/[gentoo]/xml/htdocs/proj/en/glep/glep-0022.txt
Gentoo

Contents of /xml/htdocs/proj/en/glep/glep-0022.txt

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.1 - (hide annotations) (download)
Sun Mar 7 02:20:32 2004 UTC (10 years, 1 month ago) by g2boojum
Branch: MAIN
File MIME type: text/plain
new glep

1 g2boojum 1.1 GLEP: 22
2     Title: New "keyword" system to incorporate various userlands/kernels/archs
3     Version: $Revision: 1.1 $
4     Last-Modified: $Date: 2004/02/22 02:09:07 $
5     Author: Grant Goodyear <g2boojum@gentoo.org>
6     Status: Draft
7     Type: Standards Track
8     Content-Type: text/x-rst
9     Created: 6-Mar-2004
10     Post-History: 6-Mar-2004
11    
12     Credits
13     =======
14    
15     This GLEP originated from the concerns that Daniel Robbins had with
16     the *x86obsd* keyword, and his desire to make the KEYWORDS variable more
17     "feature-rich". Drobbins' original idea was that we should
18     allow compound
19     keywords such as gnu/x86, gnu/ppc, and macos/ppc (which would be explicit
20     versions of the more familiar x86, ppc, and macos keywords). Method noted
21     that userland/arch failed to capture the full range of possibilities (what
22     about a GNU userland on a BSD kernel+libc?), and
23     the issue has languished due to a lack of reasonable solutions.
24    
25     Abstract
26     ========
27    
28     As Gentoo branches out to support non-Linux and non-GNU systems (such
29     as Hurd or the \*BSDs), the potential for an "explosion" of possible
30     keywords becomes rather large, since each
31     new userland/kernel/arch/whatever
32     combination would require a new keyword.
33     This GLEP proposes replacing the current
34     KEYWORDS variable with four variables, ARCH, USERNAME, KERNEL, and LIBC,
35     along with sensible defaults to keep the new system manageable.
36    
37     Motivation
38     ==========
39    
40     Since the beginning, Gentoo Linux has been conceived as a "metadistribution"
41     that combines remarkable flexibility with sensible defaults and exceptional
42     maintainablilty. The goal of the Gentoo-Alt_ project has been to extend that
43     flexibility to include systems other than GNU/Linux. For example, the author
44     of this GLEP has been working to create a version_ of Gentoo that uses
45     OpenBSD_ as the underlying kernel, userland, and libc. OpenBSD_ supports
46     a variety of different architectures, so, in principle, we would need a new
47     *openbsd-arch* keyword for each supported architecture. In fact, the situation
48     is even more complicated, because the Gentoo-Alt_ project would eventually
49     like
50     to support the option of "mixing-and-matching" GNU/\*BSD/whatever userlands
51     and libcs irrespective of the underlying kernel. (Debian_, for example
52     has a similar BSD project_, except that they have replaced the
53     BSD userland with a GNU userland.) The net result is that we would need
54     keywords that specified all possible permutations of arch, userland, kernel
55     and libc. Not fun.
56    
57     .. _Gentoo-Alt: http://www.gentoo.org/proj/en/gentoo-alt/index.xml
58     .. _OpenBSD: http://www.openbsd.com
59     .. _version: http://www.gentoo.org/proj/en/gentoo-alt/bsd/index.xml
60     .. _Debian: http://www.debian.org
61     .. _project: http://www.debian.org/ports/netbsd/
62    
63     Specification
64     =============
65    
66     New Variables
67     -------------
68    
69     I suggest that we replace the single KEYWORDS variable in ebuilds
70     with four separate variables: ARCH, USERLAND, LIBC, and KERNEL.
71    
72     ARCH:
73     x86, amd64, cobalt, mips64, arm, hppa, ia64, ppc64, sparc
74     USERLAND:
75     gnu, bsd
76     LIBC:
77     glibc, openbsd, freebsd, netbsd, macosx
78     KERNEL:
79     linux, selinux, openbsd, freebsd, netbsd, macosx
80    
81     (The above examples are not meant to be complete. Hurd, for example
82     is not included because I know very little about Hurd.)
83     For each variable the standard "-,-\*,~" prefixes would be allowed.
84     Similarly, `/etc/make.conf` would have ACCEPT_ARCH, ACCEPT_USERLAND,
85     ACCEPT_LIBC, and ACCEPT_KERNEL variables.
86    
87     Reasonable Defaults
88     -------------------
89    
90     To keep this system manageable, we need sensible defaults. An ebuild
91     that has missing USERLAND, KERNEL, or LIBC variables is provided
92     with implicit USERLAND="gnu", KERNEL="linux", and/or LIBC="glibc"
93     variables. However, once a variable is explicitly added (such as
94     KERNEL="openbsd"), the default is no longer assumed. That is,
95     one would need KERNEL="openbsd linux" if the ebuild is stable on
96     both openbsd and linux kernels.
97    
98     The ARCH variable, on the other hand, does *not* have a default, per se.
99     Instead, if no ARCH variable exists then portage would automatically
100     add the ebuild's KEYWORD entries to ARCH. Thus, all current ebuilds
101     would still work without changes, allowing for a gradual transition
102     to the new system as the new variables are needed.
103    
104     Profiles
105     --------
106    
107     Along with an explosion of keywords comes a concomitant explosion
108     of potential profiles. The good news is that profiles show up only
109     in a single directory, so an explosion there is easier to contain.
110     I suggest an arch-kernel-userland-libc-version naming scheme, with
111     the kernel-userland-libc terms defaulting to linux-gnu-glibc if
112     absent. (Yes, Chemists do tend to be fond of systematic naming
113     systems.)
114    
115     One drawback to having a large number of profiles is that maintainance
116     becomes a significant problem. In fact, one could reasonably argue
117     that the current number of profiles is already too many to be
118     easily maintained. One proposal that has been raised to simplify
119     matters is the idea of stackable, or cascading, profiles, so that
120     only differences between profiles would have to be maintained.
121    
122     Rationale
123     =========
124    
125     The proposed new "keywording" system is far from elegant, which is
126     a substantial drawback. On the other hand, it is simple, it requires
127     relatively minor changes (albeit ones that eventually would impact
128     every ebuild in the portage tree), and the changes can be implemented
129     gradually over time.
130    
131     Implementation
132     ==============
133    
134     Implementation of this GLEP would divide into adding
135     Portage functionality to support the new system and
136     modifying ebuilds to
137     comply with the new system.
138     The Portage support involves hacking Portage
139     to assemble and check a four-state
140     arch-userland-kernel-libc variable instead of the simpler
141     KEYWORD variable. One might quibble over algorithmic issues, but
142     the actual concept is pretty straightforward. Rewriting ebuilds,
143     on the other hand, is a massive undertaking. Fortunately, it is
144     also a process that can be done over whatever length of time is
145     required, since "legacy" ebuilds should work with no changes.
146    
147    
148     Backwards Compatibility
149     =======================
150    
151     Backwards compatibility has already been addressed in some detail,
152     with the stated goal being a system that would leave all current
153     ebuilds in a still-functioning state after the portage modifications
154     have been made. However, we are already using an ARCH variable for
155     some arcane purpose in Portage, and that issue would still need to
156     be resolved.
157    
158    
159     Copyright
160     =========
161    
162     This document is licensed under the Creative Commons - Attribution / Share
163     Alike license. (http://creativecommons.org/licenses/by-sa/1.0)

  ViewVC Help
Powered by ViewVC 1.1.20