Initially, LinkedIn was just another site where you could find jobs. It was simple to use, simple to connect with others; it even had some friendly groups with meaningful discussions.
And then it gained monopoly as the "sole" professional network where you could actually land a job. If you are not on LinkedIn now, you are quite invisible in the job market. Recruiters are concentrated there, even if they have to pay extremely high prices for premium accounts. The site is horrible now: a social network in disguise, toxic and boring influencers, and a lot of noise and bloated interface to explore.
When Google decided to close their code.google.com, GitHub filled a void. It was a simple site powered by git (not by svn or CVS), and most of the major open-source projects migrated there. The interface was simple, and everything was perfect. And then something changed.
GitHub UI started to bloat, all kinds of "features" nobody asked for were implemented, and then the site became a SaaS. Now Microsoft hosts the bulk of open-source projects the world has to offer. GitHub has become a monopoly. If you don't keep your code there, chances are people won't notice your side projects. This bothers me.
Rant over. I hate internet monopolies.
I want a build job to be triggered when a merge request is raised/changed to verify merge requests. Primarily I want it to comment/annotate changes so peer review focusses on logic and warnings are clear.
I can do this with Concourse, Circle, Jenkins and Github Actions on Azure Devops, Bitbucket Cloud, Bitbucket Server & Github. All Gitlab can tell you is pass/fail, which was good in 2003 but seriously lacking in 2023.
Similarly I want the ability to trigger a release and supply a desired version for the release (or someway to achieve that since our projects follow semantic versioning).
The release DSL is incomplete and could not work on server/cloud last time I used it. The page claims it can do alot but there is a hole in it and even the writer clearly knew.
I want the ability to specify multiple reusable pipelines, in a central place. This is not possible in cloud.
Lastly I would like to have multiple potential pipelines in a repository (e.g. smoke test and release). You can hack this in via variables. This will/won't work depending specifically on the runner for your job. if self hosted or cloud you'll notice different parsing behaviour depending on what host it runs on. This is shocking.
I have an email somewhere where I went through every GitLab CI DSL and documented which didn't consistently work, which only worked consistently on cloud and which only worked on server. Also things like release that are broken on both.
The only way to make it work is to use multi stage docker builds and if your doing that build bot and a bash script would be better.
All of what you said seems completely doable to me.
You can. See https://docs.gitlab.com/ee/ci/variables/predefined_variables.html
CI_PIPELINE_SOURCE
You have full access to the API and can do whatever you want in the MR too.
You can, with CI templates. Templates can be in a completely different repository
I do have different pipelines for staging and production in my projects with no issue.
Try to do it and get back to me.
For extra fun try to do it in cloud and server, you'll realise some stuffis server only, some cloud only and the docs don't tell you which one is true
I'm only using the hosted version, it works. I do have a separate gitlab-runner in GCP at the moment though that is working fine.
If something doesn't work for you I suggest creating a ticket?