2012-05-21 I really hate git. It does not scale down at all to small pro- jects. It took me hours to get the simple setup for http://git.marmaro.de to the present state. It's still worse than the mercurial counter-part http://hg.marmaro.de. The latter setup contains less configuration, took less time to set up and is more convenient to use. Git is prefectly suited for Kernel development, and for that it was written. For a many other kinds of project structures, it is a bad choice. I investigate into collectd and munin, cause I like to have such a tool on my server. I'm conviced of neither one, but collectd appears to be the better choice. I'll defer this decision. pdfseparate(1), which is part of poppler, is badly designed. Its interface is horrible. But it's still a better choice than ins- talling Java for pdftk. The companion pdfunite(1) is not as bad. Using them in combina- tion generates results multiple times as large as the source PDF although pages were removed. In doube-converting the result through pdftops(1) and ps2pdf(1), a final result of appropriate size is produced. One should avoid pdf2ps(1) as that variant doesn't handle fonts as well as pdftops(1). I suggested the acronym `TSS' for `Time-Sharing System' to mira- bilos for the MirBSD WTF database. My application for a talk on mmh at Entropia's GPN [0] was ac- cepted today, but unfortunately for Sunday when I cannot be there. I hope it is possible to move it to another day. I real- ly like to tell the community about mmh. [0] http://entropia.de/GPN12 http://marmaro.de/lue/ markus schnalke