Today I released v1.2.6 of ghc::filesystem
. The main reason was an
issue with fs::path::preferred_seperator
. The previous implementation didn’t compile on pre-C++17 compilers when the seperator was accessed. The Bugfix for #27, was a bit more work than expected, but now it works on all tested compilers.
Additionally, two pull requests found their way into this release, and I want to thank the supporters for their work.
Read more...
With the initial release of a working prototype of a new reLive streaming client I started to
work on reLive again. The first result is a curses based console application for macOS and
Linux. It is still a bit rough around the edges, but it fetches all metadata, has a database
backend and plays shows.

Read more...
Today I released v1.2.4 of ghc::filesystem
.
It mainly contains a fix for issue #22,
where ghc::filesystem::copy()
wouldn’t allways forward given copy_options
to
a ghc::filesystem::copy_file()
call.
Read more...
My previous server, based on Intel’s 8-Core-Atom C2750, was handling my mails,
domains and repositories for about five years. Most of the time, I was quite
happy with the system, but still, time passes and hardware ages fast.
When last month the system dropped offline and was unreachable even through
the remote serial connection, I was kind of shocked even with a relativly
recent backup.
I called my hosting provider IONOS, and they where kind enough to find the issue
(a dead mainboard) and replace the system, keeping my RAID1 HDDs, and getting
everything up and running again in about an hour, and that on a sunday night.
So thank you to all the unnamed service technicians that where involved in this!
Read more...
Today I released a bugfix version of ghc::filesystem
.
It only contains a fix for issue #21,
where compiling on Alpine Linux with musl
instead of glibc
chooses the wrong strerror_r
signature.
The musl
headers sadly don’t conform to the preprocessor expression of the linux man
file (_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600) && ! _GNU_SOURCE
for strerror_r
, so there was no well defined preprocessor way of solving this.
Read more...