#Your ROCKDB data
21-Mar-94 11:50:12
Sb: #125192-#Your ROCKDB data
Fm: Dale W. Harrison 76711,517
To: Doug Mitchell 70621,702
DM> the size, perhaps ZIP archiving would be better; it is smaller and
DM> almost as widely usable as ARC format (I just reviewed COMPRS.DOC and
ZIP for sure, include a copy in ARC format but monitor the downloads; if there
are none (or only 1 or 2) in 6 months, discontinue it.
DM> What about it folks? Am I barking up any wrong trees here?
Well, I think with some good relational design we could knock the size of the
datafiles down significantly – even more than you think. If I had a good
dataset I could design and build a shareware (or freeware – haven't thought
this out enough yet) program to access it. I'd use dBASE IV since that's my
language of choice right now, but I have a C-code generator and need something
to justify playing with it; this might be the project I work on it for. I
recommend DBFs as the universal micro data format.
Regarding shrinking the data for download (and diskette) efficiency, this is
what relational databases are for.
DM> System: abbreviation would help limit the size of the text form
I would make XLSYSTEM a one-character field: I,H,R,4,O,M,T,N for the fields
you listed above. The only non-obvious one is 4 for Tetragonal.
DM> Color: again abbreviations would help. I used only three
DM> letters and limited the number of colors: Whi(te), Bla(ck),
I would use one-character abbreviations for COLORS and provide a lookup table
for the abbreviations. I might also provide a parent-child relationship with a
separate table just for colors so that you weren't limited in the number of
colors you could include for a given mineral. If you fix the colors field at
20 characters (say), then for an ASCII file or an Xbase DBF you must include
those 20 characters for every mineral regardless of whether or not it is
needed. This can add up in a hurry.
DM> Luster: some abbreviations could be worked out here, though I
In my teaching DBF program Mineral Key we make users select between M
(metallic) and non-metallic first, and then select a non-metallic luster. Again
we used 1-character abbreviations. In the first pass at the program I used
many abbreviations in a 5- or 6-char field for minerals with multiple lusters
but in a larger system I might use a child table.
DM> Hardness: a hardness like "2-2.5" makes it tough to make greater
Absolutely must be a pair of fields. Ditto for Specific Gravity. If there is
no range, then both numbers are the same.
DM> Cleavage: I would be interested to know what you do when there
In Mineral Key we just stored a number.
I have permission from my Mineral Key coauthor to make it shareware and upload
it, just haven't gotten a round tuit. Maybe having a much larger dataset would
make it happen <g>. We could maintain the DBFs and the program as two separate
files in the LIBs, too, so once you had the program you could just get updates
to the mineral files.
Yours in freedom,
-=< Angus Scott-Fleming >=-
GeoApplications
Tucson, Arizona
* Natural drill rig * Kansas geologist wearing cleats in a tornado. *