From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 26098 invoked by alias); 15 Aug 2014 17:33:25 -0000 Mailing-List: contact zsh-workers-help@zsh.org; run by ezmlm Precedence: bulk X-No-Archive: yes List-Id: Zsh Workers List List-Post: List-Help: X-Seq: 33022 Received: (qmail 25058 invoked from network); 15 Aug 2014 17:33:14 -0000 X-Spam-Checker-Version: SpamAssassin 3.3.2 (2011-06-06) on f.primenet.com.au X-Spam-Level: X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,RCVD_IN_DNSWL_NONE autolearn=ham version=3.3.2 From: Bart Schaefer Message-id: <140815103330.ZM17468@torch.brasslantern.com> Date: Fri, 15 Aug 2014 10:33:30 -0700 In-reply-to: <20140815093255.GN6176@sym.noone.org> Comments: In reply to Axel Beckert "Re: zsh 5.0.5-dev-2 / Occassional hangs on Test/A05execution.ztst" (Aug 15, 11:32am) References: <20140812212920.67dcb116@pws-pc.ntlworld.com> <20140812223637.GS6176@sym.noone.org> <20140813140401.GX6176@sym.noone.org> <20140814095002.1d8949c5@pwslap01u.europe.root.pri> <140814092527.ZM18021@torch.brasslantern.com> <20140814163124.GK6176@sym.noone.org> <20140814173428.GL6176@sym.noone.org> <140814215053.ZM4188@torch.brasslantern.com> <20140815093255.GN6176@sym.noone.org> X-Mailer: OpenZMail Classic (0.9.2 24April2005) To: zsh-workers@zsh.org Subject: Re: zsh 5.0.5-dev-2 / Occassional hangs on Test/A05execution.ztst MIME-version: 1.0 Content-type: text/plain; charset=us-ascii On Aug 15, 11:32am, Axel Beckert wrote: } } > Actually the warning could be stronger than that, because the } > inability to setpt MONITOR renders the test invalid } } So instead of or in addition to the warning, can we skip this test in } case of no controlling terminal? The test that has the warning and the test that is breaking for you are two different tests. The test that's "invalid" is NOT the one that is causing your issue, so skipping the test that has the warning would not resolve this discussion thread. More specifically: } > (it will always succeed, even in the case we are trying to regress). } } Well, it at least hangs under some yet to be determined conditions. The "will always succeed" statement does NOT apply to the test that is hanging; sorry that wasn't clear enough. The test that is hanging in fact does NOT WANT the monitor option and is *intended* to regress something that happens with no controlling terminal! Which means that maybe it still happens, though why "kill -INT" is not sufficient to end the test may have something to do with the build environment (e.g., perhaps the auto-build blocks the signal because it doesn't want anything it's building to be able to stop the entire automation). } (The one case where it seems to hang for me with a controlling } terminal, I wonder if I was just not patient enough. I think I waited } for at least 20 to 30 seconds before pressing Ctrl-C, but I'm no more } sure.) If you had to wait more than 5 seconds plus some rounding error, you had waited long enough. I've committed a change to add the 6-second timeout on the final read at the end of the test, see if that changes anything ...