Recently Qt6-related ports are increasing and currently textproc/fcitx-qt6 doesn't exist. Fortunately, textproc/fcitx-qt5 already has build time cmake knob for Qt6 support as ENABLE_QT6 which builds plugin file only. Unfortunately, the knob doesn't seem to work when ENABLE_LIBRARY and ENABLE_QT5, which current textproc/fcitx-qt5 enables. FLAVOR'ize doesn't fit in this case, and actually, I couldn't make my prototype patch working with poudriere jail. If I understand correctly, subpackage would not fit too, as build didn't work well with all 3 knobs enabled at once.
Created attachment 247504 [details] Patch to create and support Qt6 slave port Created patch to support Qt6 slave port, slave port itself, and hook the slave port to build. Not sure this is actually enough (possibly missing something) and possibly Qt5 libraries are required even if there's no applications installed. Tested on stable/14, amd64 with some Qt5 applications and master port (textproc/fcitx-qt5) installed, additionally install (to test) graphics/heimer and its dependencies and tested fcitx can be enabled or not, input some Japanese text using japanese/mozc-server as backend in conjunction with japanese/fcitx-mozc and it worked for me. Mouse cursor became very small when over heimer window, but it should be problem of Qt6 framework and not of fcitx-qt5 port.
Yuri, this patch seems fine and poudriere test passed for textproc/fcitx-qt5 and textproc/fcitx-qt6. Do you want to commit this one? Tomoaki, do you have a plan to upgrade to fcitx5 which already has fcitx5-qt6 flavor.
Why keep Qt5? It should be just switched to Qt6 without slave ports or flavors, unless there are known problems or differences between Qt5 and Qt6 flavors.
(In reply to Yuri Victorovich from comment #3) fcitx-qt5 is not "fcitx implemented in qt5", but "fcitx support module for qt5", i.e., it's for being used with applications written in qt5. The fcitx-qt6 port provides the required module for being used by applications written in qt6. We should keep fcitx-qt5 when we still have ports using qt5.
(In reply to Li-Wen Hsu from comment #4) And as I wrote in my 1st comment (=description), cmake option for Qt6 doesn't build other than the actual immodule for Qt6, which could be also needed for Qt6. And sorry, I haven't test Qt6 immodule alone is installed, as I still have multiple Qt5 apps which need Japanese input via Qt5 version of immodule. And more, I've recently switched to fcitx5-mozc, which Ken Deguchi is porting [1]. The github repo is consisted as overlay that includes not only everything needed-to-be-added/updated for fcitx5-mozc / mozc-server but also many unrelated ports. Unfortunately, Ken's work for mozc-server / fcitx5-mozc are not ready to land, as the master port japanese/mozc-server needs to be updated to properly support textproc/fcitx5 and it requires newer devel/bazel in-tree or bazel7 (defaulted in Ken's overlay), causing downloads on build phase, which poudriere strongly dislikes (possible, though). And upstream is quickly updated. So anyone who want in-tree chinese/fcitx (aka fcitx4) for (maybe mostly) japanese/mozc-server and japanese/fcitx-mozc on Qt6 apps would still need my patch. [1] https://github.com/kdeguchi/ports