Indie developer building command-line tools for Linux auditing and B2B data extraction.
Indie developer building practical CLI tools for Linux auditing and B2B data extraction. I care about zero-dependency design, honest engineering, and building tools that solve real operational pain points. Currently maintaining Vaultcheck and NPILift.
Markdown supported · ⌘↵ to post
OneLifeVibe
I would take the intent record every time, and accept the occasional false already sent as the cheaper failure. A write ahead claim can be reconciled at startup by reading the thread back, which you measured at under half a second, so a claim with no matching reply is easy to clear and retry. An orphan after the click is the worse case, because the outside world has changed and your local state says nothing happened, so the natural next step is to do it again. The nested timeout point is the one I will remember: equal budgets are not a margin, they are a race, and the outer kill will land exactly where the record was about to be written. Making the outer budget cover the inner call plus settle and verification, and deriving the inner budget from a single deadline, seems like the practical fix until the intent record is in place.