$RBGB/bdemakeLessonsLearned080117.txt

Wed 080116:
-------------
After spending 3 days trial and error, bde almost compiles,
yet there are still still problems for linux and alpha hosts.
 
Thur 080117:
------------
bde finally compiles on saturn and mercury with and without logging.

I did not recognize the real problem[s] until today:

One Imakefile was corrupted by a missing target:dependency line
and/or a missing '\' continuation symbol in an imake macro.

The implicit (and wrong) default dprint* function prototypes
in bde/src/Imakefile resulted in cc not gcc and cxx not c++.
I assumed failure to #include dprint.h in schema.h in graph.h.
This lost time and did not work.

#ifdef USEDECC caused prototype decl'ns to fail because it 
always defined and cc or cxx was called instead of gcc or g++.
#ifdef CPLUSPLUS also led to failure because cxx was used 
and did not know about gnu's __FUNCTION__.  To fix this
I added a new #if defined symbol pair: -DUSECC or -DUSECPLUSPLUS
is now passed  to the compiler from Imakefiles via Makefile 
$(CC) commands, depending on the directory src or pr_util_*log, 
respectively.

The complexity of 4-way builds (8-way with USEDECC option) 
from a common Imakefile trio is daunting, esp. since saturn 
is old DEC UINX vintage. The alternate of converting
from imake to gnu automake or cmake or something else
takes a learning curve that I can't afford right now.