[bug#28421] [PATCH] gnu: Add uim.

[Arun Isaac]

>> * gnu/packages/xorg.scm (uim): New variable.
>
> [...]
>
>> +    (native-inputs
>> +     `(("anthy" ,anthy)
>> +       ("emacs" ,emacs-minimal)
>> +       ("gtk+" ,gtk+)
>> +       ("gtk+" ,gtk+-2)
>> +       ("intltool" ,intltool)
>> +       ("pkg-config" ,pkg-config)
>> +       ("qt" ,qt-4)))
>
> GTK+ and Qt are native inputs?

The package has gtk and qt programs to set uim preferences. At least,
those must need gtk and qt as inputs. So, you're right -- gtk and qt
should be inputs, not native inputs.

Also, I think anthy should be an input, not a native input.

> Also, is Qt 5 supported?  That would be best.

It looks there is no release yet that supports Qt 5, though there is
support in the master branch.

https://github.com/uim/uim/issues/61

> Besides, would it make sense to split the thing into several outputs to
> avoid the huge dependencies?

I don't know if it's a good idea to split the package into a gtk output
and a qt output. What if the user needs multilingual input in both gtk
and qt programs?

Also, this package can be built in many different ways -- with and
without the gtk and qt helper programs (for uim preferences), with and
without anthy support, with and without m17nlib support, etc. I don't
know where to draw the line, and choose only a few outputs. The
configure flags and dependencies, as they stand now, I copied from Arch
Linux.

https://git.archlinux.org/svntogit/packages.git/tree/trunk/PKGBUILD?h=packages/uim

> I’m not clear if it’s a later above Qt/GTK+ or something entirely
> separate?

I am not entirely sure either but I think at least the core library
(libuim) is entirely separate.

> Apart from that, LGTM with phases that return #t as you noted.

Will send an updated patch with all changes discussed.