Joseph Ferano
a9a411709c
C-c C-b is a completing-read over restart names, which is the whole UI for the one moment the dev loop exists to make survivable. It shows the names and nothing else, and it will let you pick one that cannot be taken. That last part is a bug, not a gap. §4 says restart lookup takes the first frame offering a name, and flan_find_restart does exactly that; so a second frame offering "retry" is real, is on the list, and is unreachable — picking it sends the string "retry" and the inner frame runs, silently. SBCL has shown this since forever by numbering the restarts and omitting the bracket on a name already used. Taken as is, and the shadowed row now refuses by name and says what would fix it: an index verb, which does not exist. SBCL also decides the order. invoke-debugger prints the condition, then show-restarts, and stops; the backtrace is a command you type. The restarts are the decision and the stack is the explanation for it, and a debugger that opens with forty frames has buried one under the other. What CIDER's stacktrace buffer gives is the behaviour — frames that fold in place, everything on the keyboard. Not its cause chain: a JVM exception wraps another one and a Flan condition wraps nothing. The fields, the stack and the locals are drawn as sections that say why they are empty and what each would take. A section left out cannot be told from one that happened to have nothing in it, and only one of those is a fact about the program.
