Another approach can be to generate the Review model dynamically as explained here: http://code.djangoproject.com/wiki/DynamicModels
This way, the Review model can be generated dynamically based on the Encounter model: For every field in the model create 3 for the Review. So, changes to the encounter model will be reflected to the Review model resulting in DRYness. But this will not avoid the redundancy in the db tables. Thanks... On Aug 19, 11:02 pm, omat <[EMAIL PROTECTED]> wrote: > Hi all, > > I have a model class that holds patients' encounters. It has fields > that hold some facts such as the date the record created, the doctor > who diagnosed, etc. > > Also there are about 20 other fields about the diagnosis. This data > about the diagnosis are reviewed by experts. If an expert thinks an > attribute of the diagnosis is wrong, he challenges that attribute and > supplies the corrected value, his comments and the type of mistake he > thinks exists. This is possible for all those 20 fields. > > The corrected value is of the same type as the type of the challenged > fields, therefore varies. Comments are TextField. Type of mistake is > ManyToManyField. > > The current model structure is as follows: > > class Encounter(models.Model): > # not challenged fields > doctor = ... > created = ... > # challenged fields > diagnosis = ForeignKey(...) > patient_number = CharField > operation1 = ForeignKey > operation2 = ForeignKey > activity = ForeignKey > ... > > and a Review model to hold the reviews: > > class Review(models.Model > encounter = ForeignKey(Encounter) > diagnosis = ForeignKey(..., blank=True) > diagnosis_comment = TextField() > diagnosis_type = ManyToManyField(ChallengeType) > patient_no = ForeignKey(..., blank=True) > patient_no_comment = TextField() > patient_no_type = ManyToManyField(ChallengeType) > operation1 = ForeignKey(..., blank=True) > operation1_comment = TextField() > operation1_type = ManyToManyField(ChallengeType) > # ... this goes on for every challengeable field ... > > As can be seen, this not DRY at all, very error prone and introduces a > lot of redundancy. > > I can think of some alternatives such as keeping the field name as a > reference to the field. This would reduce the Review models to > something like this: > > class Review(models.Model) > encounter = ForeignKey(Encounter) > field_name = CharField() > new_value = CharField() > comment = TextField() > type = ManyToManyField(ChallengeType) > > Two drawbacks of this approach I can think of are: > - you cannot directly access the referenced field by the convenient db > api way (or can you?) > - the new_value cannot be of the same type as the challenged fields > type, as we do not know which field it would refer to > > Another approach I can think of is XML approach. The Review model can > be as follows: > > class Review(models.Model): > encounter = ForeignKey(Encounter) > review = XMLField() > > And the 'review' XML field will hold all the data on reviewed fields > in a non-relational and flexible format. > > I cannot decide among these approaches as neither of them feels > natural. > > Any advice would be much appreciated. > > thanks, > omat --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
