feat: expose MigrationBuilder to allow using as sql builder #1285
Conversation
wenerme
changed the title
wenerme
force-pushed
the
features/expose-builder
branch
from
25573f2 to
9df3929
Compare
|
Sorry for the late response, but life... you know What is the purpose of this PR? You wrote about not having file accesses. Potentially we need to go another route and provide an option to not write files and just output to stdio for example 🤔 Also we definitely need tests... |
|
I want to use node-pg-migrate as dsl builder instead migration tool
|
wenerme
force-pushed
the
features/expose-builder
branch
from
9df3929 to
b9a601d
Compare
There was a problem hiding this comment.
I cannot accept this PR right now, because it exposes something fully new I personally don't want to maintain in the long run. Also it would throw errors on a second run when e.g. a select/query would get executed, like calling for a lock.
We would need to implement it in a totally different like adding a new option parameter to RunnerOption and then use the logger to do something like a dry-run.
Line 253 in 2b14645
runner is getting also called from the CLI, so that would also then be a new option to the CLI.
|
Would it make sense to expose this under unstable_createMigrationBuilder and return MigrationBuilder instead of MigrationBuilderImpl for now? If a dry-run mode is implemented for the runner, the builder could later incorporate additional methods for SQL generation. Feel free to close this issue if you feel the timing isn’t right. |
|
I don't feel safe with that approach. Even unstable would be used and required some day and depend on. What I could suggest right now as a quick solution for you: I know this is an unsatisfying solution, but it is a solution 😅 |
Shinigami92
force-pushed
the
features/expose-builder
branch
from
b9a601d to
33a0f48
Compare
Shinigami92
changed the title
fix #1283