luodaint

Jujutsu is trending on the homepage, and the topic is using discipline when dealing with version control. Six months into working agentially on a daily basis, something changed for me.

When each coding session has one specific aim to achieve before the agent starts: one migration, one service, one abstraction layer, the diff itself is already the unit you can review. The whole commit granularity issue went away. No more interactive rebases for me.

Conversely, coding sessions with a large enough scope, implementing dark mode, and fixing the authentication flow across three subsystems will always end up with a diff that is really hard to read, regardless of commit granularity. Inconclusive diff is the result of bad scope, not bad commits.

It’s getting easier to run two agent sessions in parallel over the same codebase. Avoiding them from making inconsistent assumptions, not so much.

My observations: parallel sessions acting on adjacent subsystems won't stay aligned without a common constraint set. The session that assumes the auth invariant will not know that another session just changed a constraint it relies on. The clash won’t manifest at commit time; it will occur at integration time, when the false assumption has already been propagated to three other files.

No approach feels entirely satisfactory. What works for you?

While Canny prices per tracked user rather than per member, their rates start at $99/month for 1,000 tracked users, $359/month for 5,000, and over $900/month for 50,000 users. The catchphrase “unlimited team members” doesn’t mean the price won’t rise as the number of users increases. However, if your goal is an open platform where any user can vote, this pricing model will backfire.

There is also a flat rate option. In my experience, most founders who selected a particular feedback platform did so based on design and not on pricing. Did anyone do the actual math before making a decision?