"Plotstyle" Data.
   These files contained output from Panak's "heave" command, and their data were for measurements of generation.
The files were in a directory called "Generation" and lacked a file 'type' appendage, such as ".dat" or ".txt"
even though they were text files. Further, some of their names when passed through the identify process matched
the identities prepared from some of the ordinary printometer files, that were not for generation but for consumption
though at the same location. Thus, Files  MARAETAI, OHAAKI, OTAHUHU, WAIRAKEI and WAITAKI could be confused with
the other files that had been assigned to NI.etc or SI.etc for the printometer data. Files TUAI and MANGAHAO
were already assigned to GEN.TUAI and GEN.MANGAHAO from the printometer data, and this seemed correct.
   Rather than attempt to distinguish between data from Printometer/etc and Generation/etc, perhaps by
including a part of the name one directory level up (except that this name changes by year for the printo files),
I decided to supply a file type of ".heaved" for the Generation collection as this would reduce the annoyance of
having untyped files as well as providing a distinguishing mark that would be used in the identification
process, since not being ".dat" or ".data", or ".txt", it would not be snipped off.
   Corresponding changes were needed for the alias list, and the still unresolved issue of data duplication
is further deferred.
   File NZTOTAL turned out to contain internally the assertion "North Island Total Generation" despite the "NZTOTAL"
so for the moment it is taken as NITOTAL, thus corresponding to a SITOTAL. To be resolved...
   File RANGIPO.heaved turned out to have four copies of the same data, each heading differing only in its time stamp.
Another such was TOKAANU.heavd with three lots, except that the second lot stopped short with 23/4/1996.

   It turned out that the "heaved" files produced numbers that matched those numbers identified as generation data
from the printometer file collection. This is good, as heaved files were produced by Panak by reading the data
of the printometer file collection, performing some appropriate combination as the years varied, and heaving the
result to a file in some variation of the "plotstyle" format. There are some heaved files that are the summation
of sequences, DCNORTH, DCSOUTH, NZTOTAL, and SITOTAL that are not in the printometer collection.
File ARNOLD adds a new data run, possibly due to confusion in the attributions of channels in the printometer data,
since as heaved files are produced from the printometer data, where have their sources gone? The values also
look odd, at ~45MW being about a factor of fourteen high until the end of 3/9/1989 when they convert to the expected
3MW (though with inadequate precision - about a 100KW step size) and continuing the style of a daily load curve
(not generation) until the end of the 17'th September whereupon values tend to be flat for days at a time.
   File STRATFD however contained a great wad of zero values for 1/4/1988 to 31/3/1991 yet the other files from
the printometer collection had many non-zero values as well as zeros for this span. Since a "heaved" file is
produced by Panak by reading printometer files, this file has lesser standing and so the section was excised.
As usual, in the absence of the log files produced by Panak, assessment of the provenance is precluded.

   Earlier versions of Panak were intended to reject negative numbers, but as soon as it became possible to
prepare summations and generate data files that were intended for input to Panak, there soon arose occasions
when negative numbers were to be written to the files. An ad-hoc approach involved representing via the error
indication facility (see the Corrigenda for the printometer collection) involving sequences such as "u(a-3)"
to represent "-3"; a data file with such sequences when read by Panak and "heaved" to a "plotstyle" format
would in turn produce "-3ua" until eventually, Panak was extended to allow negative numbers generally.
One objection to this was that it would slow the scan of data, and it did, adding about 4% to a run's cpu time.
   Which proved to let through errors, as with Cobb on Tuesday 22/8/1989 and in Matahina for 19/8/1988
which otherwise would have been rejected. So these two repairs are copied to the heaved files also.
   For HUNTLY,    3464 "ua" sequences removed.
   For MARSDEN,  38510 "ua" sequences removed.
   For MEREMERE, 11305 "ua" sequences removed.
   For NEWPLYM   12404 "ua" sequences removed.
   For OHAAKI     2127 "ua" sequences removed.
   For OTAHUHU   37899 "ua" sequences removed.
   For WAIRAKEI    413 "ua" sequences removed.
   After which there were only two error codes in the entire collection, which drew attention. File DCSOUTH
contained the sequence 19940320,2,50,0,0,0,0,0u ,0u ,0,0,0,etc and that is for the daylight saving stretch day.
This corresponds to file D:\HH\Source\Printometer\Data\GEN9196\SIDCS.DAT and given the ocean of zero values,
(and the absence of explanatory comment), it seemed safe enough to remove the two annoyances. There should
have been a second code, "g" for "Guess", or better, "a" for "approved" but there was just a blank for "blank think".
So, placing a "g" will encourage Gnash to regard each value as good, and this will never be considered again.
Except that there proved to be one "a" for "approved" code in the collection, for Huntly on my 1988 birthday,
and this appears in the corresponding source file D:\HH\Source\Printometer\Data\GEN8889\HLY.DAT, though with
no explanation. Perhaps it was overlooked during the fiddle.

   Aside from the re-encoding, various odd values have been noticed, that should have been noticed in the past.
Modification on various pretexts abates the oddness, but changes must be matched with the corresponding
other data files. That is the "heaved" and the "normal" values must be changed in the same way.
   For Cobb, miscellaneous monster values were attributed to mistypes or meter hiccoughs. Some high then low counts might be due to timing hiccoughs...
   For Highbank, some monster values looked to be mistypes by a punch hussy that weren't noticed. Changed to match again the values in GEN8990\HBK.DAT
      19890521 changed 222300 to  22300 ( 9:30pm)
      19890913 changed 120600 to  12000 ( 7:00am)
      19891012 changed 119800 to  11800 (11:00pm)
   For Coleridge, to match the changes in GEN8990\COL.DAT.
      19890921 changed 244300 to  24400 (10:30pm) though there remains puzzlement as to how autmatic data recorders could mistype...
   For DCNORTH changed  27500 to 527500 (-noon) for 31/3/92 (and the corresponding mistype in NIDCN.DAT, its obvious source)
   For Clyde,
      19940225 changed 630200 to 360200 (-10:30pm) as per the corresponding transposition in CYD.DAT.
      19960619 changed 525400 to 255400 (-midn.)   as per the corresponding transposition in CYD.DAT.
   For Waipori there were a number of values too large to be possible generation figures and on inspection it appeared
that although a mistype did not seem likely, they could be reduced under the banner of "reset error".
This is where after the half-hour's value has been printed, the counter was not set back to zero so that the next
count would appear to be about double. This hypothesis should be checked by viewing the printometer sheet to see
if an operator made an annotation or otherwise recorded an indication of mechanical error, but, the sheets are not
available to me for inspection. Actually, I don't know if these data were hand-typed, nor if the printometer
actually produced printometer sheets. The source data files are given the printometer type of "L&G" but in the 90s
the Landis & Gyr meters were recording their data electronically so that reset errors should not occur.
Except that this applied to the revenue meters measuring power exiting the system, and the metering at the
generators may not have been replaced, especially at Waipori as it was owned by the Dunedin City Council,
and the elctronic metering project was a NZED initiative.
   Anyway, the monster values continued to irk (and, why were they not noticed?), and now have been abated, as follows:
      19910824 changed 124800 to 73700.
      19920706 changed 104300, 104300, 104300 to 52200, 52100, 52200.
      19920928 changed 148200 to 79400.
      19930820 changed 118700 to 59500.
      19930830 changed 128600 to 64500.
      19940614 changed 160700 to 80700.
      19950529 changed 101000 to 51600.
      19950825 changed 102600 to 51200.
   For Highbank, some monster values were chopped without any pretext other than that they were too big to be possible.
      19910921 changed 0,44600 to 22300,22300.
      19910924 changed 0,43100 to 21600,21500.
      19930926 changed 0,55100 to 25000,25100.
      19940514 changed 50000,50000,50000, to 0,0,0.
      19940711 changed 88900 to 22900.
      19940908 changed 42900 to 21500 as a presumed reset error.
      19950613 changed 44200 to 24200 as a presumed mistype.
      19950912 changed 65500 to 25500.
      19920724 changed 41600 to 20800 as a presumed reset error.
      19921001 changed 40900 to 20600 as a presumed reset error.
      19921019 changed 40700 to 20800 as a presumed reset error.
      19930721 changed 39600 to 20200 as a presumed reset error.
      19950919 changed 36700 to 18500 as a presumed reset error.
      19930818 changed 36200 ro 18000 as a presumed reset error.
      19910527 changed 34600 to 16800 as a presumed reset error.
      19941209 changed 30100 to 14700 as a presumed reset error.
      19931008 changed 0,28100 to 14000,14100.
   For Ohau A, a value of 484800 was double the likely value and so seemed a "reset" error. It's just odd that its preceding value is 2424. I'd like to view the printometer strip.
      19930212 changed 484800 to 242400
   For Tekapo,
      19911008 changed 60800 to 20800 as being likely a mistype.
      19940312 changed 50700 to 25800 as a reset error.
      19940318 changed 48800 to 24600 as a reset error.
      19940415 changed 48500 to 23900 as a reset error.
      19911208 changed 45800 to 23200 as a reset error.
   For Pukaki,
      19911003 changed 280800 to 139900 as a reset error.
      19940725 changed 0,246700 into 123400,123300.
   For Arnold, with no other data source besides the orphan "heaved" file,
      19890629 changed 99300 to 51900 as being midway between adjacent values, similar to adjacent days.
      19900211 changed 76700 to 3200, as per all surrounding values.
   For Benmore a monster value of 719000 could not be accepted, yet the usual pretexts offer no persuasive value either.
      19930617 changed 719000 to ? as no good value could be invented.
   For Manapouri
      19960919 changed 1121600 to 574700 as a reset error. Despite the printometer type being declared L&G.