Comments on: Estimating Commit Sizes Efficiently http://dirkriehle.com/2009/02/11/estimating-commit-sizes-efficiently/ Dirk Riehle's blog about everything computer science, applied and more Fri, 03 Feb 2012 12:23:41 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Dirk Riehle http://dirkriehle.com/2009/02/11/estimating-commit-sizes-efficiently/#comment-267 Dirk Riehle Fri, 25 Sep 2009 11:37:48 +0000 http://www.riehle.org/?p=635#comment-267 Hi Michel: This is indeed a broad brush analysis. Any application to an individual project only should be done very cautiously. We really only use it on the scale of "all open source projects" which us right now is data for about 10000 projects. Hi Michel: This is indeed a broad brush analysis. Any application to an individual project only should be done very cautiously. We really only use it on the scale of “all open source projects” which us right now is data for about 10000 projects.

]]> By: Michel S. http://dirkriehle.com/2009/02/11/estimating-commit-sizes-efficiently/#comment-266 Michel S. Wed, 23 Sep 2009 20:51:43 +0000 http://www.riehle.org/?p=635#comment-266 Very interesting paper! I wonder whether the regression coefficients would vary, within one project, over time (unfortunately the dataset used in the paper does not appear to allow this analysis). The reasoning being twofold: developer turnover, and changing coding practices. With smaller projects, there is always the problem of enforcing indentation and coding styles too (and, most annoyingly for cross-platform projects, erroneously large diffs due to line ending differences). Very interesting paper! I wonder whether the regression coefficients would vary, within one project, over time (unfortunately the dataset used in the paper does not appear to allow this analysis). The reasoning being twofold: developer turnover, and changing coding practices.

With smaller projects, there is always the problem of enforcing indentation and coding styles too (and, most annoyingly for cross-platform projects, erroneously large diffs due to line ending differences).

]]>