CINXE.COM

<?xml version="1.0" encoding="UTF-8" standalone="yes"?> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>LKML: Matthias Andree: Re: Kernel SCM saga..</title><link href="/css/message.css" rel="stylesheet" type="text/css" /><link href="/css/wrap.css" rel="alternate stylesheet" type="text/css" title="wrap" /><link href="/css/nowrap.css" rel="stylesheet" type="text/css" title="nowrap" /><link href="/favicon.ico" rel="shortcut icon" /><script src="/js/simple-calendar.js" type="text/javascript"></script><script src="/js/styleswitcher.js" type="text/javascript"></script><link rel="alternate" type="application/rss+xml" title="lkml.org : last 100 messages" href="/rss.php" /><link rel="alternate" type="application/rss+xml" title="lkml.org : last messages by Matthias Andree" href="/groupie.php?aid=29" /><script> var _paq = window._paq = window._paq || []; /* tracker methods like "setCustomDimension" should be called before "trackPageView" */ _paq.push(["setDoNotTrack", true]); _paq.push(["disableCookies"]); _paq.push(['trackPageView']); _paq.push(['enableLinkTracking']); (function() { var u="//m.lkml.org/"; _paq.push(['setTrackerUrl', u+'matomo.php']); _paq.push(['setSiteId', '1']); var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; g.async=true; g.src=u+'matomo.js'; s.parentNode.insertBefore(g,s); })(); </script></head><body onload="es.jasper.simpleCalendar.init();" itemscope="itemscope" itemtype="http://schema.org/BlogPosting"><table border="0" cellpadding="0" cellspacing="0"><tr><td width="180" align="center"><a href="/"><img style="border:0;width:135px;height:32px" src="/images/toprowlk.gif" alt="lkml.org" /></a></td><td width="32">聽</td><td class="nb"><div><a class="nb" href="/lkml"> [lkml]</a> 聽 <a class="nb" href="/lkml/2005"> [2005]</a> 聽 <a class="nb" href="/lkml/2005/4"> [Apr]</a> 聽 <a class="nb" href="/lkml/2005/4/8"> [8]</a> 聽 <a class="nb" href="/lkml/last100"> [last100]</a> 聽 <a href="/rss.php"><img src="/images/rss-or.gif" border="0" alt="RSS Feed" /></a></div><div>Views: <a href="#" class="nowrap" onclick="setActiveStyleSheet('wrap');return false;">[wrap]</a><a href="#" class="wrap" onclick="setActiveStyleSheet('nowrap');return false;">[no wrap]</a> 聽 <a class="nb" href="/lkml/mheaders/2005/4/8/113" onclick="this.href='/lkml/headers'+'/2005/4/8/113';">[headers]</a>聽 <a href="/lkml/bounce/2005/4/8/113">[forward]</a>聽 </div></td><td width="32">聽</td></tr><tr><td valign="top"><div class="es-jasper-simpleCalendar" baseurl="/lkml/"></div><div class="threadlist">Messages in this thread</div><ul class="threadlist"><li class="root"><a href="/lkml/2005/4/6/121">First message in thread</a></li><li><a href="/lkml/2005/4/8/9">Linus Torvalds</a><ul><li><a href="/lkml/2005/4/8/14">Chris Wedgwood</a><ul><li><a href="/lkml/2005/4/8/18">(H. Peter Anvin)</a><ul><li><a href="/lkml/2005/4/8/42">Rogan Dawes</a><ul><li><a href="/lkml/2005/4/8/44">Daniel Phillips</a></li></ul></li></ul></li></ul></li><li><a href="/lkml/2005/4/8/43"> ross@lug ...</a><ul><li><a href="/lkml/2005/4/8/172">Linus Torvalds</a><ul><li><a href="/lkml/2005/4/8/341">Petr Baudis</a><ul><li><a href="/lkml/2005/4/9/19">"Randy.Dunlap"</a></li><li><a href="/lkml/2005/4/9/144">Phillip Lougher</a></li></ul></li><li><a href="/lkml/2005/4/9/70">Paul Jackson</a><ul><li><a href="/lkml/2005/4/9/74">Linus Torvalds</a></li></ul></li></ul></li></ul></li><li><a href="/lkml/2005/4/8/46">Andrea Arcangeli</a><ul><li class="origin"><a href="/lkml/2005/4/8/119">Matthias Andree</a><ul><li><a href="/lkml/2005/4/8/119">Florian Weimer</a></li></ul></li><li><a href="/lkml/2005/4/8/155">Linus Torvalds</a><ul><li><a href="/lkml/2005/4/8/180">Matthias-Christian Ott</a><ul><li><a href="/lkml/2005/4/8/190">Linus Torvalds</a></li><li><a href="/lkml/2005/4/8/315">Marcin Dalecki</a></li></ul></li></ul></li></ul></li><li><a href="/lkml/2005/4/8/48">Marcel Lanz</a><ul><li><a href="/lkml/2005/4/8/82">Geert Uytterhoeven</a></li></ul></li><li><a href="/lkml/2005/4/8/76">Matt Johnston</a></li><li><a href="/lkml/2005/4/12/54">Kedar Sovani</a><ul><li><a href="/lkml/2005/4/12/83">Catalin Marinas</a></li><li><a href="/lkml/2005/4/13/4">Ricky Beam</a></li></ul></li></ul></li></ul></td><td width="32" rowspan="2" class="c" valign="top"><img src="/images/icornerl.gif" width="32" height="32" alt="/" /></td><td class="c" rowspan="2" valign="top" style="padding-top: 1em"><table><tr><td><table><tr><td class="lp">Date</td><td class="rp" itemprop="datePublished">Fri, 8 Apr 2005 14:02:44 +0200</td></tr><tr><td class="lp">From</td><td class="rp" itemprop="author">Matthias Andree <></td></tr><tr><td class="lp">Subject</td><td class="rp" itemprop="name">Re: Kernel SCM saga..</td></tr></table></td><td></td></tr></table><pre itemprop="articleBody">Andrea Arcangeli schrieb am 2005-04-08:<br /><br />> On Thu, Apr 07, 2005 at 09:42:04PM -0700, Linus Torvalds wrote:<br />> > play with something _really_ nasty (but also very _very_ fast), take a<br />> > look at kernel.org:/pub/linux/kernel/people/torvalds/.<br />> <br />> Why not to use sql as backend instead of the tree of directories? That solves<br />> userland journaling too (really one still has to be careful to know the<br />> read-committed semantics of sql, which is not obvious stuff, but 99% of<br />> common cases like this one just works safe automatically since all<br />> inserts/delete/update are always atomic).<br />> <br />> You can keep the design of your db exactly the same and even the command line<br />> of your script the same, except you won't have deal with the implementation of<br />> it anymore, and the end result may run even faster with proper btrees and you<br />> won't have scalability issues if the directory of hashes fills up, and it'll<br />> get userland journaling, live backups, runtime analyses of your queries with<br />> genetic algorithms (pgsql 8 seems to have it) etc...<br />> <br />> I seem to recall there's a way to do delayed commits too, so you won't<br />> be sychronous, but you'll still have journaling. You clearly don't care<br />> to do synchronous writes, all you care about is that the commit is<br />> either committed completely or not committed at all (i.e. not an half<br />> write of the patch that leaves your db corrupt).<br />> <br />> Example:<br />> <br />> CREATE TABLE patches (<br />> patch BIGSERIAL PRIMARY KEY,<br />> <br />> commiter_name VARCHAR(32) NOT NULL CHECK(commiter_name != ''),<br />> commiter_email VARCHAR(32) NOT NULL CHECK(commiter_email != ''),<br /><br />The length is too optimistic and insufficient to import the current BK<br />stuff. I'd vote for 64 or at least 48 for each, although 48 is going to<br />be a tight fit. It costs a bit but considering the expected payload<br />size it's irrelevant.<br /><br />Committer (double t) email is up to 36 characters at the moment and the<br />name up to 43 characters when analyzing the shortlog script with this<br />little Perl snippet:<br /><br />------------------------------------------------------------------------<br />while (($k, $v) = each %addresses) {<br /> $lk = length $k;<br /> $lv = length $v;<br /> if ($lk > $mk) { $mk = $lk; }<br /> if ($lv > $mv) { $mv = $lv; }<br />}<br />print "max key len $mk, max val len $mv\n";<br />------------------------------------------------------------------------<br /><br />which prints: (key is the email, val the name)<br /><br />max key len 43, max val len 36<br /><br />-- <br />Matthias Andree<br />-<br />To unsubscribe from this list: send the line "unsubscribe linux-kernel" in<br />the body of a message to majordomo@vger.kernel.org<br />More majordomo info at <a href="http://vger.kernel.org/majordomo-info.html">http://vger.kernel.org/majordomo-info.html</a><br />Please read the FAQ at <a href="http://www.tux.org/lkml/">http://www.tux.org/lkml/</a><br /><br /></pre></td><td width="32" rowspan="2" class="c" valign="top"><img src="/images/icornerr.gif" width="32" height="32" alt="\" /></td></tr><tr><td align="right" valign="bottom"> 聽 </td></tr><tr><td align="right" valign="bottom">聽</td><td class="c" valign="bottom" style="padding-bottom: 0px"><img src="/images/bcornerl.gif" width="32" height="32" alt="\" /></td><td class="c">聽</td><td class="c" valign="bottom" style="padding-bottom: 0px"><img src="/images/bcornerr.gif" width="32" height="32" alt="/" /></td></tr><tr><td align="right" valign="top" colspan="2"> 聽 </td><td class="lm">Last update: 2005-04-08 14:06 聽聽 [from the cache]<br />漏2003-2020 <a href="http://blog.jasper.es/"><span itemprop="editor">Jasper Spaans</span></a>|hosted at <a href="https://www.digitalocean.com/?refcode=9a8e99d24cf9">Digital Ocean</a> and my Meterkast|<a href="http://blog.jasper.es/categories.html#lkml-ref">Read the blog</a></td><td>聽</td></tr></table><script language="javascript" src="/js/styleswitcher.js" type="text/javascript"></script></body></html>