Hi all,
first of all, I hope this mailing list does not turn back into Shayqeh Ofir's
the English teacher's description of a dialogue:
<<<
I shall explain: Monologue: one person talking to himself ; Dialogue: like
Monologue - two people talking to themselves.
>>>
Anyway, the highlights of what is new on the
trunk/master/default/HEAD/whatever-it-is-called-in-your-version-control-system-of-choice-will-you-make-up-your-mind-already
is:
1. The test suite now works again on 32-bit systems (including
i386/i586/Pentiums - only tested there) after fixing some stuff in the
dbm_fc_solver and elsewhere. I also made sure rpmbuild -tb works on both x86-64
and x86-32 and planning to keep it this way.
Part of the motivation for keeping it this way was to add "make test" to the
%check section of the Mageia .rpm .
Now for an exotic bug. valgrind warned me about this structure generating
uninitialized data upon bcmp()/memcmp() :
typedef struct
{
fcs_encoded_state_buffer_t key;
uintptr_t parent_and_refcount;
#ifdef FCS_EXPLICIT_REFCOUNT
unsigned char refcount;
#endif
} fcs_dbm_record_t;
I wondered what I didn't initialise until I released I did not memset it and
that on 32-bit platforms there was trailing padding. This involved a lot of
investigation and a final "ah-hah" moment.
2. Some cleanups and modernisations (and some anti-MSVCisms) to the code.
3. I have added all the question I had in mind to the F.A.Q.:
http://fc-solve.shlomifish.org/faq.html
Please let me know if you can find any obvious errors.
Regards,
Shlomi Fish
--
-----------------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
Chuck Norris/etc. Facts - http://www.shlomifish.org/humour/bits/facts/
Knuth is not God! Unless you confuse him with Dijkstra.
Please reply to list if it's a mailing list post - http://shlom.in/reply .
Received on Sat Dec 15 2012 - 13:02:54 IST