In my last BorgBackup post I described my new setup using BorgBackup for the backup of my private data. After two weeks of use, I have now first experiences on the way the incremental backups perform and some additional information how I apply this to backup a virtual FreeBSD server hosting some of my stuff. Incremental Backup Performance ;=) I must confess, I am positively impressed. As told before, my duplicity setup did have a very bad performance (even just in the locally to my NAS).

Backup is always something you under-appreciate until you need it. Therefore, since I lost a lot of my mails due to non-existing backups in the past, I tried to keep proper backups around the last decade. At least this allowed me to keep all my Kate related development mails intact since 2001 or so, not that bad ;=) My old way of doing backups: rsync & duplicity So far, I did my backup on my Unix machines by using rsync to a second host and duplicity to archive stuff on my old Synology NAS.

The KDE e.V. is a registered non-profit organization that represents the KDE Community in legal and financial matters. The KDE e.V. is for example responsible for paying the servers that run our GitLab instance and all our other web services. The e.V. takes care of sponsoring developer sprints and contributor travel costs, too. You did participate at some Akademy? This wouldn’t have been possible without the KDE e.V., both by sponsoring and helping to organize the event!

I tried to compile the Picolibc that uses the Meson build system with CompCert. The CompCert install I tried uses the GNU/GCC toolchain for preprocessing and linking. That makes the compiler relatively compatible to a GCC, beside that not all command line options are supported. I was used to CMake projects and there it is normally enough to just set CC=ccomp and you are ready and set (at least if you use the GNU/GCC toolchain).

I bought a MacBook end of 2014. My initial reason to buy it was to improve the Kate port for macOS. Beside that, I wanted to try if the Apple ecosystem and macOS are really that great and will solve all my issues in life (short answer: no, they aren’t, at least not for me). The HiDPI screen looked nice, too :=) After some initial “not a lot works”, if you don’t go the Homebrew or MacPorts way, the Kate port improved over the following years.

At work we use the LLVM/clang libraries for pre-processing and parsing C/C++ stuff. During some experimenting with the current master version of these libraries, I stumbled on unexpected crashes inside our unit-tests on Windows. We work with MSVC 2019 on Windows and all worked fine with LLVM 9.x, but with master, close to all my tests did now segfault in aligned_free. I first thought that might be some current regression in LLVM master, but after tracing it back to having DenseSet/Map crashing during destruction, this seemed unlikely.