GitLab

By setting the seed value for the random number generator explicitly we
can reproduce the random number time series. This lets us exactly
reproduce the value used for the publication on TCCA approximation
errors.

With the new UIPC software installation on 2019-07-30 the name of
channel 31 was changed from LHeLvl to LHe-Alt, which gives a proper
distinction to LHe-Lvl. With the new library version LHeLvl and LHe-Alt
both are translated to LHX.

Adjust doc-markers such that GPL phrase is not included in libttsynt.doc.
Remove obsolete CVS keywords.
Adjust make rule for libttsynt.doc to include include-files and separate
sections from different files.

The dims variable passed to function mxCreateStructArray was declared int,
while the declaration of mxCreateStructArray makes the compiler expect a
variable of type const mwSize*

Apparently /bin/logger is confused, if the tag parameter contains
several whitespace-separated strings. It behaves very much like the
logger function, if a single string "DL1logger" is passed as tag.

this is just a quick hack unfortunately; the code and in particular
diagram settings require a major revision

The log_frequency scale cannot contain exactly the requested number of
values for the lowest decade. This test mode allows to check whether the
higher value decades are affected by this. They are not.

The reason for not producing the requested number of value is, that the
smallest frequency on the logarithmic scale than would be smaller than
the smallest value on the linear scale. So log_frequency lets the
smallest value of the logarithmic scale be as small as the smallest
value of the linear scale, but not smaller.

In cases, where no PSD is required, the PSD computation is skipped. This
was not properly accounted for in the checks done in the program:

PSD is not required for transfer of phase and the existence of these
value shall not be checked in these cases; check for frequency array
being smaller than 2 (rather than 1); add verbosity output

- use correct way of specifying units
- avoid Hollerith constants