Larry,
That is an excellent question. Perhaps I can describe what I am doing, and then 
the list can give me a better idea on how to do it.
I have a few hundred thousand PDF files that are being created from a scanner. 
 
For the most part, we are scanning oldler documents first. All the scans go 
into a single shared folder. Although the files might be from 2005, the file 
date, created by the scanning program is the date of the scan, ie the date the 
paper was turned into a PDF. So, more or less, older files have older file 
dates. 
 
To get the names of the files into Rbase, I have to capture the output of the 
DIR command. As you know, the Rbase version of the DIR command does not support 
any sorting.
 
So, I load the data from the output of the DIR command into a table.
I extract the file name, file size, file date, file time for each row of data, 
ie 1 row per PDF file.
 
Now I want my app to show me the oldest file first, and move it to a folder 
that makes sense in our office. I would like to imagine that the most recent 
files end up "On Top of Pile" in each folder. I call that the MROT system, 
(Most Recent On Top). Same as the organization on my desk.
 
So now I have to find the oldest file in the table based on the file date and 
the file time. Since many of these files were created on the same date by the 
scanner, I can't just ask for the minimum date. I need to look at the minimum 
time within the minimum date. And if the scanner was working very quickly, even 
the file times might be the same, so I use the name of the file which the 
scanner creates for me....ie xxxx0001.pdf was scanned before (and is therefore 
older) than xxxx0002.pdf
 
Now I could do a search for the lowest file date, and then the lowest file time 
within that group of lowest file dates, and then the lowest filename within the 
lowest file time...but that means 3 searches each time I move a file from the 
scanned files folder to the destination folder. Not a bad solution, but I was 
trying to be efficient. 
 
I tried using a cursor to go through the pile of files, but sometimes my girls 
skip a few files because they are hard to read, or they don't know what is the 
best folder to put them in. So I had to start flagging rows to indicate what 
happened to them, ie skipped or filed. I have a next and previous button in the 
app. The logic started giving me a headache if they skipped a few files, then 
moved a few files, then went backwards a few files, etc. I also have an 
"unfile" last button...which moves the file back from the destination folder 
and into the unsorted folder. So now I have to unflag that row. I started 
getting a really bad headache.
 
Further more, when we go back to look at the files, usually we want to look at 
the last file that went into the folder first. To figure that out, I would have 
to do a triple search, or set up a cursor and move through the pile of files. 
Mentally, is is easier for me to wrap my brain around a single value in a 
single column. Once the files are in their folder, they are renumbered, 
starting with 1. If a file is ever deleted out of the middle of the pile, it 
makes it easy to see...there will be a number missing.
 
So, my idea was to sort them just once by date, time and name, give them a 
number starting with a 1, and then use the numbers to sort the files.
 
My solution is far from perfect, but it beats the heck out of file cabinets!!!
 
Mike


________________________________
From: Lawrence Lustig <[email protected]>
To: RBASE-L Mailing List <[email protected]> 
Sent: Sunday, March 18, 2012 8:41 PM
Subject: [RBASE-L] - Re: What is the best way to autonumber a column for 
sorting?


<<
I have a table with a few thousand rows. The first column is an integer and 
initially it has no data. That is the column I want to use to number my rows. I 
want to put a 1 in the column for the row that has the lowest value for the 
2nd, 3rd and 4th columns. They all have data that is easy to sort (date, time 
and filename).

>>

Why do you need a proxy column if you can sort the data by the existing 
columns?  What happens if the data in the underlying columns changes, should 
the sort order column also change?
--
Larry

Reply via email to