================================================
Hello all.  Thanks to NaiyanaT and JoeF for reminding me about genv11 for bde.

The correct genmerge=genv11 executable command to run before (or while)
making bde/pr_util is:
	$CASE/99s523/genmerge/chgenv11/chgen/executables/chgen	
This shell proc is platform-independent, and runs ./$HOSTTYPE/chgen,
which does exist for decstations as well as alpha's:
---------------------------------
jupiter.cs.uml.edu(106)> pwd
/usr/proj3/case/99s523/genmerge/chgenv11/chgen/executables
jupiter.cs.uml.edu(107)> lg alpha dec*
alpha:
total 1240
-rwxr-xr-x   1 lechner  98s523   1261568 May 12  1999 chgen11

decstatiol 616
-rwx--x--x   1 lechner  98s523    615552 Sep 24 16:46 chgen11
jupiter.cs.uml.edu(108)>
------------------------------------
Although you don't have to recompile chgen  every time you build bde,
you hould re-run chgen, to get an up-to-date pr_util library. 
With gen 10.2, vdhamoda says the backup (and altered)
pr_util library in her apr directory must be used. 
When using genmerge=genv11 she claims that is not necessry.
 (there were parallel bug-fixes I presume).

The value of HOSTTYPE is a system environment variable: E.g.,  
-------
cs.uml.edu(63)> echo $HOSTTYPE
alpha
------
Unfortunately only csh and tcsh know about it - sh may not.
$HOSTTYPE  is similar to $MACHINE, which is compiler-defined 
but not known at shell command level:
-------------
jupiter.cs.uml.edu(65)> echo $MACHINE
MACHINE: Undefined variable.
--------------
The repository versions of chgen  should all have a shell proc 
in genversion/executables/chgen that runs the CORRECT platform-specific
chgen executables, like bde/executables/bde. Please incorporate and USE it. 

Do NOT bind in HOSTTYPE to your test commands or Imakefiles.
When you make bde on one CPU, pr_util/Makefile should run the 
chgen executable on the same CPU type.
Building chgen was never part of pr_util's Imakefile, only running it was.

(Although in principle  chgen should produce the same pr_util content
regardless of its binary's HOSTTYPE, there is no use tempting fate here,
or risking accidental HOSTTYPE conflicts by: not knowing which platform
your latest executable was compiled for, ignoring the abort message, 
and trusting the earlier version of pr_util that your make run did not change.)

Note to Jeff Feinberg: 
---------------------
Genmerge may not have its own platform-independent shell proc.
If it does, it should run the chgen in gen/exec*/$HOSTTYPE or gen/bin/$HOSTTYPE
(bin should be renamed executables for consistency with bde).
--------
Both gen and bde Imakefiles probably leave the compiled gen or bde  in */src,
but they should move it to */exec*/$HOSTTYPE, otherwise you must remember to 
move it manually before it can be used properly.

R Lechner
=================================================
> 
> Hi Joe,
> 
> > 	I checked the Makefile in the parent directory, bde, and determined that 
> > it did indeed have the macros that I wanted.  Remember, though, that the make 
> > failed on the pr_util references.  I simply opened the Makefile in the bde 
> > directory, found a line that said:
> > 	
> > 	SUBDIRS = pr_util src
> > 	
> > 	and took the pr_util reference out.  I ran the Makefile in the bde 
> > directory, which drove the Makefile in the src subdirectory and everything built 
> > cleanly and I was able to start the BDE utility that got built as a result.
> 
> I think we don't need to remove pr_util.  I have changed Makefile in
> subdirectory pr_util at line :
> 
> CHGEN
> =/usr/proj3/case/97f522/bde2java/hhuang/chgen/executables/${HOSTTYPE}/chgen10.2
> 
> from ${HOSTTYPE} -> alpha because there is no declaration of HOSTTYPE.
> 
> >From changing this, I can completely buid bde. However, I haven't run bde
> yet since I can't run bde at home. Hopefully, it works.
> 
> Naiyana
>