The article is about spinning up multiple staging servers, maybe one per branch/feature and possibly leaving them there for hours or days, to be inspected at each own URL until somebody confirms that the implementation is OK. All of this is automated (a git push?)
The usual staging and/or preproduction environments would be the preview of their own branches.
It sounds nice. The only problems I can see are:
1) The cost, because each one of those instances could have its own queues, databases, connections with third party APIs, etc.
2) Seeding the database with a meaningful amount of data. I often see preproduction database built by testers month by month after a minimal seeding. Each instance here has to be seeded with all data plus what's required to demo the new feature. That must be saved before destroying the database, reused for further branches, merged with the data for other features.
The usual staging and/or preproduction environments would be the preview of their own branches.
It sounds nice. The only problems I can see are:
1) The cost, because each one of those instances could have its own queues, databases, connections with third party APIs, etc.
2) Seeding the database with a meaningful amount of data. I often see preproduction database built by testers month by month after a minimal seeding. Each instance here has to be seeded with all data plus what's required to demo the new feature. That must be saved before destroying the database, reused for further branches, merged with the data for other features.