From: CSBVAX::CSBVAX::MRGATE::"SMTP::UUNET.UU.NET::MCVAX!MONCAM!PAUL" 5-DEC-1988 20:17 To: MRGATE::"ARISIA::EVERHART" Subj: extra extra warning #2 Received: from uunet.UU.NET by prep.ai.mit.edu; Mon, 5 Dec 88 17:44:27 EST Received: from mcvax.UUCP by uunet.UU.NET (5.59/1.14) with UUCP id AA05696; Mon, 5 Dec 88 17:52:02 EST Received: by mcvax.cwi.nl via EUnet; Mon, 5 Dec 88 23:38:35 +0100 (MET) Received: from acorn by kestrel.Ukc.AC.UK with UUCP id aa21269; 5 Dec 88 21:05 GMT Received: by acorn.co.uk (5.51/AM10) id AA18754; Mon, 5 Dec 88 20:38:59 GMT Received: by marvin.moncam.uucp (3.2/SMI-3.2) id AA04953; Mon, 5 Dec 88 18:35:06 GMT Date: Mon, 5 Dec 88 18:35:06 GMT From: Paul Hudson Message-Id: <8812051835.AA04953@marvin.moncam.uucp> To: info-gcc@prep.ai.mit.edu Subject: extra extra warning #2 The following diffs cause gcc to give a warning if an expression statement has no effect. This catches (among others) the reverse finger trouble to my previous change, typing == where I meant =. It only objects if the whole expression has no side-effect - it might be argued that f() + 1; should be pointed out. I have a feeling that this would happen too often as a result of macros returing a value which is ignored, but where the whole expression has a side-effect. It could be added to expand_expr if (ignore && !TREE_VOLATILE(exp) && extra_warnings) warning("...."); should do it. Paul Hudson Snail mail: Monotype ADG Email: ...!ukc!acorn!moncam!paul Science Park, paul@moncam.co.uk Milton Road, "Sun Microsysytems: Cambridge, The Company is Arrogant (TM)" CB4 4FQ *** stmt.c.orig Mon Dec 5 18:16:09 1988 --- stmt.c Mon Dec 5 18:17:38 1988 *************** *** 758,763 **** --- 758,765 ---- expand_expr_stmt (exp) tree exp; { + if (expr_stmts_for_value == 0 && !TREE_VOLATILE(exp) && extra_warnings) + warning("statement with no effect"); last_expr_type = TREE_TYPE (exp); last_expr_value = expand_expr (exp, expr_stmts_for_value ? 0 : const0_rtx, VOIDmode, 0);