I have a copy of a letter (at home) from Oracle where it says that when they
do the ports they try their best to use the platform at its best.
Not anymore it seems.
If I don`t forget I`ll post it tomorrow (it`s short)
--
Syltrem
http://pages.infinit.net/syltrem (OpenVMS related web site)
To reply to myself directly, remove .spammenot from my address
"Malcolm Dunnett" <nothome_at_spammers.are.scum> a écrit dans le message news:
BTOlo8TcCSsJ_at_malvm5.mala.bc.ca...
> In article <VKbR6.9890$Sm6.680125_at_weber.videotron.net>,
> "Syltrem" <syltrem_at_videotron.ca.spammenot> writes:
>
> > As for myself, I know that since 8.1.6, you can't access a dump file
through
> > Decnet anymore (using node"user pwd"::disk:[dir]file.dmp syntax).
> > I was told they do not use RMS anymore to translate file names, but
instead
> > they use some internal code.
>
> That would be consistant - it appears they no longer use RMS to write
> the dump file.
>
> > In essence, they tried to reivent the wheel but it's not round anymore.
> >
>
> My guess is that their goal is to make the VMS port just another
> Unix port ( ie remove any VMS specific features which might have enhanced
> performance and just do it the same way they do on Unix ). That is good
> in the sense that it should make the VMS ports available sooner, but it's
> unfortunate when it leads to serious performance degradations.
>
> I suppose Oracles response might be that Compaq should fix the
> operating system. As has been mentioned in other contexts in comp.os.vms,
> there are applications that run much faster on Unix than on VMS purely
> because they do inefficient things in terms of disk writes and rely on
> Unix buffer caching to hid the performance penalties this would otherwise
> introduce. If VMS offered a similar "write it whenever you get around to
> it" model this wouldn't be an issue. I understand VMS 7.3 may address
> this to some degree, but I don't have a copy yet so I can't test that.
>
Received on Thu May 31 2001 - 09:55:30 CDT
