On 16/04/2017 21:14, Ken Moffat wrote:
On Sun, Apr 16, 2017 at 07:43:00PM +0200, Pierre Labastie wrote:
On 15/04/2017 19:49, Pierre Labastie wrote:
On 15/04/2017 17:37, Roger Koehler wrote:
I just completed a new build of LFS SVN using jhalfs, including
blfs_root, but it bombed out when building scons:
running install_lib
error: [Errno 38] Function not implemented
I tried it manually following BLFS with same error.
I could reproduce it on a laptop with a core2duo, and debian 8 as a host.
Actually, after building LFS, mount virtual devices using the instructions
on the page "Preparing Virtual Kernel Filesystems", chroot using the
instructions on the page "Cleaning up". Then build Python 2.7.13, then run:
---------------------
$ python
[...]
import os
os.urandom(1)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
OSError: [Errno 38] Function not implemented
I only have 64-bit systems now, but your above example seems to work
fine for me (it returned 'q').
---------------------
Note that _there is_ a file /dev/urandom, and that:
---------------
$ hexdump /dev/urandom
---------------
returns random output...
Actually, looking at logs from building Python, I see that this error occurs
while building Python...
But it is not caught because of the "&&"...
I do not have the logs of working installations.
Pierre
Doing an install of scons to /tmp/SCONS appears to work for me. So,
I'd offer you my log, but it is 442KB and I don't think the list
will like that. gzip -9 brings it down to 38K, but that is awkward
to read and quote.
Is there any particular part of it which will be of interest ?
Aside: I don't normally read the log, but it looks stupid:
checking for g++... no
configure: WARNING:
By default, distutils will build C++ extension modules with "g++".
If this is not intended, then set CXX on the configure command line.
So, it claims it didn't find g++ but warns it will use it anyway.
Stupid snake! I suppose that might be very much a homage to Python,
Monty.
The only obvious references to any sort of random are gcc warnings
for -Wunused-function in Python/random.c, I assume the problem you
saw was well before that ?
Also, how does '&&' not catch it ?
And before anyone replies: "I didn't expect the Spanish
Inquisition for disparaging Python".
Lol
First the &&: an instruction such as:
do_this && do_that
in bash does not fail, even if "do_this" fails, because it is taken by bash
as semantically identical to:
if do_this; then do_that; fi
Now, jhalfs extracts its scripts from the book, so all the instructions
(almost) use the &&. For example, on the Python page:
make install && chmod -v 755 /usr/lib/libpython2.7.so.1.0 If "make
install" fails, the above instruction does not return an error code, so
the script does not stop (all the scripts created by jhalfs have "set
-e" at the beginning), and the failure may get unnoticed (specially when
those instructions have been working for ages). Now, the failure with
urandom. I too have the "defined but not used" warning about
dev_urandom_xxx functions, when compiling Python. The failure itself
occurs much later, during installation. Actually, it fails when calling
system functions, and that may be dependent on the host system, since it
occurs while in chroot. I have to build a kernel in order to run them
from a booted system. Pierre
--
http://lists.linuxfromscratch.org/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
