[GSoC] About the idea: Unicode support in vi

[Matthias Andree]

Am 23.03.2011 10:13, schrieb Alexey Shuvaev:
> On Wed, Mar 23, 2011 at 12:39:44AM -0500, Zhihao Yuan wrote:
>> Hi,
>>
>> I'm a Computer Science student at Northern Illinois University, and I
>> used FreeBSD for a long time. I'm interested in the idea that to
>> improve the nvi in the base system. My proposal is slightly different:
>> I want to fork nvi and make it iconv-awared (or mbyte-mode tunable,
>> like tcsh), so that it can deal with more encodings. Can that be a
>> GSoC project proposal?
>>
> +1 here!
> 
> ports/editors/nvi-devel is another starting point here. As far as I understand
> it is a further development of nvi which is in base. What I don't like
> about it is a dependency on databases/db3 and changed (worse, in my opinion)
> handling of keystrokes in 'insert' mode. But it is iconv-aware implementation
> already.

nvi-devel is bit-rotten. Most releases date from 2004, and there was a
patchlevel-release in 2007 apparently, since then it's been left to bit rot.

I'm thinking about just killing databases/db3 and see what happens with
nvi-devel.  I tried convincing it to work with db41, and while it
compiles, it somehow abuses Berkeley DB in a way I don't see during
debugging and barfs with "Invalid argument" on a DB->open call on a
recovery file.

Also, the documentation says it depends on 3.1, but then we've been
using 3.3 for ages, but even the first release of nvi-devel to use
Berkeley DB was released when 4.2 was already out.  There seems to be
some code to make it work (which in itself is buggy it uses broken
comparisons for its version checks), but it doesn't work for reasons I
don't see with gdb.  Berkeley DB doesn't like the way it's being used
and errors out with EINVAL.  However, I don't care enough to build a
debug-enabled version of Berkeley DB to see where abandoned nvi-devel
might abuse bdb.

vim works for me, supports Unicode, and for "fewer dependencies", we
have vim-lite.