I just compiled a recent master on a Debian-Trixie machine: apt install -y autoconf autoconf-archive automake autoupdate binutils g++ gcc gettext git libX11-dev libc++ \ libc++-dev libfribidi-dev libgettextpo-dev libimlib2-dev libstdc++-14-dev libstdc++-dev \ libticonv-dev libtool libtoolize libx11-dev libxext-dev libxft-dev libxinerama-dev libxpm-dev \ libxrandr-dev libxrender-dev libxt-dev make pkg-config And with that it just compiles.
https://github.com/fluxbox/fluxbox is the / a git-repository, you can create a PR there. fluxbox.org is referring to that repository. if you host your code / branch on another git-hoster (gitlab, gitea, etc): just tell here where to fetch it.
https://github.com/fluxbox/fluxbox/tree/fix/dont-deliver-sigalarm will be merged as soon as some logistics are cleared. thanks for putting the attention on that patch.
infinite loop in Menu::cycleItems()
https://github.com/fluxbox/fluxbox/pull/70/commits/8a54c1ab241071fa1245b6f75a55b44bbfcb7d86
Good catch! For code clarity I would prefer an extra loop / check upfront if there is any selectable item anyways … similar to the m_items.empty() check. If you dont mind … I ll provide patch, towards github and will provide a thanks to your patch. But very good catch of the issue.
Yah, that's what I have found as well :) Which made me wonder in the first place on why the spec states that every element (as stored in the property) must contain the 0 byte … when existing utility functions already deal with the issue. The original spec (before 1.0) was refering to 0-separated strings … so, n 0-terminated strings for the names … and then represented by a 0-separated list of utf8-encoded strings inside of the property … which is exactly what Xutf8TextListToTextProperty is actually...
_NET_DISPLAY_NAMES does not null-terminate the last name.