On Sat, Sep 26, 2009 at 5:24 PM, Gennadiy Rozental wrote:

Василий Старостин wrote:

...

So I expected that Boost.Test can use this function return code and increase errors counter for leaks. If it cannot, this can be improvement for the future.

The problem is that this function is not invoked till after testing framework is done. in fact I do not invoke it at all. It is invoked by the CRT once program execution is completed.

Running this function earlier most probably will produce huge number of false positives.

For note, the google memory tracker allows you to even partition segments of your program off, letting you test for memory leaks just from those sections and so forth, it is very powerful, as well as multi-platform.

On Sun, Sep 27, 2009 at 12:01 AM, Gennadiy Rozental wrote:

OvermindDL1 wrote:

...

For note, the google memory tracker allows you to even partition segments of your program off, letting you test for memory leaks just from those sections and so forth, it is very powerful, as well as multi-platform.

Can you point to docs/download location? How portable is it? At what point it performs instrumentation?

Main site: http://goog-perftools.sourceforge.net/ The memory allocator and leak checker is very portable as I recall, the profiler is not as portable (but you do not need it from the sounds of it), but they are working on it. This is what Google actually uses in their own products and servers, very well tested. It will report results either at program end, or at any specific time you specify, as stated, it is very powerful, and very easy to use, I have been quite happy with it, and the memory allocator has speed up some of my memory intensive programs by a surprising amount. The code is actually very generic, it could actually, quite easily, become a boost library (if Google would ever offer it up as such), maybe with some namespace changes and such though.