I have a csv file with one of the columns containing data such as 0x62, 0x68, etc.
When importing the csv into JMP using JSL open(infile), it is found that JMP automatically converts the column into a Numeric datatype, changing 0x62 to 98 (decimal) in the process.
Manually importing the csv not using script, the same issue (auto-converts).
It only happened to one of the computers, other computers did not auto-convert.
Text import settings are the same between the computer that auto-converts, vs. those that don't.
Appreciate if there's any advice on what to look into to prevent JMP from auto converting? And if there's a script line that could be added to ensure all computer systems will be pre-conditioned to the same settings prior to text import?
One the one computer that is doing the auto convert, I would suggest that when selecting the file for input, that you specify the "Data (Using Preview)".
You can then specify that the Hex column not be interpreted as being numeric, but rather to keep it as a character. In the preview window below, I have chosen to keep the Age column as character.
Additionally, once you have read in the .csv file once, you can go to the Source entry for the table, and Edit it and save the JSL code that used to read in the file. You will see in the JSL how JMP sets the column to be character.
Using Multiple File Import might be another work-around in JMP 14 or later. MFI expects a directory and a filename pattern, but it is easy enough to make that pattern be a complete file name.
Tried the Data (Using Preview) option and indeed that one computer showed up as Number. That computer is using JMP13 -- is there a dependence to JMP version on how JMP interprets data types differently during csv text import?
For now, I might add lines in script to force the data type to Character every time a user imports, but this will have the penalty of performance (run time).
As the releases of JMP come out, there definitely can be changes such as you see. The JMP folks are always trying to improve such things, which at times end up with what looks like inconsistencies, not as improvements.
What you may want to do, is to modify your code to check for the release of JMP being used, and then only the versions that need the fix will have it's script's efficiency impacted.