It has been a few years, but I think we came to basically the same
conclusion on build/test skipping in autojobs for CassCI. Most wanted
the job runs all the time, but a few wanted to skip runs for scratch or
WIP branches, in order to preserve resources when they didn't care what
the result was.
I don't know the details of CircleCI regex, if it will grok both leading
and trailing wildcards, or we could be slightly more strict and skip
/nobuild-.*/ for any branches that match 'nobuild-*' - this would give
everyone a single standard name scheme to follow.
User-defined skipping has been useful for a few people in the past to
allow them to do what they wanted, which is nice. When they were ready
for a final build/test check before setting 'Patch Available', they just
pushed a new branch without the CI-skip name. Done, off it goes running
test builds.
This seems like a pretty easy yaml change that should tick the box for
users that want to push scratch/WIP branches without using testing
resources.
-1 for disabling by default.
Michael
On 3/25/20 11:33 AM, David Capwell wrote:
we should probably just look into further Circle config and/or local
scripting to flip bits in the config on branch creation for people that
want auto testing on push disabled
If there are improvements we can make to the config to allow opt-out, I
am +1 to adding them (not uncommon to flag a PR as Preview or WIP and omit
CI, as long as this gets reverted once ready). If there are ways to change
the defaults in the Circle CI site I am +1 to documenting these steps.
On Wed, Mar 25, 2020 at 8:43 AM Joshua McKenzie <jmcken...@apache.org>
wrote:
Seems there's not an easy consensus. Rather than digging into any data and
reasoning behind the differing opinions that have been shared, we should
probably just look into further Circle config and/or local scripting to
flip bits in the config on branch creation for people that want auto
testing on push disabled. Seems like the fastest path to resolution here.
Any dissent?
On Wed, Mar 25, 2020 at 11:21 AM Jon Haddad <j...@jonhaddad.com> wrote:
I think there's more we can do to customize the circie setup, and am
also a
hard -1 on disabling testing by default. We should figure out a way to
opt
out of it, but the default should be on.
On Wed, Mar 25, 2020 at 7:51 AM Oleksandr Petrov <
oleksandr.pet...@gmail.com>
wrote:
If you just want to disable them for your own branch/branches - I don't
see
any problem with that, I'm just not sure if it's a good default.
In a similar spirit, I know that some folks are running the script to
set
up HIGHRES environment for circle ci every time they create a branch
even
though it's not a default.
On Wed, Mar 25, 2020 at 2:26 PM <e.dimitr...@gmail.com> wrote:
Hi Aleks,
Thanks for those fixes and pointing out this issue.
I didn’t ask for tests not to be used and disabled. I asked whether
we
can
make them not to run on every single push to circleci as they are not
always needed. Intermediate pushes just to save pieces of work do not
require immediate CI run. That was my point.
Ekaterina
Sent from my iPhone
On 25 Mar 2020, at 9:07, Oleksandr Petrov <
oleksandr.pet...@gmail.com>
wrote:
I recently had to fix at least two problems that could've been
prevented by
running tests and checks that are about to be turned off by default
([1]
and [2]).
I'd like to point out that this has happened while checks were
theoretically enabled, and both problems could've been prevented.
This
doesn't seem to be a merge problem, or something that showed up
only
after
the merge.
I might be misunderstanding motivation for this, but my impression
was
that
we, as a community, are striving to be able to have working version
on
every commit merged to master, and possibly even block merging in
case
tests don't pass. It'd be great to hear more about why this could
be
helpful.
[1] Ninja fix: fix eclipse warnings that were broken during
CASSANDRA-15528
<
https://github.com/apache/cassandra/commit/a01d05d9a73211fb91c068e133d78ef8ccf34b4e
[2] Ninja fix: Fix unit tests that were broken during
CASSANDRA-15303.
<
https://github.com/apache/cassandra/commit/b29af2925cddacb4ab8b429b31917748781fbe5d
On Tue, Mar 24, 2020 at 9:01 PM Joshua McKenzie <
jmcken...@apache.org
wrote:
Am I understanding correctly - this isn't disabling tests, just
changing
when they're triggered (i.e. automatic to manual)?
So, for instance, smaller interim commits don't trigger a CI run
and
thus
costs?
On Tue, Mar 24, 2020 at 3:34 PM David Capwell <dcapw...@gmail.com
wrote:
I want to change it so it could be a manual choice whether to do
it
or
not.
Could you explain the motivations for disabling the tests by
default?
My
personal stance is all tests should run (we disable a lot, at
least
HIGHER
should enable all...), not a fan of disabling tests.
On Tue, Mar 24, 2020 at 12:14 PM Ekaterina Dimitrova <
ekaterina.dimitr...@datastax.com> wrote:
Hello everyone,
Hope this email finds you well!
Just a heads up that I plan to open a Jira and change the
CircleCI
config.
Currently unit tests and in-jvm tests are triggered
automatically
on
every
commit.
I want to change it so it could be a manual choice whether to do
it
or
not.
Anyone against that who really needs the current setup? Is there
any
background information I miss?
Ekaterina Dimitrova | Software Engineer
ekaterina.dimitr...@datastax.com | datastax.com
<
http://datastax.com/?utm_campaign=FY20Q2_CONSTELLATION&utm_+medium=email&utm_source=signature
--
alex p
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@cassandra.apache.org
For additional commands, e-mail: dev-h...@cassandra.apache.org
--
alex p
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@cassandra.apache.org
For additional commands, e-mail: dev-h...@cassandra.apache.org
