Networks CI start only on the tag, but different pipe on each branch, how?

Hi all.

Task:
start pipeline when you create the tag. It is necessary to divide Branches
ie, create a 1.0.0 Tag from a virgin single process
but if Tag 1.0.0 from the master is another

Try:
To respond to a tag added in .networks-ci.yml to each stage
only:
 - /^(\d+\.)?(\d+\.)?(\*|\d+)$/

tags type 1.0.0 and it is working well, but I need to have different tasks depending on the source of the branches, to develop its own, for its own master, etc.

And that's when I got up... found a bunch feature_requests, but all of them are closed without a solution.

Can anyone recommend something?

In the extreme case will be to make a tag, type dev-master and 1.0.0-1.0.0 and catch it in only, but suddenly got a better idea?

11.6.4 CE networks on the LAN

Thanks in advance.

UPD: Yes, I know that commit can be at all branches simultaneously, and that the logic of explanation is not correct. But the challenge is the place to be. Tag a trigger for each branch their tasks...
April 4th 20 at 13:12
2 answers
April 4th 20 at 13:14
1. in pipline with alternating or direct access to the git to determine which branch to execute a specific action. The only problem is that it is not very clear. And such script can be hurriedly hard to write
2. Use advanced syntax networks:
https://docs.gitlab.com/ee/ci/yaml/#rules or https://docs.gitlab.com/ee/ci/yaml/#onlyexcept-advanced
Thanks for the links. Unfortunately, I've already checked in search and the only thing I could help it
CI_COMMIT_BRANCH - but in the case of the tag is empty ;( - nigel commented on April 4th 20 at 13:17
All right. Because the tag can not be both on the master and on the other thread - he did commit. Ie in fact little difference between the branch (branch) and tag there is some floating pointers inside the structure of the tree of commits. Why CI_COMMIT_BRANCH no branches - it just explains it, because you are not guaranteed to determine to what branch is a commit, if more than two. And the trigger worked exactly according to the Assembly tag.... Difficult question how to do the right thing, but it looks like you have a process initially incorrectly designed. And adds fuel to the fire that the networks pulls the repository into the detached mode...... - Dalton_Braun commented on April 4th 20 at 13:20
April 4th 20 at 13:16
For this task you need to broadcast the conditions specified using only and exceptusing double negatives.

https://gitlab.com/gitlab-org/gitlab-foss/issues/2...

Find more questions by tags GitLab