On 30 Jul 2015, at 20:54, Michael Burman wrote:
But this [id] or id has nothing to do with encoding. It could have
just as well been 'hawkular.id' or 'id' and be an issue. This is more
You are right, it has nothing to do with encoding.
It is rather a mis-understanding of semantics where some parts/parties
were thinking the [] belong to the id while others don't.
We need to define those semantics going forward.
to do with namespacing issue than limiting what characters we could
use in the metrics ids. Hawkular-Metrics still returned the exact same
name as was written to it. For namespacing (if we want to have
This is nice, but not enough for Hawkular overall.
Also if John says he needs a special character for some tags, then
we need to make sure that this special character does not show up
in a resource id, when the resource id is part of the metric name,
as otherwise it could be misinterpreted by Hawkular-Metrics.
Not to mention the normal browser usage. If the user wants to use
easily remembered ids, he should not use generated ones (who could
remember uuids if they can't remember encoding rules?). If this is a
question of some certain agent then it's still not a global issue.
I am not proposing UUIDs.
But for humans to remember a sequence of %5D%20%2F%20%5B, it is
as readable as a UUID.