Hacker News new | past | comments | ask | show | jobs | submit login

I don't know of anybody doing that - unless you had automated checks that it was a fast and non-destructive operation, that would be highly risky.

I'm simply talking about using the current production database as a basis for generating the required changes, rather than "here's what I think my production database currently looks like based on a version number and these 75 chained migration scripts I've run on it in the past"




I'm still confused. When do you generate your diff then?

I think migrations ought to be done by the developer at the same time that they write their other code changes, because they have all the context at that time.

However, it doesn't seem like that could work for diffs, because the version of the production database you're diffing against may not be the same as the version you end up deploying to.


Generate it whenever you prefer, probably with the rest of the code changes as you say.

Immediately before applying, check that the database is still in the same state as it was when you generated the diff script. If yes, apply. If no, abort.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: