Author: njn Date: 2007-09-26 10:43:23 +0100 (Wed, 26 Sep 2007) New Revision: 6912
Log: add notes Modified: branches/MASSIF2/massif/ms_main.c Modified: branches/MASSIF2/massif/ms_main.c =================================================================== --- branches/MASSIF2/massif/ms_main.c 2007-09-26 09:36:11 UTC (rev 6911) +++ branches/MASSIF2/massif/ms_main.c 2007-09-26 09:43:23 UTC (rev 6912) @@ -31,6 +31,16 @@ // XXX: //--------------------------------------------------------------------------- // Todo: +// - make file format more generic. Obstacles: +// - unit prefixes are not generic +// - preset column widths for stats are not generic +// - preset column headers are not generic +// - "Massif arguments:" line is not generic +// - get rid of --stacks, unless/until I do it fully and properly -- ie. +// track every stack alloc/dealloc -- necessary if peak-taking is to be +// accurate. Stacks stuff is hard to regtest, unfortunately. +// - consider 'instructions executed' as a time unit -- more regular than +// ms, less artificial than B // - do a graph-drawing test // - do tests with complicated stack traces -- big ones, ones that require // XCon_redo, ones that exceed --depth, etc. @@ -39,6 +49,7 @@ // documentation // - Check MALLOCLIKE_BLOCK works, write regtest // - work out peak-taking +// - do snapshots on client requests (after peak-taking is done) // - make everything configurable, eg. min/max number of snapshots (which // also determine culling proportion), frequency of detailed snapshots, // etc. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Valgrind-developers mailing list Valgrind-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/valgrind-developers
