19991013: Discussion items at the MUG meeting (cont.)

>From: Dee Wade <address@hidden>
>Organization: SSEC
>Keywords: 199910131730.LAA13670 Unidata concerns for the MUG

Dee,

>I am assuming that your question was covered in the meeting.
>I must admit I expected there to be more discussion about your
>and Dave's differences of opinion than there was.

Quite frankly, even before going to the meeting I had the feeling that
Dave S. had decided on a course of development for configuration of
ADDE subservers that was unchangeable.  Since I didn't want to keep
beating a dead horse, I opted to not make too much noise about the
situation during the meeting.  Is it your feeling that the subject
was/is not closed?

>Do you feel you got the information you were after?

No. Actually, I was told by more than one MUG member that I didn't
press hard enough on two points:

o can we get assurances that new development efforts will be funded to
  their completion.  In particular, the development strategy adopted by
  Dave S. regarding configuration of ADDE dataset subservers had my
  option (sort of) as the last stage in implementation.  I am very
  worried that, like so many things in the past, monies will be cut off
  from the effort before it is finished.  This will leave those of us
  who need a way to control access on a dataset-by-dataset basis out in
  the cold.  I have had long talks with both Jeff Wilson of the BoM and
  Bryan Batson of JSFC about this issue, and they stand firmly behind
  the goal of IP mask access to datasets

o I don't feel that I ever really got a good answer about an inquiry
  being closed when there is dispute as to its status.  Again, the real
  issue that is the basis for my question is the closed status on my
  inquiry about what McInitLocalServer should do in regards to filling
  out the values in the servacct structure.  My and other's reading of
  the Programmer's Reference Manual suggests that all I need to do is
  call McInitLocalServer to get all of the fields in the servacct
  structure.  I was informed by Russ Dengle that at least one of the
  fields was not available anywhere (password).  I can live with this,
  but I think that the routine ought to provide the IP address of the
  requesting client.  My inquiry was closed when John Benson responded
  that I can get the information by using a system call.  This is fine
  AND I am doing so, but I feel that this should be rolled into
  McInitLocalServer.

As far as my concern about development tasks getting finished: I am
still of the mind that ADDE is not finished enough to fully transition
my users to it.  This is disturbing given that quite a few non-ADDE
routines are not Y2K capable!  The most glaring case in point is the
inability to do things like temperature advection, vorticity, theta,
theta-e, etc from point source plotting routines.  McIDAS users are
used to having this capability, so the first question that they ask is
where it is when they try to use routines like SFCPLOT, SFCCON,
RAOBPLOT, RAOBCON, etc.

I have to say that it is very hard to promote the use of McIDAS at my
sites when things like this are not ready for prime time.  The promise
of ADDE access to remote datasets is sparking the enthusiasm for a
number of sites that had basically stopped using McIDAS.  If the
functionality is not full, however, even ADDE will not pursuade them to
keep/restart using the package.

Tom