--- /dev/null
+Return-Path: <tomi.ollila@iki.fi>\r
+X-Original-To: notmuch@notmuchmail.org\r
+Delivered-To: notmuch@notmuchmail.org\r
+Received: from localhost (localhost [127.0.0.1])\r
+ by olra.theworths.org (Postfix) with ESMTP id 1FCE3431FAF\r
+ for <notmuch@notmuchmail.org>; Mon, 22 Oct 2012 07:32:56 -0700 (PDT)\r
+X-Virus-Scanned: Debian amavisd-new at olra.theworths.org\r
+X-Spam-Flag: NO\r
+X-Spam-Score: 0\r
+X-Spam-Level: \r
+X-Spam-Status: No, score=0 tagged_above=-999 required=5 tests=[none]\r
+ autolearn=disabled\r
+Received: from olra.theworths.org ([127.0.0.1])\r
+ by localhost (olra.theworths.org [127.0.0.1]) (amavisd-new, port 10024)\r
+ with ESMTP id 0Hra137KXHE7 for <notmuch@notmuchmail.org>;\r
+ Mon, 22 Oct 2012 07:32:54 -0700 (PDT)\r
+Received: from guru.guru-group.fi (guru.guru-group.fi [46.183.73.34])\r
+ by olra.theworths.org (Postfix) with ESMTP id 8D220431FAE\r
+ for <notmuch@notmuchmail.org>; Mon, 22 Oct 2012 07:32:54 -0700 (PDT)\r
+Received: from guru.guru-group.fi (localhost [IPv6:::1])\r
+ by guru.guru-group.fi (Postfix) with ESMTP id CF8F410036F;\r
+ Mon, 22 Oct 2012 17:32:53 +0300 (EEST)\r
+From: Tomi Ollila <tomi.ollila@iki.fi>\r
+To: Jani Nikula <jani@nikula.org>,\r
+ Ethan Glasser-Camp <ethan.glasser.camp@gmail.com>, notmuch@notmuchmail.org\r
+Subject: Re: [PATCH] lib: fix warnings when building with clang\r
+In-Reply-To: <87vce3vf5q.fsf@nikula.org>\r
+References: <1349076971-2065-1-git-send-email-jani@nikula.org>\r
+ <87pq4c61hc.fsf@betacantrips.com> <87vce3vf5q.fsf@nikula.org>\r
+User-Agent: Notmuch/0.14+51~g62cd13b (http://notmuchmail.org) Emacs/24.2.1\r
+ (x86_64-unknown-linux-gnu)\r
+X-Face: HhBM'cA~<r"^Xv\KRN0P{vn'Y"Kd;zg_y3S[4)KSN~s?O\"QPoL\r
+ $[Xv_BD:i/F$WiEWax}R(MPS`^UaptOGD`*/=@\1lKoVa9tnrg0TW?"r7aRtgk[F\r
+ !)g;OY^,BjTbr)Np:%c_o'jj,Z\r
+Date: Mon, 22 Oct 2012 17:32:53 +0300\r
+Message-ID: <m2lieyk25m.fsf@guru.guru-group.fi>\r
+MIME-Version: 1.0\r
+Content-Type: text/plain\r
+X-BeenThere: notmuch@notmuchmail.org\r
+X-Mailman-Version: 2.1.13\r
+Precedence: list\r
+List-Id: "Use and development of the notmuch mail system."\r
+ <notmuch.notmuchmail.org>\r
+List-Unsubscribe: <http://notmuchmail.org/mailman/options/notmuch>,\r
+ <mailto:notmuch-request@notmuchmail.org?subject=unsubscribe>\r
+List-Archive: <http://notmuchmail.org/pipermail/notmuch>\r
+List-Post: <mailto:notmuch@notmuchmail.org>\r
+List-Help: <mailto:notmuch-request@notmuchmail.org?subject=help>\r
+List-Subscribe: <http://notmuchmail.org/mailman/listinfo/notmuch>,\r
+ <mailto:notmuch-request@notmuchmail.org?subject=subscribe>\r
+X-List-Received-Date: Mon, 22 Oct 2012 14:32:56 -0000\r
+\r
+On Sun, Oct 21 2012, Jani Nikula <jani@nikula.org> wrote:\r
+\r
+> On Sun, 21 Oct 2012, Ethan Glasser-Camp <ethan.glasser.camp@gmail.com> wrote:\r
+>> Jani Nikula <jani@nikula.org> writes:\r
+>>\r
+>>> Building notmuch with CC=clang and CXX=clang++ produces the warnings:\r
+>>>\r
+>>> CC -O2 lib/tags.o\r
+>>> lib/tags.c:43:5: warning: expression result unused [-Wunused-value]\r
+>>> talloc_steal (tags, list);\r
+>>> ^~~~~~~~~~~~~~~~~~~~~~~~~\r
+>>> /usr/include/talloc.h:345:143: note: expanded from:\r
+>>> ...__location__); __talloc_steal_ret; })\r
+>>> ^~~~~~~~~~~~~~~~~~\r
+>>> 1 warning generated.\r
+>>>\r
+>>> CXX -O2 lib/message.o\r
+>>> lib/message.cc:791:5: warning: expression result unused [-Wunused-value]\r
+>>> talloc_reference (message, message->tag_list);\r
+>>> ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\r
+>>> /usr/include/talloc.h:932:36: note: expanded from:\r
+>>> ...(_TALLOC_TYPEOF(ptr))_talloc_reference_loc((ctx),(ptr), __location__)\r
+>>> ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\r
+>>> 1 warning generated.\r
+>>>\r
+>>> Check talloc_reference() return value, and explicitly ignore\r
+>>> talloc_steal() return value as it has no failure modes, to silence the\r
+>>> warnings.\r
+>>> ---\r
+>>> lib/message.cc | 4 +++-\r
+>>> lib/tags.c | 2 +-\r
+>>> 2 files changed, 4 insertions(+), 2 deletions(-)\r
+>>>\r
+>>> diff --git a/lib/message.cc b/lib/message.cc\r
+>>> index 978de06..320901f 100644\r
+>>> --- a/lib/message.cc\r
+>>> +++ b/lib/message.cc\r
+>>> @@ -788,7 +788,9 @@ notmuch_message_get_tags (notmuch_message_t *message)\r
+>>> * possible to modify the message tags (which talloc_unlink's the\r
+>>> * current list from the message) while still iterating because\r
+>>> * the iterator will keep the current list alive. */\r
+>>> - talloc_reference (message, message->tag_list);\r
+>>> + if (!talloc_reference (message, message->tag_list))\r
+>>> + return NULL;\r
+>>> +\r
+>>> return tags;\r
+>>> }\r
+>>\r
+>> Hi! What you did with talloc_steal is obviously fine. \r
+>>\r
+>> I'd be happier about what you did with talloc_reference() if there were\r
+>> precedent, or a clearly-articulated convention for notmuch. Instead this\r
+>> is the third use in the codebase that I can see, and the other two are\r
+>> each unique to themselves. In mime-node.c we print an "out-of-memory"\r
+>> error and in lib/filenames.c we cast (void) talloc_reference (...), I\r
+>> guess figuring that we're pretty much hosed anyhow if we run out of\r
+>> memory.\r
+>>\r
+>> Why return NULL here? It seems like if talloc_reference fails, we're\r
+>> going to crash eventually, so we should print an error to explain our\r
+>> impending doom. I'd guess you're uneasy printing anything from lib/, but\r
+>> still want to signal an error, and the only way you can do so is to\r
+>> return NULL. I guess that silences the compiler warning, but it's not\r
+>> really the correct way to handle the error IMO. On the other hand, it's\r
+>> such a weird corner case that I don't even think it merits a FIXME\r
+>> comment.\r
+>>\r
+>> How about an assert instead of a return NULL?\r
+>\r
+> No. I don't think a library should assert, exit, or print to stderr on\r
+> this sort of thing. It's up to the calling application. Even if it\r
+> probably doesn't have many choices left, given how much memory\r
+> talloc_reference needs (not much).\r
+>\r
+> Ignoring the talloc_reference return value with (void) is just wrong,\r
+> and the caller of notmuch_message_get_tags should anticipate a NULL\r
+> return. So IMHO that's the pragmatic thing to do in this mostly\r
+> theoretical situation, the biggest change being silencing the warning.\r
+\r
+I agree that the best library can do is to return NULL (if talloc had\r
+a place in ctx to store error indication that could be used but I did\r
+not see any in quick look -- and using global there is not a good idea)\r
+\r
+but, before returning NULL should 'tags' be freed.\r
+\r
+Additionally, should lib/filenames.c be changed to have code:\r
+\r
+if (unlikely (talloc_reference(filenames, list) == NULL)) {\r
+ talloc_free (filenames);\r
+ return NULL;\r
+}\r
+\r
+> BR,\r
+> Jani.\r
+\r
+(btw, what are the chances that program crashes before returning NULL\r
+due to page fault in stack frame allocation ???)\r
+\r
+Tomi\r
projects / notmuch-archives.git / commitdiff