I have received about 85 replies to the survey so far. First, I would like to
encourage ALL BITNET/NetNorth maintainers to answer the survey, as there are
presently much more EARN answers than BITNET ones, and we run the risk of
having too little BITNET answers to get a significative view of the opinion of
BITNET people.
I would like to start a discussion about some of the plans I am starting to
think about; I need to collect opinions and impressions that cannot fit into a
survey with numeric replies, and for this I would like to use the LSTSRV-L
list only. I am sending a copy of this note to EARNTECH, but I invite
interested EARNTECH people to join the LSTSRV-L list and to continue the
discussion there, in order to avoid cross-posting.
As a matter of principle, I do not want to disclose too much about the results
of the survey before it is closed. However, we are not electing the President
of the InterGalactical Forces, and I need to give you some background so that
you can understand my concerns.
I will probably not surprise anybody if I say that, on the average, people are
quite satisfied with the services provided by LISTSERV today, and that there
is a clear majority of people who want to keep things the way they are today
(this holds true regardless of the "category" - EARN, non-EARN, maintainer,
user). However, and as I expected, there is no clear "winner" or "loser" in
the other 3 choices (sell to EARN, to BITNET, make into a product): it all
depends on the "category" you are looking at, and even so the differences are
not so big.
Clearly, from the technical point of view, solution 1 (keep things the way
they are) is the most satisfactory, and it is also the one that the majority
wants. Well, I may end up doing just that, but it doesn't solve the problem
unfortunately. None of the other solutions is entirely satisfactory either.
What I noticed while browsing through the actual replies (in an attempt to
identify typed "sub-groups" of people) is that there is a non-negligible,
strongly-typed minority of EARN users who answer in the following way:
- Quality of today's product: not satisfactory (2-3 on all topics - remember,
<3 = bad, 3 = neutral, >3 = good).
- Vendor solution: replies vary, but usually not satisfactory (0-3),
especially if I am the one doing the maintenance.
- EARN solution: excellent idea (5), dramatic improvement of both software and
maintenance quality (5), provided that I am not the one doing the work; if I
am, the software will become completely useless (-5).
- BITNET solution: very bad (0-1).
- Recommended solution: sell to EARN.
- Time to rewrite the software: around 6 months of a type-3 programmer, actual
cost depending on the country (taxes vary a lot from one country to the
other in Europe, and are generally much higher than in the States).
Clearly, these people think that I'm doing a very poor job of developing and
maintaining LISTSERV, and that it would be really much better if handed to
someone else. They are not satisfied with the present functions, which can
only mean that they would like LISTSERV to be migrated to OSI. They know that
I will not do it, they know that BITNET will not do it, they doubt that a
vendor would do it if he can make benefits with the present version, and
therefore they are so strongly advocating the "EARN without Eric" solution.
Now, this is one of the fundamental problems. EARN will undoubtedly want to
migrate LISTSERV to VapourNet protocols, there is no way I will accept to do
it (at least not until the vapour solidifies into palpable rock :-) ), and it
is politically unacceptable for EARN to be unable to perform this migration. I
would personally hate to prevent them from doing that anyway, but it is clear
that I do not want to be involved and that, once the code has been OSIfied, I
will no longer want to be involved its support.
Another fundamental requirement is that BITNET and NetNorth should not have to
suffer from the political nonsense that is going on in EARN. The only solution
I found so far would be to "split" the product, giving EARN something they can
work on and providing continuing service to BITNET+NetNorth, unaffected by the
EARN changes. I would like to hear your opinion on the following proposal:
The LISTSERV product would be "forked" into two separate branches, each of
which would originally be the EXACT SAME CODE (with cosmetic changes to the
configuration files).
1. The first one would retain the name of LISTSERV, and would have the same
attributes as today's product. I would keep developing and maintaining it,
free of charge, in the way that I have been doing since the very beginning.
This new LISTSERV would be available (both code and support) free of charge
to BITNET and NetNorth ONLY.
2. The second one would be renamed to, say, LISTEARN, and I would issue an
unlimited, network-wide license to EARN. The conditions would be,
basically, that EARN is allowed to distribute the LISTEARN code to any EARN
site and to use it for as long as the sun shall rise, that EARN is allowed
to make changes to this code and to distribute them in the same fashion,
but that they are not allowed to sell any of the code or otherwise
distribute it to institutions outside EARN. As a matter of principle, I
would ask EARN for some reasonable amount of money in return, but this is
irrelevant at this point. LISTEARN would be "owned" by EARN (with the small
restrictions I mentioned above), maintained by EARN through the EARN staff
budget (not by me), and available free of charge to all EARN sites.
I would like to insist on the fact that, when the "forking" is done, the two
products are identical. That is, at t=0, LISTSERV == LISTEARN.
Another important thing is that I would support LISTSERV but not LISTEARN.
This should be very clear: any question, problem or bug report about LISTEARN
would be sent back to its originator, with instructions to send it to whomever
EARN chooses to maintain LISTEARN. Whatever happens to LISTEARN is not my
business, and if users are not satisfied with the quality of the product, they
should complain to their BoD member (who is the normal channel of
communication for this kind of problems).
As far as backbone and peering is concerned, LISTEARN would be a "different"
type of list server, in much the same way as the BITNIC LISTSERV was. The
LISTSERVs would not know about the LISTEARNs, which might in fact evolve into
something that is no longer compatible with LISTSERV. Similarly, the LISTEARNs
will not know about the LISTSERVs, and we will have two separate backbones:
one in the US+Canada, one in Europe. This is of course not very good from the
technical point of view, but then that is exactly what will have to happen
anyway when EARN migrates to OSI protocols, or when BITNET migrates to domain
naming, etc.
I am open to suggestions on this topic. However please keep in mind that any
solution you may propose has to meet the following requirements:
- Little or no impact to the LISTSERV service presently provided to BITNET and
NetNorth; this includes continued LISTSERV maintenance and development.
- Ability for EARN to make whatever changes they want to make to LISTSERV, in
their time scale, without being subjected to uncontrollable external
elements like the "so-called volunteers".
- Complete removal of my involvement and responsibility insofar as EARN
changes to the software are concerned.
- Respect of the results of the survey, which will be available on the 18th of
february and on which we can only make assumptions for the time being.
Thanks for your time and involvement, Eric
PS: Remember - all future discussions to be held on LSTSRV-L only.
|
