10 Dec
2006
10 Dec
'06
6:21 p.m.
I am not a boost iostreams expert... just a user. Having said that, if iostreams::read() can return -1 during the copy_impl then it looks like the original code could be a problem.
and it happens... it is indeed the test used to detect the end of the loop...
What was your test case to demonstrate this?
was something like
short *data = new short[w*h];
for(int index_ligne=0;index_ligne