On Wednesday 20 January 2016 23:48:59 Tim Graham wrote: > A proposal has been raised to change the validation for > CommaSeparatedIntegerField to allow negative numbers [1]. This is obviously > backwards-incompatible, so I don't think it'll fly without some option to > control the behavior, however, I wanted to ask if there's any reason not to > deprecate the field at this point. From a quick glance, I believe it could > be replaced with > CharField(validators=[validate_comma_separated_integer_list]). The only > database storage difference is on Oracle: > > 'CharField': 'NVARCHAR2(%(max_length)s)', > 'CommaSeparatedIntegerField': 'VARCHAR2(%(max_length)s)', > > Is this difference meaningful?
<guess from="memory"> The two field types use different encodings, and on oracle the maximum field length for character columns is in bytes rather than characters; so, in many settimgs, I think you'll be able to store twice as many numbers in a CommaSeparatedIntegerField as in the equivalent CharField. </guess> It's an edge case, either way (who puts hundreds of numbers in such a field?) Shai.
