| commit | 21a43d0caf5c14b3924166c86899d263739bb423 | |
| author | Christophe Rhodes <csr21@cantab.net> | |
| Thu, 30 Apr 2015 11:27:06 +0000 (30 12:27 +0100) | ||
| committer | Christophe Rhodes <csr21@cantab.net> | |
| Thu, 30 Apr 2015 11:27:06 +0000 (30 12:27 +0100) | ||
| tree | e4a51fdb7840de9951b1409feefb0d1b350dfe5f | treesnapshot (tar.gz zip) |
| parent | 8f549215df1741d0ca9c0073ebf76df4e695f267 | commitdiff |
INFINITE-ERROR-PROTECTOR never returns normally
In which case, having it used as
(unless (infinite-error-protector)
...)
only serves to confuse the compiler, making it insert extra
%COMPILE-TIME-TYPE-ERRORs to check that ERROR itself never
returns. These had no particular effect, other than to show up
when I started investigating string context forms in cross-compiled
fasls.
Eliminating all %COMPILE-TIME-TYPE-ERRORs in the cross-compiler
would make a worthy project, complicated a little by the existence
of (ARRAY NIL) objects :-/
In which case, having it used as
(unless (infinite-error-protector)
...)
only serves to confuse the compiler, making it insert extra
%COMPILE-TIME-TYPE-ERRORs to check that ERROR itself never
returns. These had no particular effect, other than to show up
when I started investigating string context forms in cross-compiled
fasls.
Eliminating all %COMPILE-TIME-TYPE-ERRORs in the cross-compiler
would make a worthy project, complicated a little by the existence
of (ARRAY NIL) objects :-/
| src/code/toplevel.lisp | diffblobblamehistory |