I'm using field.represent for reference fields basically everywhere across my app. The recent commit
"make grid formatters_by_type always overrule field.represent for automatic Columns" completely breaks that, as field.represent is skipped for all reference fields.
I would highly encourage you to revert this commit, or at least ensure field.represent can still be used. Currently, I would need to use a manual Column() in each Grid, each time I include the field. This is a lot of boilerplate that previously wasn't necessary.
You even said yourself that the formatters_by_type can probably be deprecated entirely. So why did you suddenly do a complete 180 on this and make it the ONLY way to format reference fields?
> Yes this broke when fields in pydal were given all a default represent function. I think we should just deprecate FORMATTERS_BY_TYPE.
> Do we really want need the checkboxes?
For now, i will have to stay on an older version just before that commit.
FYI: My logic for field.represent on reference fields works like this: I created a DB view for each table which adds *_represent columns for each reference field. These are then added to Grid.required_fields by my Grid subclass, and used in field.represent if they are part of the row passed to field.represent. This is done to avoid repeated single-row queries caused by the built-in methods of representing reference fields, as those were a significant performance issue for some of my tables which have 10+ reference fields. (15 rows per grid page * 10 reference fields = 150 single-row queries per grid page)