AccessConversion - Revision history

KarlPinc: /* The Process */ Search path is not really necessary to run prepare_data().

2013-04-11T02:29:18Z

‎The Process: Search path is not really necessary to run prepare_data().

2013-04-10T21:24:54Z

‎The Process: New page_followtype conversion table.

2013-03-22T21:12:52Z

‎The Process: Mention follow_page table

2013-03-20T21:39:36Z

‎A year's worth of example conversion sql: Line break

2013-03-20T21:27:17Z

‎A year's worth of example conversion sql: Add schema

2013-03-15T19:49:48Z

‎Obtaining Multiple Errors: sqlbyline program -- make separate sub-sections

2013-03-08T17:29:17Z

‎The Process: left and right groups exist in access

2013-03-08T16:44:30Z

‎The Process: Note search_path

2013-03-08T05:34:24Z

‎The Process: Wording

2013-03-08T05:33:32Z

‎The Process: Grammer

@@ Line 34: / Line 34: @@
 ** follow
 * Import the Access csv files into their corresponding tables in the <code>conversion</code> schema.  Import in order of creation as listed above.
 * Prepare the Access data for conversion by running <code>prepare_data()</code> as follows:
-** Be sure that <code>prepared_data</code> shows up as part of your search path.
+** Either
-** Type (paste/etc.) the following into your SQL window and execute it:<br/><code>select prepare_data();</code>
+*** Be sure that <code>prepared_data</code> shows up as part of your search path.
 * Use convert_access() to convert a desired time period, the dates are inclusive.  Repeat as necessary with different time periods.  If you get an error the database is unchanged. <br/> <code>select convert_access('1990-01-01', '1990-12-31');</code>
 * Fix errors in the data and repeat until done.

@@ Line 7: / Line 7: @@
 * Import all the support (lookup) tables
 * Import the <code>BIOGRAPHY_DATA</code> table
-* (If necessary) upload the Cross_reference_forprogram.xlxs spreadsheet.
+* (If necessary) upload the Cross_reference_forprogram.xlxs spreadsheet into the <code>conversion</code> schema.
 ** Export it from Excel as csv
 ** Upload into the cross_reference_forprogram table.
 * Export the following tables from Access as csv files:
 ** follow

@@ Line 12: / Line 12: @@
 * Export the following tables from Access as csv files:
 ** follow
 ** follow_1_minute
 ** follow_1_minute_infant
@@ Line 27: / Line 28: @@
 ** follow_1_minute_infant
 ** follow_1_minute
 * Import the Access csv files into their corresponding tables in the <code>conversion</code> schema.  Import in order of creation as listed above.
 * Prepare the Access data for conversion by running <code>prepare_data()</code> as follows:
@@ Line 36: / Line 38: @@
 * When finished query the following tables in the <code>prepared_data</code> schema looking in the column named <code>done</code> for rows with a value of <code>FALSE</code>.  These rows were not converted.  Figure out why and try again:
 ** follow
 ** follow_1_minute
 ** follow_1_minute_infant
@@ Line 42: / Line 45: @@
 ** follow_5_minute
 ** follow_5_minute_group
 == Obtaining Multiple Errors ==

@@ Line 48: / Line 48: @@
 in which case the data in the database remains untouched.  If there are many errors in the data it can be tedious converting many days with a single invocation of <code>convert_access()</code>, fixing a single error, and repeating.  The solution is to run the program many times, and, regardless of error or lack thereof, discard all changes to the database so that the process can be repeated as many times as desired.
 The following example code can be cut and paste to find up to an error per day within entire year's worth of data.  Cutting and pasting the example, together with a little search and replace, and attention to leap year, can adapt it to any year or to multiple years.
 <pre>
 begin; select convert_access('1996-01-01', '1996-01-01'); rollback;

@@ Line 17: / Line 17: @@
 ** follow_1_minute_sibling
 ** follow_5_minute
 * Delete all rows from the following tables in the <code>conversion</code> schema (if necessary), in the following order:
 ** follow_5_minute
 ** follow_1_minute_sibling