> From: Lassi Kortela <xxxxxx@lassi.io> > Date: Sunday, August 16, 2020 4:06 AM > >>> [ Shiro on scoping: ] >>> >>> My intention is rather to split srfi-198 from generic >>> debugging/inspection interface (but not assuming the latter is >>> possible). Specifically, scheme-procedure and args optional (at >>> least for srfi-170). >> >> Ah, yes, that goal makes complete sense. Although we might state a requirement, or it is "strongly suggested", that if the environment doesn't allow you to find out the Scheme procedure throwing an exception, the status object it throws must include it. > I also agree that Shiro's advice is wise. Let's make the procedure and > argument info optional in SRFI 170 (and naturally in 198 as well). My concept is something like this: required by SRFI 198: 'set required by e.g. SRFI 170: ??, but fewer than the current required "strongly suggested" by SRFI 198: 'message, 'foreign-interface, ?? "strongly suggested" by e.g. SRFI 170: ?? "nice to have, and we have a Schemeregistry standard": 'heritage The last is to e.g. tell a user of R7RS-large into which SRFI 170 has been assimilated where he might look to find historical information. For example, the SRFI doc of course does not explain why all of the particular decisions were made about its API. But someone wondering why so many but not all SRFI 170 file system procedures take only a string filename can search the archived discussion and discover the reason was to lessen the burden on implementors using the 80/20 Pareto Principle as a guiding principle. >> [...] - Harold