From exmh-workers-admin@redhat.com Thu Aug 29 11:03:28 2002 Return-Path: Delivered-To: yyyy@localhost.netnoteinc.com Received: from localhost (localhost [127.0.0.1]) by phobos.labs.netnoteinc.com (Postfix) with ESMTP id 0C98F43F9B for ; Thu, 29 Aug 2002 06:03:24 -0400 (EDT) Received: from phobos [127.0.0.1] by localhost with IMAP (fetchmail-5.9.0) for jm@localhost (single-drop); Thu, 29 Aug 2002 11:03:24 +0100 (IST) Received: from listman.example.com (listman.example.com [66.187.233.211]) by dogma.slashnull.org (8.11.6/8.11.6) with ESMTP id g7T5kvZ00327 for ; Thu, 29 Aug 2002 06:46:57 +0100 Received: from listman.example.com (localhost.localdomain [127.0.0.1]) by listman.redhat.com (Postfix) with ESMTP id CD4923EAF8; Thu, 29 Aug 2002 01:47:09 -0400 (EDT) Delivered-To: exmh-workers@listman.example.com Received: from int-mx1.corp.example.com (int-mx1.corp.example.com [172.16.52.254]) by listman.redhat.com (Postfix) with ESMTP id 0B5373F712 for ; Thu, 29 Aug 2002 01:40:41 -0400 (EDT) Received: (from mail@localhost) by int-mx1.corp.example.com (8.11.6/8.11.6) id g7T5ebR07573 for exmh-workers@listman.redhat.com; Thu, 29 Aug 2002 01:40:37 -0400 Received: from mx1.example.com (mx1.example.com [172.16.48.31]) by int-mx1.corp.redhat.com (8.11.6/8.11.6) with SMTP id g7T5ebY07569 for ; Thu, 29 Aug 2002 01:40:37 -0400 Received: from blackcomb.panasas.com (gw2.panasas.com [65.194.124.178]) by mx1.redhat.com (8.11.6/8.11.6) with SMTP id g7T5PRl05635 for ; Thu, 29 Aug 2002 01:25:28 -0400 Received: from medlicott.panasas.com (IDENT:welch@medlicott.panasas.com [172.17.132.188]) by blackcomb.panasas.com (8.9.3/8.9.3) with ESMTP id BAA23712; Thu, 29 Aug 2002 01:40:22 -0400 Message-Id: <200208290540.BAA23712@blackcomb.panasas.com> To: Chris Garrigues Cc: Brent Welch , Robert Elz , exmh-workers@redhat.com Subject: Re: New Sequences Window In-Reply-To: <1030562749.28110.TMDA@deepeddy.vircio.com> References: <1030562749.28110.TMDA@deepeddy.vircio.com> Comments: In-reply-to Chris Garrigues message dated "Wed, 28 Aug 2002 15:25:47 -0400." From: Brent Welch X-Url: http://www.panasas.com/ X-Face: "HxE|?EnC9fVMV8f70H83&{fgLE.|FZ^$>@Q(yb#N,Eh~N]e&]=>r5~UnRml1:4EglY{9B+ :'wJq$@c_C!l8@<$t,{YUr4K,QJGHSvS~U]H`<+L*x?eGzSk>XH\W:AK\j?@?c1o List-Post: List-Subscribe: , List-Id: Discussion list for EXMH developers List-Unsubscribe: , List-Archive: Date: Wed, 28 Aug 2002 22:40:21 -0700 X-Pyzor: Reported 0 times. X-Spam-Status: No, hits=-10.2 required=7.0 tests=IN_REP_TO,KNOWN_MAILING_LIST,REFERENCES,SPAM_PHRASE_01_02, X_LOOP version=2.40-cvs X-Spam-Level: Well, I've used the check-the-modify-time cache trick for files in many places (not just exmh) so some part of me certainly thinks it is effective. However, it occurred to me that if we do checkpoint state, then aren't we modifying the sequences file for the current folder on every message read? Perhaps we look at the sequences file more than once per message view? Just idle speculation - we can stick in some time calls to find out how expensive things are. Someone asked about increasing the time resolution in the exmh log. We could make that conditional on some support available in 8.3 - Tcl has had "clock seconds" (like gettimeofday) and "clock clicks" (high resolution timer) for some time. But in 8.3 we've calibrated clock clicks values to microseconds. It is still only useful for relative times, but each call to Exmh_Log could emit the microsecond delta since the last log record. Of course, we are measuring all the overhead of taking the log record, etc. I'll try it out. >>>Chris Garrigues said: > > From: Brent Welch > > Date: Wed, 28 Aug 2002 10:32:42 -0700 > > > > > > >>>Robert Elz said: > > > Mh_Sequence also goes and rereads the files (.mh_sequences and the > > > context file) but I'm not sure how frequently that one is called. > > > > In some places I maintain caches of files by checking their modify > time, > > but the sequence files are soo small that by the time you stat them to > > check their date stamp, you could just read them again. > > Do you really think this is true? I added a modify time check thinking > that > it would make an improvement since we were reading it a *lot* more times > in > the new code because we're trying to use the sequences. > > On the other hand, the sequences files are probably being read out of > cache > when that happens anyway. > > Even with a small file, I'd think that the time taken to do a > [file mtime $filename] would be worth it. My code is in proc > MhReadSeqs. > > Chris > > -- > Chris Garrigues http://www.DeepEddy.Com/~cwg/ > virCIO http://www.virCIO.Com > 716 Congress, Suite 200 > Austin, TX 78701 +1 512 374 0500 > > World War III: The Wrong-Doers Vs. the Evil-Doers. > > -- Brent Welch Software Architect, Panasas Inc Pioneering the World's Most Scalable and Agile Storage Network www.panasas.com welch@panasas.com _______________________________________________ Exmh-workers mailing list Exmh-workers@redhat.com https://listman.redhat.com/mailman/listinfo/exmh-workers