[Info-vax] problem with LSE installation

[Stephen Hoffman]

On 2015-05-18 02:00:51 +0000, David Froble said:

> So, you have a common code base.  At some time you make changes, or add 
> something.  Will the code base then fork?  If not, then it seems your 
> only recourse would be to issue something, patches, new version, 
> whatever, for all three, and if you wished, and it fit, VAX too.

The use of a common code base and whether the particular source code 
changes are compatible across architectures or across versions are 
entirely separate details.  The former makes the latter easier, but 
does not dictate anything about the latter.

> Assume that the modifications are non-compatable with earlier versions, 
> and Alphas could not remain in a cluster without the new stuff ....
> 
> What then ?

You either deprecate the old releases or the older architectures for 
certain features, or you issue a "compatibility patch"; a patch kit 
which allows an older version to coexist with a newer version.

Both of these approaches have happened with OpenVMS in fairly recent 
times.  The former with IPSCS, which effectively requires every host to 
be at V8.4, which kicks out all VAX systems.  The latter with 
<http://h71000.www7.hp.com/doc/84final/6677/6677pro_sm2.html>.   Some 
of which may well be what is bagging Phillip with that cluster of his, 
but I digress.

Having a common code base does mean that there's less effort involved 
with getting the patches into all of the builds, but VSI is not going 
to be shipping OpenVMS Alpha updates for the immediate future (yeah, 
VSI might convince HP to provide a compatibility kit, or they might 
not), and by the time shipping patches and new releases for older 
architectures is potentially in play for VSI, the Alpha and VAX 
hardware will be five or ten years older than now, and VSI will have 
that OpenVMS x86-64 port either well along or available.



-- 
Pure Personal Opinion | HoffmanLabs LLC