Path: chuka.playstation.co.uk!scea!sumner From: sumner@austin.metrowerks.com (Joel Sumner) Newsgroups: scee.yaroze.programming.codewarrior Subject: Re: 5 minutes and counting ! Date: Fri, 20 Feb 1998 15:32:13 -0600 Organization: Metrowerks Lines: 38 Message-ID: References: <6ci19m$ar1@chuka.playstation.co.uk> NNTP-Posting-Host: mtwks203.metrowerks.com Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Newsreader: Yet Another NewsWatcher 2.4.0 In article <6ci19m$ar1@chuka.playstation.co.uk>, "Steve Dunn" wrote: >I have the following code in a class > > m_lIndex = 0 ; > > for ( u_long l = 0 ; l < MAX_OBJECTS ; l++ ) > objectTable[ l ] = 0 ; > > othWorld[0].length = ORDERING_TABLE_LENGTH; > >By the way, MAX_OBJECTS = 248000. > >I step to the line that sais objectTable[ l ]=0, then 'step over' that line. [snip] >On a twin P2 300 machine, the debugger took over 5 minutes to come back, and >when it did, the cursor was on the othWorld[0].length = ... line. > >Is this by design ? Uh, well... Umm.. Not intentionally. ;-) The single step algorithm sets a breakpoint, runs, and waits for a breakpoint. The Net Yaroze debugger's algorithm is (as you have seen) very slow when stepping over source lines that contain loops. (it's only looking ahead one branch and thus it is doing the run/breakpoint cycle MAX_OBJECTS times). As you can see, there is room for improvement. One thing I might suggest is to use alt-click to step over multiple instructions. You can alt-click on a tick mark and the debugger will interpret that as a "run until". -Joel -- PlayStation Development Tools Dude Metrowerks