We have to replace report['id'] in the copy we write, or else its not visible to someone reading the report later.
I've noticed a defect in the branch - we don't inherit the id with the lognamer based id allocator - I'll fix that (we won't use the combination, but someone may) and push that up.
I had the thing in my head that I needed to eject any existing id before hashing, but you're quite right, we can hash with an existing id in there and its no less unique - though if we are not inheriting the id:
- someone publishes the same oops twice with differing ids, we'd end up writing it twice to disk - big deal
- publishing the same oops with no id in it would behave as we do today
- publishing different oopses with the same id will behave as we do today
I'll have a fiddle when I fix the defect and see how it looks.
So thats a really good question.
We have to replace report['id'] in the copy we write, or else its not visible to someone reading the report later.
I've noticed a defect in the branch - we don't inherit the id with the lognamer based id allocator - I'll fix that (we won't use the combination, but someone may) and push that up.
I had the thing in my head that I needed to eject any existing id before hashing, but you're quite right, we can hash with an existing id in there and its no less unique - though if we are not inheriting the id:
- someone publishes the same oops twice with differing ids, we'd end up writing it twice to disk - big deal
- publishing the same oops with no id in it would behave as we do today
- publishing different oopses with the same id will behave as we do today
I'll have a fiddle when I fix the defect and see how it looks.
Thanks,
Rob