You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Long lists of revisions are pretty cumbersome to reason about, search, etc., especially if I continue to enforce the "no downgrades" thing. It's almost certain that, to get to a final revision state for any given feature, there are likely to be at a minimum 5+ revisions. Even if downgrades are introduced, there are still likely to be several.
And given that revisions are often logically related (ie. a feature release might include 3+ revisions to get to a "final" state), it might make sense to bundle those together.
One-off revisions should still be allowed, I think, because...
It's conceivable that there often can be single-revision "features" (or tasks)
Backwards compatibility
Some questions...
Should the tongue-in-cheek language be even more try-hard?
ie. revisions can be in "trips" with "steps", or they can be one-off "errands"?
What should generating a new revision do by default?
Should hashing and comparison change?
How to decide which "trip" a revision should be generated for
The latest one by default?
Or, especially if merging new work into a branch (which could include a later "trip"), always require a trip id?
One novel characteristic of this is that, by viewing the steps in a given trip in relationship to one another, it can be easier to see how a schema changes for a given feature. Additionally, without downgrades, it really clearly shows the iteration within a feature that upgrade/downgrade/repeat cycles can't, but that might not really be a feature (particularly since successive revisions in a 'trip' should be expanding on, rather than fixing, previous steps)
The text was updated successfully, but these errors were encountered:
Long lists of revisions are pretty cumbersome to reason about, search, etc., especially if I continue to enforce the "no downgrades" thing. It's almost certain that, to get to a final revision state for any given feature, there are likely to be at a minimum 5+ revisions. Even if downgrades are introduced, there are still likely to be several.
And given that revisions are often logically related (ie. a feature release might include 3+ revisions to get to a "final" state), it might make sense to bundle those together.
For example, rather than a structure like...
... there could be something like...
One-off revisions should still be allowed, I think, because...
Some questions...
One novel characteristic of this is that, by viewing the steps in a given trip in relationship to one another, it can be easier to see how a schema changes for a given feature. Additionally, without downgrades, it really clearly shows the iteration within a feature that upgrade/downgrade/repeat cycles can't, but that might not really be a feature (particularly since successive revisions in a 'trip' should be expanding on, rather than fixing, previous steps)
The text was updated successfully, but these errors were encountered: