IMO, resource IDs need to be opaque - the fact that it went from [id] to id should not
have mattered. clients should never care about a format or syntax of an ID. Now, of
course, I say that knowing that I myself just hacked up the server through my knowledge
that IDs have the feed ID in it :-) but I have a big TODO in the code to change that
soon.
----- Original Message -----
On 30 Jul 2015, at 18:38, Michael Burman wrote:
> Why are we fighting encoding in 2015? Mainframes had these things
Because:
a) the recent change of resource id format has broken code in several
places where the semantics were not really known and/or implicitly applied
Metrics are stored with a [id] while the inventory now only has id and no
longer [id]. That must not happen again
b) not all consumers/clients are machines. We need to make it easy to
consume our services
c) the current format of specifying resource data + resource type information
may not be what we want going forward.
d) we need to ensure that if e.g. ':tag' all of a sudden gets a special
meaning in a metric id, clients do not send that as part of a normal id,
as metrics or any other part of Hawkular may behave in ways that are
not obvious.
If we keep what we have fine. But then we need to document it and
stick to it in a consistent way all over the place, so that ids or partial
urls are/become predictable.
_______________________________________________
hawkular-dev mailing list
hawkular-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hawkular-dev