Hi Havard
On 09/03/2016 22:02, Havard Eidnes wrote:
I would like abuse-c: much more if it were changed in two ways:
- permit abuse-c: in inet(6)num: objects
- permit abuse-c: to point to a normal person: object, not only role:
This boils down to what I thought would have been the better
implementation all along.
Better is a relative word. It would be better to improve the
operational usability and solve the problems you have without breaking
the design.
I have worked on the design, development and support of this database
for 15 years. I know it inside out. I know what problems it has. I
have seen the mistakes both new and old users have made. I have seen
the crazy things users have done because the database semantics,
syntax, business rules allowed it.
Frankly, I fail to see how that is that relevant to the above
suggestion.
Because when you have a long term perspective on how people have used
this database, the above suggestions fit into the category of those
crazy things, imo.
If you make the changes you are asking for I guarantee within 2 or 3
years people will be saying abuse-c is a failure, lets invent
something new to fix the mess.
What sort of mess? Can you please be a bit more explicit?
First of all references to PERSON objects. The original design, from old
documents I have read over the years, was that a PERSON object is for
holding personal information about a real person. These should only be
referenced in ROLE objects and the ROLE objects should be referenced
every where else in the database. But the constraints were never
implemented and PERSON and ROLE just became a fuzzy mess. The RIPE NCC
has spent a lot of time over the years helping people out when a person
leaves an organisation and HIS PERSON object is referenced in 10s or
100s of thousands of objects.
I know Gert is particularly upset that he has to create a ROLE object.
But having some standardisation in how things work, in the long term,
leads to more efficient usage and better understanding of how the
database works.
But this also highlights one of my issues with the people who discuss
these points on these mailing lists. Most of these people have used this
database for so long they 'just know' how it works. They don't seem to
understand the problems new users face when trying to understand this
thing. It is a case of decimalisation vs imperial measurements. Once you
know the imperial measurements you never forget them, but it takes a
long time to learn them.
Then there is the issue of allowing abuse-c in multiple places some of
which will override others. Over time this will get confused, more
specific values will be missed and reports will be sent to the wrong
addresses. Also some people will still dig into the database manually
thinking they know how it works and will inevitably find the wrong values.
The trouble with these technical mailing lists is it is the same very
very small number of people out of the 12k members and other
interested parties who keep pushing the same ideas to fix your
problems regardless of the consequences. As long as it works OK for
you, everything must be fine. None of you are willing to think out of
the box. I proposed some options for fixing these problems 2 years ago
on RIPE Labs. I am not saying they are the best solutions, but no one
has even commented on them in 2 years.
This is not an open, bottom up process. This is a cartel of old timers
who make all the decisions so they get their own way. This needs to be
fixed.
"Thanks" for the ad hominem. Instead of discussing the merits of the
above suggestion in a reasoned manner, I get told that I'm part of an
old-mans-club or worse. I may be old, but that was nevertheless
uncalled for, IMHO.
I hit reply but this was intended as a generalised comment. However you
cannot argue against the facts. Ignoring the AP WG, where half the
people are trying to maximise the value of IP addresses and the other
half are trying to minimise costs in the crazy IP gold mine, for the
technical mailing lists it is the same small group of people who make
the decisions. In many cases they already know each others viewpoints.
They can choose to blank out discussions they don't want to even
acknowledge, like the data model. I have been trying to start a
discussion on having a review to see if anything would be improved by a
change. That is so far removed from actually changing anything. But
still no one will even acknowledge my points. Only Shane and one other
person have ever commented on any point I have made about the data model
in the last 8 months. Some 'senior'/'elders'/'long established' people
on these lists, who respond to other points I make, noticeably cut out
the bit where I mentioned the data model in the reply. So nothing I said
about the data model gets perpetuated in any discussion.
But now I don't think there is any point in having that discussion. It
will be the same people who discuss it and I already know their
views...leave it alone, change nothing, 'we' don't need it simplifying.
It will help new members more than established ones, although it could
benefit all members. And there aren't many new members on these lists.
Two years ago when I used to go to all RIPE Meetings and talk to people
about the data model and explain some ideas there was quite a bit of
interest. But those people don't talk on mailing lists....except the
previous co-chairs of the DB WG who all liked some of my ideas when I
presented to them.
The mailing list archives are there for anyone to analyse. I did this
once and the number of individuals (excluding chairs and NCC staff who
you expect to be on the lists) who commented on the technical lists over
a year was very small indeed. I don't expect much to change in the near
future...
cheers
denis
Best regards,
- Håvard
