Bug 555428

Summary:	Proposal: generate /usr/share/dict/words from installed dictionaries like aspell/hunspell instead of relying on sys-apps/miscfiles
Product:	Gentoo Linux	Reporter:	Coacher <itumaykin+gentoo>
Component:	[OLD] Unspecified	Assignee:	Gentoo's Team for Core System packages <base-system>
Status:	UNCONFIRMED ---
Severity:	enhancement
Priority:	Lowest
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---

Description Coacher 2015-07-20 08:41:19 UTC

Hello.

Currently /usr/share/dict/words files is available from sys-apps/miscfiles package. Unfortunately miscfiles repository looks dead. On the other hand many users who care about spell checking already have aspell/hunspell installed.

Some modern programs rely on /usr/share/dict/words file for spell checking (for example https://github.com/git-cola/git-cola). It would be nice to install a more modern dictionary for such programs and to avoid pulling a collection of antique BSD files. We can generate this file from installed aspell/hunspell dictionaries.

ArchLinux distro uses aspell dictionaries to provide such file (see https://www.archlinux.org/packages/community/any/words/). A similar approach should work for hunspell (there exist a bugreport regarding this on hunspell sourceforge page, but the site is down atm).

We can refine ArchLinux approach: generate words file for each lang in LINGUAS, then put a symlink to the first one, which is considered default (ArchLinux generates words files for a limited numbers of languages only and determines the default one based on the LANG value).

I can work on the needed additions for aspell and hunspell packages if someone is interested.

But I don't see how to resolve a conflict between aspell and hunspell if we choose to generate words files for both of them.

Reproducible: Always

Comment 1 Coacher 2015-07-20 08:47:39 UTC

Also http://wordlist.sourceforge.net/ project exist, which provides a limited number of words files and is used in Debian.