Re: BGL leaky?

From: Vladimir Prus (yg-boost-users_at_[hidden])
Date: 2003-05-06 01:41:03

Hi Martin,

Martin Okrslar wrote:

> Dear Boost users,
> 
> is the BGL leaky? For example the code added below produces memory leaks
> (as seen from also added valgrind output).
> When I comment out the line marked by /* THIS LINE DOES IT */, there are
> no leaks. I am running BGL 1.29 and gcc 3.2.
> 
> Any ideas for a work-around?

[...]

> ==23473== definitely lost: 0 bytes in 0 blocks.
> ==23473== possibly lost:   0 bytes in 0 blocks.
> ==23473== still reachable: 384 bytes in 1 blocks.
> ==23473==
> ==23473== 384 bytes in 1 blocks are still reachable in loss record 1 of 1
> ==23473==    at 0x4004A3A8: __builtin_new (vg_clientfuncs.c:125)
> ==23473==    by 0x4004A3F6: operator new(unsigned) (vg_clientfuncs.c:138)
> ==23473==    by 0x4035E4F1: ??? (bits/stl_alloc.h:482)
> ==23473==    by 0x4035E40D: ??? (bits/stl_alloc.h:532)

Strictly speaking, I'm not sure this is memory leak. If there non-zero
values in "definitely list" column, it's surely a leak. But here it can be
something reasonable. For example, memory allocator may reserve a small
chunk of memory. I think there's is not problem here.

- Volodya