Work around PosgreSQL index issue

[TG1999]

Update model to use an index on a checksum of the large summary text field.
Otherwise there can be times when this exceeds PostgreSQL limits.
See #632 (comment)
The failing test before this patch #659
Reference: #650
Signed-off-by: Tushar Goel tushar.goel.dav@gmail.com

[pombredanne]

This should work, but IMHO we need first a minimal test case that reproduces the error.

[pombredanne]

Looking great!
You may be missing one migration though and this should be a data migration that computes the md5 for the actual data?

[TG1999]

When I run ./manage.py makemigrations command again, it does nothing, md5 hash computation part is not getting stored as a migration file.

[ziadhany]

Why do we have three fields in the model (summary_md5, package affected_md5, references_md5) separately. IMHO we can have one field and hash all the fields inside it. This will help us not to store useless data in the database
instead using unique_together ,we can use unique=True .
ex:

md5_field = models.CharField(
                    max_length = 32,  
                    unique = True
                    )

def save(self, *args, **kwargs):
        self.md5_field = hashlib.md5(self.summary.encode("utf-8") + self.affected_packages.encode("utf-8") + ... 
        super(Advisory, self).save(*args, **kwargs)

[pombredanne]

Thank you!
See a few comments for you consideration!