> One thing we decided a while back was to put the " (blaming ...)" > near the end, but the blame information is taken into account in that > first line. Perhaps there should be a more refined piece of code that > computes that first line in this case, tho?
On Thursday, NWU-Max gave a bb talk at NEU on blame in the TR/R case. It was more or less a categorical analysis of the Wadler-Findler paper that says “we should always blame both parties involved, plus the spec”, which would mean contract here. (His analysis points out an asymmetry in WF, which is — in his framework — a symptom of a mistake.) It was of course music in my ears to hear this. Christos pointed out that Max’s suggested fix could be generalized so that a contracted value keeps track of all crossed boundaries and the error system could report those crossings to help the developer even more. After all if A is blamed, fixing A may just move the bug to B and then to C and then to D. The cost is high (wrapped information goes). But if we could validate the conjecture practically, it would neat generalize this error message and finally wrap up the first NSF on contracts :) — Matthias -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to racket-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
