Yes, I too have been tackling this problem since Allbase "D"
Every release has some external that did not exist on the
previous one. On one occasion HP added a parameter to sqlxcon
(or similar) and this was on a PASS!. So, I had to have different
versions as follows:
 
     Allbase E
     Allbase F pre Pass 33
     Allbase F pass 33 and above
     Allbase G
 
I've scrapped the E and pre 33 versions now, but still have the
other two!
 
Why oh why was sqlxcon (?) not made option extensible to begin with?
 
The Allbase team has known about this problem for ages (I've reported
it at least four times over as many years). From an ISV's point of view
it makes life awkward to say the least. From an IMAGE/SQL User's point
of view he simply cannot develop on one of his systems then expect to
distribute code to all others unless they happen to be on the same
release.  If I remember correctly, the reason why HP will not fix this
problem (or say it won't happen in the future) is that features on
one version of image/sql are not compatible with previous versions.
True, but surely developers are intelligent enough to program according
to version.
 
Thanks for highlighting this real problem in public Denys.
 
On HP-UX, it's even worse.....because of the lack of shared libs for
Allbase, you have to link in the parser with your program. The parser
does its work according to the version of allbase you linked on. This
means,for example, that if you linked with the F libsql and found
yourself running on a G Environment and tried to emit a valid TRUNCATE
TABLE command, the F libsql would reject it because it knows nothing
about TRUNCATE (not available on F). So I have to supply two versions
on UX also.
 
Mike Whiteley