Bryan, sorry for the double email, as always I forgot to CC ports.
On Sun, Jan 19, 2020 at 10:48:18AM +0900, Bryan Linton wrote:
> On 2020-01-18 13:16:58, Omar Polo <[email protected]> wrote:
> > Hi,
> >
> > On Thu, Jan 09, 2020 at 11:34:34PM +0900, Bryan Linton wrote:
> > > Hello ports@
> > >
> > > I was attempting to make some changes to inputmethods/anthy for
> > > another purpose when I noticed it was woefully out of date.
> > >
> > > Version 9100h was released in 2009. Version 0.4 was released six
> > > months ago (in 2019).
> > >
> > > [snip]
> > >
> > > Concerns:
> > >
> > > * I have not tested the emacs module because I do not use emacs. I
> > > have however tested it in Firefox, leafpad, a Japanized xterm, and
> > > editors/nvi running in said Japanized xterm. More testing would
> > > be appreciated though.
> >
> > Thanks for working on this! It compiles here, but I cannot judge
> > the quality of the patch. Some comments after a bit of testing with
> > inputmethod/uim:
> >
>
> Hello, thank you for testing! Some comments follow inline.
>
> > - firefox (and possibly other programs) need some directories to
> > be unveiled [0].
> >
>
> Yes, I was the one who sent that email too :)
I've realized that, but I've left the comment primarly for other
interested in testing this.
> It's not specific to this update though. Even the old version of
> anthy would need those directories unveiled.
>
> I have another update to anthy that switches it to using a common
> directory in line with what Theo suggested later in that thread. But
> I wanted to hopefully get this update in-tree before going forward
> with that.
Oh, this is great! I'm looking forward to it.
> > - with gajim (gtk3) and emacs (gtk3) works as expected
> >
>
> Great, thanks! This was the one use-case I couldn't test myself
> since I don't use emacs.
>
> > - xterm and emacs (compiled with the lucid toolkit) sort of. While
> > typing, the characters are displayed as rectangles (similarly
> > to when a font is missing), but after pressing enter the whole
> > word is displayed properly. Also, the selection box does not
> > appear in these programs (but these probably are issues with uim
> > rather than anthy.)
> >
>
> Yes, this definitely sounds like a uim issue. To be clear, is
> this a regression? I.e. Did this work OK before, but broke with
> this update?
It's not a regression, it's a problem I have with uim plus any
programs that use the native X11 input method (and I don't have
knowledge on that matter.) I have XMODIFIERS set up as per uim
description.
Also, I'm sorry, but I've written a confusing comment. (at the
time of the last mail) I had tested only emacs with the standard x
input method (uim in that case), because I don't have much knowledge
of "native emacs input methods" and I confused myself with anthy.el
and uim.el...
However, I've now tested anthy.el properly and I can confirm that
it's *awesome*. The only drawbacks is that it needs LANG=ja_JP.UTF-8,
otherwise you get glibberish for some characters (i.e. "\343\200\202"
instead of "。")
Another thing that I've noticed, the description for anthy says
With its complement package anthy-emacs, [...]
but here the package is emacs-anthy:
$ pkg_info | grep emacs-anthy
emacs-anthy-0.4p4 emacs files for anthy
(patch below)
> Also as a follow-up, can you check whether you're using the
> "Anthy" or the "Anthy (UTF-8)" input method in UIM? If it's the
> former, does it fix this if you switch it to UTF-8?
>
> As I mentioned in the initial email, the internals of anthy have
> switched to be completely UTF-8. If you're using the old input
> method, does switching to "Anthy (UTF-8)" in uim fix this?
Ups, I didn't mentioned in the mail, but I have only used "Anthy
(UTF-8)", since I got strange (for the lack of a better term) input
with "Anthy" (non utf8) too. I have installed anthy to test your
patch (I was procastinating installing an input method for japanese.)
> Failing that, does running xterm with the script I've pasted in
> below fix this?
Unfortunately not. I've also installed the font you are using,
without success.
> snip...
So, to recap, anthy.el works, firefox and iridium too, and I have
only an issue with uim that's not a regression.
Cheers!
--
/Omar Polo
--- pkg/DESCR-main.orig Tue Nov 21 01:12:40 2006
+++ pkg/DESCR-main Mon Jan 20 14:38:51 2020
@@ -1,7 +1,7 @@
Anthy is a japanese input method library that can be used
from many setups.
-With its complement package anthy-emacs, it can be used with
+With its complement package emacs-anthy, it can be used with
emacs, using the simple anthy-agent wedge for communication.
It can also be accessed from uim.
