Re: [Chicken-users] SRFIs 34, 35 and 36

[William Annis]

 >To be frank, I don't particularly like SRFI-34. The semantics of `raise'
 >are (IMHO) unnecessarily complicated, and
 >the whole idea of condition-type defining macros a la ML is not my
 >idea of the "Scheme Way" (of which exist about as many as Scheme
 >implementors ;-).

        Well, I don't particularly care which one is available, though
I favor accepted standards by instinct when I have nothing else to go
on.  I just want more flexibility in error handling and reporting than
dynamic-wind offers.

        It looks like only SISC currently surrorts 34, or the SRFI
implementations page hasn't been updated in a while.

 >> SRFI 36, which requires conditions for I/O operations will take more
 >> work, but would be incredibly useful.  I've looked at library.scm and
 >> posix.scm a bit, and some of the conditions would be reasonably
 >> straightforward to add by consulting errno after ##sys#update-errno.
 >> But there may be hidden problems I've not noticed yet.
 >
 >Yes, most errors triggered by the system have a tag that can be
 >used to create specific condition objects for specific error situations.

        I'll have to dig into that more.  My python programs are
regularly full of exception handling code, and I'm fond of that
model.  Bare scheme is a horror if your program ever talks to a file.
So much can go wrong!

--
wm