GLEP 31: Character Sets for Portage Tree Items

Author Ciaran McCreesh <ciaranm@gentoo.org>
Type Standards Track
Status Final
Version 1
Created 2004-10-27
Last modified 2014-01-17
Posting history 2004-10-28, 2004-11-01, 2004-11-11
GLEP source glep-0031.rst

Abstract

A set of guidelines regarding what characters are permissible in the portage tree and how they should be encoded is required.

Status

Approved on 2004-11-08 assuming that implementation will include documentation for correctly encoding files within nano.

Motivation

At present we have several developers and many more users whose names require characters (for example, accents) which are not part of the standard 'safe' 0..127 ASCII range. There is no current standard on how these should be represented, leading to inconsistency across the tree.

Although the issues involved have been discussed informally many times, no official decision has been made.

Specification

ChangeLog and Metadata Character Sets

It is proposed that UTF-8 ([1]) is used for encoding ChangeLog and metadata.xml files inside the portage tree.

UTF-8 allows the full range of Unicode ([2]) characters to be expressed, which is necessary given the diversity of the Gentoo developer- and user-base. It is character-compatible with ASCII for the 0..127 characters and does not significantly increase the storage requirements for files which consist mainly of American English characters. It is widely supported, widely used and an official standard.

The ISO-8859-* character sets ([3]) would not be appropriate since they cannot express the full range of required characters.

Ebuild and Eclass Character Sets

For the same reasons as previously, it is proposed that UTF-8 is used as the official encoding for ebuild and eclass files.

However, developers should be warned that any code which is parsed by bash (in other words, non-comments), and any output which is echoed to the screen (for example, einfo messages) or given to portage (for example any of the standard global variables) must not use anything outside the regular ASCII 0..127 range for compatibility purposes.

files/ Entries Character Sets

Patches must clearly be in the same character set as the file they are patching. For other files/ entries (for example, GNOME desktop files), consistency with the upstream-recommended character set is most sensible.

Suitable Characters for File and Directory Names

Characters outside the ASCII 0..127 range cannot safely be used for file or directory names. (Of course, not all characters inside the ASCII 0..127 range can be used safely either.)

Backwards Compatibility

The existing tree uses a mixture of encodings. It would be straightforward to fix existing ChangeLogs and metadata files to use UTF-8.

The echangelog tool is character-set agnostic. In order to properly enter UTF-8, developers would have to switch to a UTF-8 shell session. This only applies if the developer is entering new text which uses 'fancy' characters -- existing characters are not mangled.

Certain text editors are incapable of handling UTF-8 cleanly. However, since the echangelog tool is generally the correct way to generate ChangeLog entries, this should not be a major problem. Generating metadata.xml files correctly in these editors could become problematic. The vim and emacs editors, which appear to be most widely used, are both capable of handling UTF-8 cleanly -- for vim, this could be configured automatically via the gentoo-syntax ([4]) package.

References

[1]RFC 3629: UTF-8, a transformation format of ISO 10646 http://www.ietf.org/rfc/rfc3629.txt
[2]ISO/IEC 10646 (Universal Multiple-Octet Coded Character Set)
[3]ISO/IEC 8859 (8-bit single-byte coded graphic character sets)
[4]The app-vim/gentoo-syntax package, https://developer.berlios.de/projects/gentoo-syntax/