MDXXnnnn

MD files contain data records addressed by row and column coordinates. All records in a row also share information called a row header. Similarly, all records in a column share a column header. A complete record thus consists of the row header, column header and data. The headers hold parameters common to the data records. Row headers are located in column 0; column headers are in row 0. Up to 1.6 million columns are allowed in MD files and are supported by the point data clients and servers. Fields in a record are identified by 4-character names called keys. Data values are stored as integers. Each record can be up to 400 words long; individual fields in a record are one word (4 bytes) long.

The MD file structure is self-contained. All information for accessing one of these binary files exists in the 4096-word header, which contains the schema specifying the default number of rows and columns in an MD file; the composition of the row headers, column headers and data records; and the names, scale factors, and units of the keys.

A copy of the schema resides in the MD file and in the McIDAS-X disk file named SCHEMA, which contains all schemas recognized by the system. The copy held in SCHEMA serves as a blueprint for all MD files of a particular kind of data. When an MD file is created, the schema is copied from SCHEMA to the MD file header block with all appropriate modifications.

MD file numbers can be between 1 and 999999. If an MD file number is five or six digits, the file name begins with only MDX or MD. For example, MD file number 12345 has the file name MDX12345, but MD file number 123456 has the file name MD123456.

MD files do not have to follow this standard naming convention. The file masking option of DSSERVE may be used to access a data file of any name through the ADDE.

For MD file Application Program Interfaces (APIs), refer to the API functions list at the end of this section.

Word allocation

Word	Description
0	schema name
1	schema version number
2	schema registration date, yyyddd
3	default number of rows
4	default number of columns
5	total number of keys in the record
6	number of keys in the row header
7	number of keys in the column header
8	number of keys in the data portion
9	1-based position of the column header
10	1-based position of the data portion
11	number of repeat groups
12	size of the repeat group
13	starting position of the repeat group
14	missing data code
15	integer ID of the file
16 - 23	text ID of the file
24	creator's project number
25	creation date, yyyddd
26	creator's ID
27	zero-based offset to the row header
28	zero-based offset to the column header
29	zero-based offset to the data portion
30	first unused word in the file
31	start of the user record
32	start of the key names
33	start of the scale factors
34	start of the units
35 - 38	reserved
39	beginning Julian day of the data, ccyyddd
40	beginning time of the data, hhmmss
41	ending Julian day of the data, ccyyddd
42	ending time of the data, hhmmss
43 - 60	reserved
61 - 62	file name
63	MD file number
64 - 463	user record, MD coordinates (0,0); not described by the schema; use for storing arbitrary information
464 - 863	names of the file keys
864 - 1263	scale factors for the keys
1264 - 1663	units of the keys
1664 - 4095	reserved

The row header begins at word 4096. The row header's length in words is rows * number of keys in the row header.

Following the row header is the column header. The column header's length is columns * number of keys in the column header.

The MD data follows the column header. The data's length is rows * columns * number of keys in the data portion. Any remaining words are available for user purposes.

Words 39-42 are filled in only during the production of real-time files. When real-time file data is copied to other MD files, words 39-42 in the destination files are set to null.

Function	Description
m0pthdr	opens a connection to read a point source data header
m0ptrdhdr	reads the point source data header
m0ptget	opens a connection to read point source data
m0ptrd	reads point source data
mdo	writes point source data
mdquit	deletes a point source file
m0psort	gets point source selection parameters from the command line