[MIT-Scheme-devel] failure while re-reading microcode tables

[Taylor R Campbell]

If the user has no permission to read some directory in the pathname
returned by the MICROCODE-TABLES-FILENAME primitive, the procedure
RE-READ-MICROCODE-TABLES! fails, even if the subsequent microcode
identification would match.  This is because FILE-EXISTS? signals an
error when it can't tell whether the file exists.  Using FILE-EXISTS?
in the first place leads to a race condition anyway, but probably we
don't care much about that.

Is it safe to use the condition system in RE-READ-MICROCODE-TABLES!,
or should there be a variant of FILE-EXISTS? that means not so much
`does this file exist?' but `give me #T if this file exists, and #F
for any other state of affairs'?