Hi Jeremy: On Thu, Feb 7, 2013 at 3:18 PM, Jeremy Sanders <[email protected]> wrote: > Hi Joseph > > This is a long-standing bug (probably from the 1st release). Each import > filter is responsible for remembering what the name of imported dataset is. > When the user renames the dataset, the import filter doesn't know about the > change, and so it doesn't get recorded in the saved file. > > It's a bit hard to fix. The standard import filter has the name encoded in a > text descriptor entered by the user (which is hard to change), or in a file > (very hard to change). Likewise, CSV files have the dataset names within > them. There is no way for an import plugin to get to know about a name > change.
I'm starting to see that these dataset names are very important. If you import two CSV files with the same heading names without giving them a prefix, the second import just overwrites the datasets from the first import. Also, if you change the name of a dataset, then plot properties will lose their references to the data and take on empty values. Would it be possible to associate each dataset with the column number it came from instead of its heading name? In my experience, the column number is the better identifier. I often make cosmetic changes to headings in CSV files, but I usually don't change the order of the columns. I also have a lot of data without column headers. It seems to me that if all internal references to datasets were stored as (filename, column number), and the dataset names were just cosmetic labels, then all problems associated with renaming datasets and name collisions would be eliminated. Or are things not that simple? I don't want to give the impression that I'm suggesting you make these changes. I'm just trying to understand the design issues better. In any case, I still find Veusz to be the best graphing program I've used. Joseph _______________________________________________ Veusz-discuss mailing list [email protected] https://mail.gna.org/listinfo/veusz-discuss
