We had a CM program that worked fine for all test given it from inhouse.
When posted to the production machine our clients would receive STACK
OVERFLOWS on the executable. We would ask client what did you do and we
would repeat the procedure inhouse on the production machine with no
problem. Come to find out that all inhouse connection were by NS/VT over a
LAN and the clients were coming in on a X.25 Pad through the DTC. The
definition of terminal characteristics was different. Buffers allocated for
LAN connetions were different than phone connections, thus the STACK
OVERFLOW. Since CM is so limiting and those who code for it are becoming a
rare breed, data is being allocated at a rapid pace. Ha, on our HP9000 our
programmers came up with a executable that needed more than 1.6GB Data to
compile and as you may have already guessed it didn't compile. Data use is
being exchanged for good program logic and speed. The quicker you can move
to NM the better, but in the long run, the better you train your programmers
the better the programs will be.

James

> -----Original Message-----
> From: HP-3000 Systems Discussion [mailto:[log in to unmask]]On
> Behalf Of Sletten Kenneth W KPWA
> Sent: Thursday, May 04, 2000 3:24 AM
> To: [log in to unmask]
> Subject: Re: System Abort Finale?
>
>
<snip>
> We just gained a bunch of in-house customers that were
> running an old but fairly large and complex app on the LAST
> MPE/V machine I know of locally (a Series 70).  What may
> (or may not) be semi-associated with this thread:
>
> The local owners of our Series 70 apps tested their software
> fairly extensively on our 959 in 1999;  all appeared to be well
> (and their application manager kept good notes, indicating
> which sub-systems had been tested and passed on our 959).
>
> Then now when they moved to our 959 for real, they found out
> at least one of the (unchanged) CM PROGs that worked fine
> in 1999 crashed with a stack UNDERFLOW....  now... what
> has changed ??..  hmmm...:  well (obviously), we are now in
> the post-Y2K era.  Note I have no hard evidence of any kind
> that our CM problem was caused by going through the Y2K
> boundary;  just that it's the only thing we know that is different.
>
> Anyway, I note in Jim's above their QUERY abort occurred on
> what he thinks is a vanilla 5.5 system.  IIRC HP sez you should
> be on >= PP6 to be fully Y2K compliant on everything;  even
> though a lot of stuff was "generally" Y2K compliant well prior
> to that....  but also IIRC, 5.5 PP zippo is NOT Y2K compliant
> in a number of respects....   NOW:
>
> Can we therefore conclude Jim's QUERY abort is a latent Y2K
> problem ??..  of course not...  but:  I ditto Gary's question:
> CM or NM QUERY ??....
>
<snip>