GNOME Bugzilla – Bug 135523
Text import drops extra columns that are added a long way down file
Last modified: 2009-06-11 15:20:35 UTC
I have a text file I'm importing that is very long (over 10,000 rows). The number of columns varies all the way down the file, but there are 2 columns for most of the file, and then 3 columns near the bottom. During the text export stage, Gnumeric detects that there are 2 columns (in how ever many rows at the top of the file it looks at), which I don't have a problem with, because examining the whole file would probably take to long. However, when importing the file, the region of the file with 3 columns has the 3rd column completely dropped, presumably because only 2 columns were visible in the top part of the file. Expected behaviour: If extra columns are found than were originally determined, they should just be added with General format type (or perhaps parsed to see if they are text or numbers). The data shouldn't just be dropped.
Hmm, You may have noticed that yo can choose which columns to import. You apparently did not choose that third column to import (of course you are not even given that opportunity). The default assumption is that you only want to import checked columns. Perhaps that needs to be changed.
I don't understand the general value of making the default "import everything, including columns I uncheck". Columns that were not detected initially are a special case -- they should probably always be tacked onto the side, I would think -- then the user can decide what to do with them.
I just fixed a problem with empty columns. In "separated" mode we detect all three columns. In "fixed" mode you need to add the third column by hand. (And now you actually can do that.) I see nothing further wrong at this point. This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report.