Re: [jamboost] bjam -d0 regression

From: Rene Rivera (grafik666_at_[hidden])
Date: 2002-11-13 10:28:26

[2002-11-12] Vladimir Prus wrote:

>
>Rene, this checkin of yours introduces a regression:
>
>2002-08-09 09:14 grafik
>
> * make1.c: This is a reworking of make1 to be non-recursive. Pros:
> 1) Removes bug in stack limited platforms of blowing the stack
> limits. Like win32.tLiek win32. 2) Slightly faster (10% or so).
> Cons: 1) More heap usage.
> 
> The code is from Sta
>
>Compare this outputs:
>
>bash-2.05a$ /tmp/bjam -d0 run-add
>run-program run-add
>no errors detected
>You have mail in /var/mail/ghost
>bash-2.05a$ /tmp/bjam -d0 run-add
>no errors detected
>
>The extra line in first output should not be there.
>
>The offending diff hunk is:
>
>@@ -286,9 +436,8 @@
> /* from all the actions then report our completion to all the */
> /* parents. */
>
>- if( cmd && t->status == EXEC_CMD_OK )
>+ if( cmd && pState->t->status == EXEC_CMD_OK )
> {
>- if( DEBUG_MAKE )
> if( DEBUG_MAKEQ || ! ( cmd->rule->actions->flags &
RULE_QUIETLY ) )
> {
> printf( "%s ", cmd->rule->name );
>
>I believe this condition should be
>
> if (DEBUG_MAKEQ || ! (cmd->rule->actions->flags & RULE_QUIETLY) &&
DEBUG_MAKE )
>
>but not sure. What do you think?

If the precedence makes it:

( DEBUG_MAKEQ || ( ( ! (cmd->rule->actions->flags & RULE_QUIETLY) ) &&
DEBUG_MAKE ) ) )

Then yes, that looks right.

-- grafik - Don't Assume Anything
-- rrivera_at_[hidden] - grafik_at_[hidden]
-- 102708583_at_icq - Grafik666_at_AIM - Grafik_at_[hidden]