(they don't really exist before version 7), so that solution was toast.
Instead, let's do it the way it's done on Unix, but then remove older
versions of the file.
That new mechanism *may* fail for some unixly formated file spec,
although I wouldn't worry too much about it.
this is what we now use to read $RANDFILE / $HOME/.rnd.
(Previously, after 'cat'ting lots of stuff into .rnd
only the first MB would be looked at.)
Bugfix for apps/enc.c: Continue if RAND_pseudo_bytes returns 0
(only -1 is an error).
opening the output file with "wb" to truncate it except on VMS
(where the file now keeps its original length because it is opened
with "rb+" -- does VMS have ftruncate?)
returns int (1 = ok, 0 = not seeded). New function RAND_add() is the
same as RAND_seed() but takes an estimate of the entropy as an additional
argument.
don't try to detect fork()s by looking at getpid().
The reason is that threads sharing the same memory can have different
PIDs; it's inefficient to run RAND_seed each time a different thread
calls RAND_bytes.
between SSLeay 0.8.1b and 0.9.0b with no apparent reason).
If we *want* an error when DEVRANDOM is not defined (it always is with
the current e_os.h) we should use #error.
This will soon be complemented with MacOS specific source code files and
INSTALL.MacOS.
I (Andy) have decided to get rid of a number of #include <sys/types.h>.
I've verified it's ok (both by examining /usr/include/*.h and compiling)
on a number of Unix platforms. Unfortunately I don't have Windows box
to verify this on. I really appreciate if somebody could try to compile
it and contact me a.s.a.p. in case a problem occurs.
Submitted by: Roy Wood <roy@centricsystems.ca>
Reviewed by: Andy Polyakov <appro@fy.chalmers.se>
in cryptlib.h (which is often included as "../cryptlib.h"), then the
question remains relative to which directory this is to be interpreted.
gcc went one further directory up, as intended; but makedepend thinks
differently, and so probably do some C compilers. So the ../ must go away;
thus e_os.h goes back into include/openssl (but I now use
#include "openssl/e_os.h" instead of <openssl/e_os.h> to make the point) --
and we have another huge bunch of dependency changes. Argh.
There were problems with putting e_os.h just into the top directory,
because the test programs are compiled within test/ in the "standard"
case in in their original directories in the makefile.one case;
and in the latter symlinks may not be available.
