Package Details: emacs-mozc-ut-full 2.31.5712.102.20250110-1

Git Clone URL: https://aur.archlinux.org/mozc-ut-full.git (read-only, click to copy)
Package Base: mozc-ut-full
Description: Emacs engine module for Mozc with UT dictionary
Upstream URL: https://github.com/fcitx/mozc
Licenses: custom
Conflicts: emacs-mozc
Replaces: emacs-mozc
Submitter: brli
Maintainer: brli
Last Packager: brli
Votes: 6
Popularity: 0.004558
First Submitted: 2022-07-04 17:27 (UTC)
Last Updated: 2025-01-19 07:19 (UTC)

Required by (0)

Sources (21)

Latest Comments

« First ‹ Previous 1 2 3 4 Next › Last »

brli commented on 2023-08-31 19:09 (UTC)

@Score_Under: please rephrase the term: "Every time I get to it", because I can't understand that.

Let me explain a bit:

  1. The main repo, aka. fcitx's fork of mozc, is pinned to a certain commit, in the line _mozc_commit=e4617bf, which should be unique-hashed enough.
  2. The other libs are defined by the git submodule document in THAT mozc commit.
  3. The utuhiro78 additional dictionary helper is maintained by the author, and is fine to always follow up to main branch IMHO, as it won't pollute final binary.
  4. I admit that it isn't the best solution to use git repo as source. Also, Noted that I'm not always following the latest commit.
  5. The dictionaries are prefixed with the date-of-latest-push (well, not that accurate anyway)

Score_Under commented on 2023-08-31 10:07 (UTC)

Would it be possible to pin the resources in this pkgbuild to a specific git hash permalink instead of the main branch? Every time I get to it, the hashes are incorrect already.

Bitals commented on 2023-08-08 11:33 (UTC)

  1. It just runs makepkg in a container and puts the result into the repo, it's not black magic. It does what PKGBUILD tells it to do.
  2. I only tried to execute it standalone to get an error message, because it didn't run at all via ibus on Gnome as I actually use it. The Properties dialog is actually fine to run via cmd. That problem was fixed by this commit: https://aur.archlinux.org/cgit/aur.git/commit/?h=mozc-ut-full&id=c8f75cae67a60f099fb2e3220b6ab6d59246ffb4 that introduced qt5-base as makedep. Because before it had no way to know that this needs qt5-base to actually build properly.

Now my problem is that autocompletion popup only stays for half a second before completing with the first suggestion and closing itself. Not sure how to debug this, but I think it's a mozc' issue, not PKGBUILD'.

brli commented on 2023-08-06 15:25 (UTC)

  1. well...I'm not pretty sure but, the Arch's official devtools returns that they're using systemd-nspawn as official container solution. So, maybe no much help if you're using docker...(still, make sure your docker steps are cohere to upstream's steps might help)

  2. mozc isn't meant for execute and use as standalone app (can tell by the pkgnames in PKGBUILD), if you're telling me that, it doesn't work on XX input framework, then there probably be some workaround or hint of breakage. You can't tell me by "it can't be executed alone, thus I can't input a thing through it." FYI, I mainly use fcitx5-mozc, and when it fails me, there comes a recompile due to library mismatch.

Bitals commented on 2023-07-15 12:12 (UTC)

@BrLi It got installed once it was in makedeps. Docker image is rebuilt weekly and immediately on large updates when they happen. Only problem now is the completion prompt being displayed for just a couple of frames, making it impossible to use.

brli commented on 2023-07-14 14:00 (UTC)

@Bitals

The qt5-base is in makedepends, thus should be installed via makepkg.

I can't help much if you're not using executable wrapped around makepkg when compiling.

Also, it is highly possible that your package set in the container isn't up-to-date.

Plz make sure to sync repos and do a full system upgrade before doing any compilation.

Bitals commented on 2023-07-14 06:52 (UTC)

@BrLi I have qt5-base installed, but it was missing from the build environment (docker container on a different host), so it was built fine, but seems like that corrupted the final binaries. Yesterdays commit finally runs fine, t9 cancels immidiately, but I think that's not related, will try to solve it.

brli commented on 2023-07-13 12:14 (UTC)

@Bitals thanks

  1. have you installed qt5-base? as it seems that mozc_tool requires that it to display GUI (check with ldd /usr/lib/mozc/mozc_tool), I'll add that as optdepends

  2. the mozc_server isn't meant to be executed alone.

  3. please check that you're packaging under clean chroot.

  4. I run the packages on my machine before upload, so, they shall be compiled and run fine IMHO.

@eponymous-patch and else:

Noted that the source tarballs remains the same name, thus you should have clean up the cached source files before build.

I might make mistakes also that I didn't check out the cache first sometime.

Bitals commented on 2023-07-13 11:11 (UTC)

@BrLi https://share.bitals.xyz/s/nrrPbw4BX884XJ4 Password: OccasionalReturn4b13L@mebr@!n

brli commented on 2023-07-13 10:41 (UTC)

since osdn is down

can someone please provide cached x-ken-all-202110.zip and jigyosyo-202110.zip?

« First ‹ Previous 1 2 3 4 Next › Last »