From joe at gilith.com Tue Sep 21 17:47:24 2010 From: joe at gilith.com (Joe Hurd) Date: Tue, 21 Sep 2010 10:47:24 -0700 Subject: [metis-users] Welcome! Message-ID: Welcome to the metis-users mailing list! Users of the Metis theorem prover are encouraged to join and discuss bug reports, new features and applications, and anything else relevant to Metis. Cheers, Joe From joe at gilith.com Wed Sep 29 00:39:53 2010 From: joe at gilith.com (Joe Hurd) Date: Tue, 28 Sep 2010 17:39:53 -0700 Subject: [metis-users] New --time-limit command line argument Message-ID: Responding to a feature request made by Andr?s Sicard-Ram?rez, I have implemented a --time-limit N command line argument, which will cause Metis to give up if it cannot complete within N seconds. Metis will return "SZS status Unknown" it if is working on a problem and the time limit kicks in. The feature is implemented in Metis 2.3 (release 20100928), which can be downloaded from http://www.gilith.com/software/metis/ There are some caveats to this feature: 1. The timer is checked every time a clause is removed from the waiting set and added to the active set. This is an expensive operation, so Metis might continue to run past the time limit. 2. Metis has many failure modes beyond running out of time, such as eating up all the memory it can find, so in most circumstances it is preferable to use the ulimit shell command to set limits rather than this command line argument. Cheers, Joe From andres.sicard.ramirez at gmail.com Wed Sep 29 04:36:32 2010 From: andres.sicard.ramirez at gmail.com (=?UTF-8?B?QW5kcsOpcyBTaWNhcmQtUmFtw61yZXo=?=) Date: Tue, 28 Sep 2010 23:36:32 -0500 Subject: [metis-users] New --time-limit command line argument In-Reply-To: References: Message-ID: Hi, On Tue, Sep 28, 2010 at 7:39 PM, Joe Hurd wrote: > Responding to a feature request made by Andr?s Sicard-Ram?rez, I have > implemented a --time-limit N command line argument, which will cause > Metis to give up if it cannot complete within N seconds. Metis will > return "SZS status Unknown" it if is working on a problem and the time > limit kicks in. > @Joe: Thanks for the implementation of this feature. > There are some caveats to this feature: > > 2. Metis has many failure modes beyond running out of time, such as > eating up all the memory it can find, so in most circumstances it is > preferable to use the ulimit shell command to set limits rather than > this command line argument. I agree the ulimit command can be a better option. I am calling Metis from an Haskell application using threads and I had some problems killing the threads using ulimit + Metis. The new --time-limit argument worked with my set-up. All the best, -- Andr?s