> Hi Craig, Sergey, et.al. Hi Jim. > > Attached are 2 additional patches to the library, both dealing with the > recent file2str dynamic buffer change. > > The first one just makes that routine more efficient by reducing the number > of read() and xrealloc() calls. Jim, please, do more simulations. This is not more efficient. For example for sizes from 5152 to 8192 the new code actually does one extra reallocation in compare to the current code. --- TESTING 8191 xcalloc size = 4096 xrealloc size = 5151 tot_read = 4096 xrealloc size = 10270 tot_read = 8191 result = 8191 strlen = 8191 --- Even if you reuse such block of memory, the benefit is uncertain and the code looks strange. I believe you focus too much on stuff that is unimportant. IMO there are other problems in the code which need our attention. Regards, Jaromir.