Message 23/432 Harro Verkouter                    Oct 3, 2000 04:19:09 pm +0200

Date: Tue, 03 Oct 2000 16:19:09 +0200
Organization: JIVE
X-Accept-Language: en
CC: evntech@jb.man.ac.uk
Subject: EVNtech: [Re: TOG Reports and Agenda] - RTC report
Sender: owner-evntech@jb.man.ac.uk
Precedence: bulk


Hi All,


Please find a report on the RTC project enclosed.


The RTC test of February 2000 revealed a couple of operational problems
with the software. Small problem in DBGET (which is fixed, tested only
at Wb so may need more). The major problem(s) are in the back-end. The
infrastructure of the SW (both internal as well as external) is not
suitable for the demands we require for reliable, predictable and
moreover usable operation. Too much human intervention is needed
(=source of errors/unpredicableness).

In order to make it (the whole of the RTC) work I thought it necessary
to start all anew. The rewrite is not as difficult as it may seem
because it's not a huge project. I primarily focussed on information
infrastructure. This consists (at least) of the following areas (which
are mostly working as we speak):

* incoming data-area:  rtc data files arrive here and are checked if
they really are rtc datafiles.

* experiment database area: all relevant information for an experiment
is now read from the drudge file rather than specifying them on the
commandline

* output area:  the webpages are created here. Effort has gone into
assuring that (e.g.) already correlated experiments do not get lost from
the Archive page; also an extra level of indexing has been added: the
data will be found in the following hierarchy:

        <experiment>/<scan>/<baseline>

as opposed to

        <scan>/<baseline>

where <scan> was being interpreted as <experiment>.


Still need to do:

* the actual correlation (which is not very difficult)
* conversion  from RTC dataformat to Station Data File (SDF) format,
which is used by JPL's sotware correlator. This is also not very
difficult: a program already exist to do this. Need to make it smarter    
though. Can easily incorporate it in the code since it basically is one
or two routines...
* Model Specification File generation. JPL's software correlator reads
its model from this file. Very important. Once you have the right
numbers it's not that difficult. Routines for writing the file
syntactically correct already exist (from the previous rtc-code); needs
incorporating + adding smartness + take advantage of better information
infrastructure.
* Finish the webpage-content. The layout/creation etc is taken care of
Ok only need to write more decent stuff into the actual pages.


Cheers,

--

Groeten,


Harro Verkouter                 Joint Institute for VLBI in Europe
verkouter@jive.nl               Radiosterrenwacht Dwingeloo
                                Postbus 2
tel.: (+31)(0)521-596516        7990 AA DWINGELOO, the Netherlands
fax.: (+31)(0)521-597332

------------------------------------------------------------------

Accident, n.:
        A condition in which presence of mind is good, but absence of
        body is better.
                -- Foolish Dictionary

------------------------------------------------------------------