Signed-off-by: Lukas Holecek hluk@email.cz
@lucarval, @gnaponie Anything else we discussed that should be in the documentation?
message => messages
message
messages
in Greenwave => in the Greenwave
in Greenwave
in the Greenwave
have matching subject type, product version and test case name. => have the subject type, product version, and test case name match.
have matching subject type, product version and test case name.
have the subject type, product version, and test case name match.
contains list => contains the list and new => and the new
contains list
contains the list
and new
and the new
Previous => The previous
Previous
The previous
have same => have the same
have same
have the same
receiving new => receiving a new
receiving new
receiving a new
product version but => a product version, but
product version but
a product version, but
from subject => from the subject
from subject
from the subject
product version => the product version
product version
the product version
It publishes single message for a decision change. Maybe a decision change message? (I'm bad with articles.)
a decision change message
rebased onto 89b20914d97ea9a1d33061c4d8090d8216e9bffa
message => messages It publishes single message for a decision change. Maybe a decision change message? (I'm bad with articles.)
That sounds good to me.
rebased onto e4f382ea0593a2e83f98ae1dd0f581cc514d6015
Updated.
That changes the previous decision. Or no message is sent. I would emphasize this concept and put an example.
I would also explicitly say that if there's no change, the message doesn't get published, because, as it is clear from the topic, it is about a "decision update". I know it's a lot of repetition of the same concept, but many people didn't get this in the past and they were confused about why Greenwave doesn't just publishes a message every time a new result or waiver comes in.
Can we remove the 2 "and"s? Just "the list of applicable policies, the new and previous decision."
isn't it "when - a millisecond" or something like that?
But actually... is it really important for the user to know this information?
Yeah. I would add that people are free to configure "custom" product version because maybe it doesn't apply to their use case. But this means that "If the product version cannot be guessed, policies with any product version are considered...)
@gnaponie No idea what you mean here.
isn't it "when - a millisecond" or something like that? But actually... is it really important for the user to know this information?
I think this could be useful for some trouble shooting. But mentioning the removed millisecond is maybe too much ... not sure.
I would also explicitly say that if there's no change, the message doesn't get published
This is mentioned below:
The decision change message is not published if the new and previous decisions have the same satisfied and unsatisfied requirements.
rebased onto 62936e1ddc293aa5b1313c8a5be0e371cc3ae177
I've added:
Some policies may use custom unique value for product version when test subjects are not related to any real world product.
Can we remove the 2 "and"s?
Done.
Yeah sorry, I guess I said it in an unclear way. I mean that I would say something like (but maybe better rephrased): "when there is a new test result in ResultsDB or a new waiver in WaiverDB related to a predefined policy, and the new result or waiver changes the previous decision. Or no message is sent."
Because from how you wrote this sentence I would understand that a message get published every time there's a new result/waiver, no matter if the decision changed.
Nevermind. I guess the sentence is fine.
Optional: But can't we just say here... "Its value is immediately before the submit time of the new result/waiver" or something?
I mean that I would say something like (but maybe better rephrased): "when there is a new test result in ResultsDB or a new waiver in WaiverDB related to a predefined policy, and the new result or waiver changes the previous decision. Or no message is sent."
OK, changed.
rebased onto 9d354f357889364602e33ae5f5a12386a2546035
thanks +1 \o/
rebased onto 1beb84c5d32490e5514ff259e1f6bd3f562c334e
Pull-Request has been merged by lholecek
Signed-off-by: Lukas Holecek hluk@email.cz