Closed AMUZY closed 10 months ago
I also want to ask, is it done on purpose that if we filter the Firefox
product by a channel release-cdntest
, we still see all the release
and release*
channels showing up ?
I also want to ask, is it done on purpose that if we filter the
Firefox
product by a channelrelease-cdntest
, we still see all therelease
andrelease*
channels showing up ?
You should not be seeing release
rules on release-cdntest
. You should be seeing release*
- which is a glob that matches anything starting with release
. (Note that if a Rule has a scheduled change pending, it will show up if either the current version of the Rule or the scheduled change matches the channel filter, eg: if either contains release*
it will show up for all channels that start with release
.)
Yeah you're right. It's just the release*
channel that shows up. But please have you tested it out yet? @gabrielBusta
Hi @bhearsum , I have refactored the conditions to better match what is required of the filter function and it works. Please take a look thanks.
Apologies that I haven't been able to give this a full review yet - I will try to get to it tomorrow. One quick comment: you should use
to compare the rule to a channel rather than doing it by hand here. (As you can see, that function covers even more cases as well.)
I have tried this function on two occasions here and it didn't work perfectly in this case. It doesn't consider the product
, rather checks only with the channel
. It's suitable for other cases just not here. I'd try it again and give you a feedback later.
Apologies that I haven't been able to give this a full review yet - I will try to get to it tomorrow. One quick comment: you should use https://github.com/mozilla-releng/balrog/blob/a4536abd23c8e9d3d103f6fae1c10ef593fe58db/ui/src/utils/rules.js#L2
to compare the rule to a channel rather than doing it by hand here. (As you can see, that function covers even more cases as well.)
I have tried this function on two occasions here and it didn't work perfectly in this case. It doesn't consider the
product
, rather checks only with thechannel
. It's suitable for other cases just not here. I'd try it again and give you a feedback later.
Thanks for calling this out -- I think you're right, as this considers a scheduled change with a null channel to be a match. With that being the case, we should refactor it a bit to at least be able to make use of the matchesGlob
function currently inside of it. (That's the main thing I don't want to see re-implemented elsewhere.)
Apologies that I haven't been able to give this a full review yet - I will try to get to it tomorrow. One quick comment: you should use https://github.com/mozilla-releng/balrog/blob/a4536abd23c8e9d3d103f6fae1c10ef593fe58db/ui/src/utils/rules.js#L2
to compare the rule to a channel rather than doing it by hand here. (As you can see, that function covers even more cases as well.)
I have tried this function on two occasions here and it didn't work perfectly in this case. It doesn't consider the
product
, rather checks only with thechannel
. It's suitable for other cases just not here. I'd try it again and give you a feedback later.Thanks for calling this out -- I think you're right, as this considers a scheduled change with a null channel to be a match. With that being the case, we should refactor it a bit to at least be able to make use of the
matchesGlob
function currently inside of it. (That's the main thing I don't want to see re-implemented elsewhere.)
Hah! Told ya I was right. Don't worry, I'd refactor that function to accommodate the current conditions and give you feedback soon👍
Hi @gabrielBusta , this PR solves the issue about scheduled deletions showing for unrelated channels