| commit | c6900bba94a721534f4e67fc62ef611da1a5a679 | |
| author | Jan Oravec <jan@fb.com> | |
| Fri, 3 Oct 2014 01:57:27 +0000 (2 18:57 -0700) | ||
| committer | hhvm-bot <hhvm-bot@fb.com> | |
| Sat, 4 Oct 2014 00:00:26 +0000 (3 17:00 -0700) | ||
| tree | d16cf0646f6a5dcb9b1b44b901d75ff0d35b9608 | treesnapshot (tar.gz zip) |
| parent | 522ff52a36bdb462eb2facbc5b568d1dc105ee5e | commitdiff |
Let hhbbc know that invariant_violation always fails
Summary: hhbbc does not know about call_user_func_array(), so when
invariant(condition(), ...);
is rewritten as
if (!condition()) {
invariant(false, ...);
}
it has trouble figuring out that invariant(false, ...) never
falls through.
After this diff, hhbbc figures out correctly that invariant(false, ...)
returns TBottom, but still cannot fully use this knowledge.
Reviewed By: @ptarjan, @elgenie
Differential Revision: D1595071
Summary: hhbbc does not know about call_user_func_array(), so when
invariant(condition(), ...);
is rewritten as
if (!condition()) {
invariant(false, ...);
}
it has trouble figuring out that invariant(false, ...) never
falls through.
After this diff, hhbbc figures out correctly that invariant(false, ...)
returns TBottom, but still cannot fully use this knowledge.
Reviewed By: @ptarjan, @elgenie
Differential Revision: D1595071
| hphp/system/php/lang/invariant.ns.php | diffblobblamehistory |