{"thread":{"id":"12","subject":"Trying base32","startedAt":"2005-04-14T07:26:47Z","lastAt":"2005-04-14T07:26:47Z","messageCount":1,"participants":["H. Peter Anvin"],"isPatch":false,"patchVersion":null,"patchTotal":null},"messages":[{"id":"48","messageId":"425E1B37.20408@zytor.com","threadId":"12","inReplyTo":null,"subject":"Trying base32","fromName":"H. Peter Anvin","fromEmail":"hpa@zytor.com","sentAt":"2005-04-14T07:26:47Z","receivedAt":"2005-04-14T07:26:47Z","isPatch":false,"sender":{"key":"hpa@zytor.com","avatar":null},"body":"Out of curiosity, I cooked up a version of git which used a flat base32, \ninstead of base64, repository.  The goal of this is, of course, to \nprovide for compatibility with case-deficient filesystems.\n\nThe answer is that the base32 format is measurably slower than base64, \nwith a cache-hot checkout time of 19.19 seconds (as compared to 15.68 \nfor base64.)  So the length of the filenames does matter.\n\nAnyway, I have uploaded the base32 patch onto kernel.org, and I have \nmoved all the git patches into:\n\nftp://ftp.kernel.org/pub/linux/kernel/people/hpa/git/\n\nAt this point, though, I would recommend the base64-flat patch.\n\n\t-hpa\n"}]}