Date: Thu, 9 Jul 2009 22:37:02 -0500 From: Mark Linimon <linimon@lonesome.com> To: Ion-Mihai Tetcu <itetcu@freebsd.org>, Paul Schmehl <pschmehl_lists@tx.rr.com>, freebsd-ports@freebsd.org Subject: Re: can you PLEASE _read_ the QAT mails? (was: Re: Question about a failure report) Message-ID: <20090710033702.GA20975@lonesome.com> In-Reply-To: <20090709202550.GD2145@acme.spoerlein.net> References: <23B9298D1A02526B4646B131@utd65257.utdallas.edu> <E1MMoEv-0000YG-D3@daland.home> <91D041090FA2939A0942293F@utd65257.utdallas.edu> <E1MMp3f-0000hV-0m@daland.home> <4C41B4EAEF8C942A8D085916@utd65257.utdallas.edu> <994EF1F5-0A23-4A4F-B853-48AED6A5529F@tandon.net> <20090705111532.61c2dd65@it.buh.tecnik93.com> <4ED7FF030DD8AEF9F1B2A039@Macintosh-2.local> <20090706005148.15d15cd1@it.buh.tecnik93.com> <20090709202550.GD2145@acme.spoerlein.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Jul 09, 2009 at 10:25:50PM +0200, Ulrich Spörlein wrote: > I am always confused about what portsmon is doing vs. pointyhat. portsmon tracks all the errorlog data from pointyhat, with the caveat that only the last 2 runs (somewhat erroneously termed "full" and "latest") of each buildenv are tracked. e.g., if package foo failed more than 2 runs ago and has not be re-run since, portsmon will not show a build error for it. It will, however, show you the last package that was uploaded. portsmon's idea of metadata for a port is based on a single buildenv -- this is a limitation of the database schema ATM. The default buildenv that is used is i386-7 right now*, but I should be switching it to i386-8 pretty soon. This means there is no way to ask it "show me packages that fail on amd64-6" or "show me ports that are marked BROKEN on sparc64, for instance. Therefore, all of the periodic emails describe the state of a port only on i386-7. Finally, since portsmon is looking at errorlogs that have already been produced, it to some degree is reporting the state of past ports tree(s). However, its metadata is updated every hour. mcl * careful observers will note that the environment in which the metadata are evaluated may not _exactly_ match pointyhat's setup, but it's as close as I can get it. The last time I checked, this affected only a handful of ports; in particular, a few false positives for IGNORE based on conflicts.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090710033702.GA20975>