Contents of /xml/htdocs/proj/en/glep/glep-0044.txt

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1.8 - (show annotations) (download)
Sun Jan 11 19:40:56 2009 UTC (9 years, 11 months ago) by betelgeuse
Branch: MAIN
Changes since 1.7: +3 -3 lines
File MIME type: text/plain
Update Status from Accepted to Final

1 GLEP: 44
2 Title: Manifest2 format
3 Version: $Revision: 1.7 $
4 Last-Modified: $Date: 2006/10/14 02:55:39 $
5 Author: Marius Mauch <genone@gentoo.org>,
6 Status: Final
7 Type: Standards Track
8 Content-Type: text/x-rst
9 Created: 04-Dec-2005
10 Post-History: 06-Dec-2005, 23-Jan-2006, 3-Sep-2006
13 Abstract
14 ========
16 This GLEP proposes a new format for the Portage Manifest and digest file system
17 by unifying both filetypes into one to improve functional and non-functional
18 aspects of the Portage Tree.
21 Motivation
22 ==========
24 Please see [#reorg-thread]_ for a general overview.
25 The main long term goals of this proposal are to:
27 - Remove the tiny digest files from the tree. They are a major annoyance as on a
28 typical configuration they waste a lot of disk space and the simple transmission
29 of the names for all digest files during a ``emerge --sync`` needs a substantial
30 amount of bandwidth.
31 - Reduce redundancy when multiple hash functions are used
32 - Remove potential for checksum collisions if a file is recorded in more than one
33 digest file
34 - Difference between filetypes for a more flexible verification system
37 Specification
38 =============
40 The new Manifest format would change the existing format in the following ways:
42 - Addition of a filetype specifier, currently planned are
44 * ``AUX`` for files directly used by ebuilds (e.g. patches or initscripts),
45 located in the ``files/`` subdirectory
47 * ``EBUILD`` for all ebuilds
49 * ``MISC`` for files not directly used by ebuilds like ``ChangeLog`` or
50 ``metadata.xml`` files
52 * ``DIST`` for release tarballs recorded in the ``SRC_URI`` variable of an ebuild,
53 these were previously recorded in the digest files
55 Future portage improvements might extend this list (for example with types
56 relevant for eclasses or profiles)
58 - Only have one line per file listing all information instead of one line per
59 file and checksum type
61 - Remove the separated digest-* files in the ``files/`` subdirectory
63 Each line in the new format has the following format:
65 ::
67 <filetype> <filename> <filesize> <chksumtype1> <chksum1> ... <chksumtypen> <chksumn>
70 However theses entries will be stored in the existing Manifest files.
72 An `actual example`__ for a (pure) Manifest2 file..
74 .. __: glep-0044-extras/manifest2-example.txt
77 Compability Entries
78 -------------------
80 To maintain compability with existing portage versions a transition period after
81 is the introduction of the Manifest2 format is required during which portage
82 will not only have to be capable of using existing Manifest and digest files but
83 also generate them in addition to the new entries.
84 Fortunately this can be accomplished by simply mixing old and new style entries
85 in one file for the Manifest files, existing portage versions will simply ignore
86 the new style entries. For the digest files there are no new entries to care
87 about.
89 Scope
90 -----
92 It is important to note that this proposal only deals with a change of the
93 format of the digest and Manifest system.
95 It does not expand the scope of it to cover eclasses, profiles or anything
96 else not already covered by the Manifest system, it also doesn't affect
97 the Manifest signing efforts in any way (though the implementations of both
98 might be coupled).
100 Also while multiple hash functions will become standard with the proposed
101 implementation they are not a specific feature of this format [#multi-hash-thread]_.
103 Number of hashes
104 ----------------
106 While using multiple hashes for each file is a major feature of this proposal
107 we have to make sure that the number of hashes listed is limited to avoid
108 an explosion of the Manifest size that would revert the main benefit of this proposal
109 (reduzing tree size). Therefore the number of hashes that will be generated
110 will be limited to three different hash functions. For compability though we
111 have to rely on at least one hash function to always be present, this proposal
112 suggest to use SHA1 for this purpose (as it is supposed to be more secure than MD5
113 and currently only SHA1 and MD5 are directly available in python, also MD5 doesn't
114 have any benefit in terms of compability).
116 Rationale
117 =========
119 The main goals of the proposal have been listed in the `Motivation`_, here now
120 the explanation why they are improvements and how the proposed format will
121 accomplish them.
123 Removal of digest files
124 -----------------------
126 Normal users that don't use a "tuned" filesystem for the portage tree are
127 wasting several dozen to a few hundred megabytes of disk space with the current
128 system, largely caused by the digest files.
129 This is due to the filesystem overhead present in most filesystem that
130 have a standard blocksize of four kilobytes while most digest files are under
131 one kilobyte in size, so this results in approximately a waste of three kilobytes
132 per digest file (likely even more). At the time of this writing the tree contains
133 roughly 22.000 digest files, so the overall waste caused by digest files is
134 estimated at about 70-100 megabytes.
135 Furthermore it is assumed that this will also reduce the disk space wasted by
136 the Manifest files as they now contain more content, but this hasn't been
137 verified yet.
139 By unifying the digest files with the Manifest these tiny files are eliminated
140 (in the long run), reducing the apparent tree size by about 20%, benefitting
141 both users and the Gentoo infrastructure.
143 Reducing redundancy
144 -------------------
146 When multiple hashes are used with the current system
147 both the filename and filesize are repeated for every checksum type used as each
148 checksum is standalone. However this doesn't add any functionality and is
149 therefore useless, so the new format removes this redundancy.
150 This is a theoretical improvement at this moment as only one hash function is in
151 use, but expected to change soon (see [#multi-hash-thread]_).
153 Removal of checksum collisions
154 ------------------------------
156 The current system theoretically allows for a ``DIST`` type file to be recorded
157 in multiple digest files with different sizes and/or checksums. In such a case
158 one version of a package would report a checksum violation while another one
159 would not. This could create confusion and uncertainity among users.
160 So far this case hasn't been observed, but it can't be ruled out with the
161 existing system.
162 As the new format lists each file exactly once this would be no longer possible.
164 Flexible verification system
165 ----------------------------
167 Right now portage verifies the checksum of every file listed in the Manifest
168 before using any file of the package and all ``DIST`` files of an ebuild
169 before using that ebuild. This is unnecessary in many cases:
171 - During the "depend" phase (when the ebuild metadata is generated) only
172 files of type ``EBUILD`` are used, so verifying the other types isn't
173 necessary. Theoretically it is possible for an ebuild to include other
174 files like those of type ``AUX`` at this phase, but that would be a
175 major QA violation and should never occur, so it can be ignored here.
176 It is also not a security concern as the ebuild is verified before parsing
177 it, so each manipulation would show up.
179 - Generally files of type ``MISC`` don't need to be verified as they are
180 only used in very specific situations, aren't executed (just parsed at most)
181 and don't affect the package build process.
183 - Files of type ``DIST`` only need to be verified directly after fetching and
184 before unpacking them (which often will be one step), not every time their
185 associated ebuild is used.
188 Backwards Compatibility
189 =======================
191 Switching the Manifest system is a task that will need a long transition period
192 like most changes affecting both portage and the tree. In this case the
193 implementation will be rolled out in several phases:
195 1. Add support for verification of Manifest2 entries in portage
197 2. Enable generation of Manifest2 entries in addition to the current system
199 3. Ignore digests during ``emerge --sync`` to get the size-benefit clientside.
200 This step may be ommitted if the following steps are expected to follow soon.
202 4. Disable generation of entries for the current system
204 5. Remove all traces of the current system from the tree (serverside)
206 Each step has its own issues. While 1) and 2) can be implemented without any
207 compability problems all later steps have a major impact:
209 - Step 3) can only be implemented when the whole tree is Manifest2 ready
210 (ideally speaking, practically the requirement will be more like 95% coverage
211 with the expectation that for the remaining 5% either bugs will be filed after
212 step 3) is completed or they'll be updated at step 5).
214 - Steps 4) and 5) will render all portage versions without Manifest2 support
215 basically useless (users would have to regenerate the digest and Manifest
216 for each package before being able to merge it), so this requires a almost
217 100% coverage of the userbase with Manifest2 capabale portage versions
218 (with step 1) completely implemented).
220 Another problem is that some steps affect different targets:
222 - Steps 1) and 3) target portage versions used by users
224 - Steps 2) and 4) target portage versions used by devs
226 - Step 5) targets the portage tree on the cvs server
228 While it is relatively easy to get all devs to use a new portage version this is
229 practically impossible with users as some don't update their systems regulary.
230 While six months are probably sufficient to reach a 95% coverage one year is
231 estimated to reach an almost-complete coverage. All times are relative to the
232 stable-marking of a compatible portage version.
234 No timeframe for implementation is presented here as it is highly dependent
235 on the completion of each step.
237 In summary it can be said that while a full conversion will take over a year
238 to be completed due to compability issues mentioned above some benefits of the
239 system can selectively be used as soon as step 2) is completed.
242 Other problems
243 ==============
245 Impacts on infrastructure
246 -------------------------
248 While one long term goal of this proposal is to reduce the size of the tree
249 and therefore make life for the Gentoo Infrastructure easier this will only
250 take effect once the implementation is rolled out completely. In the meantime
251 however it will increase the tree size due to keeping checksums in both formats.
252 It's not possible to give a usable estimate on the degree of the increase as
253 it depends on many variables such as the exact implementation timeframe,
254 propagation of Manifest2 capable portage versions among devs or the update
255 rate of the tree. It has been suggested that Manifest files that are not gpg
256 signed could be mass converted in one step, this could certainly help but only
257 to some degree (according to a recent research [#gpg-numbers]_ about 40% of
258 all Manifests in the tree are signed, but this number hasn't been verified).
261 Reference Implementation
262 ========================
264 A patch for a prototype implementation of Manifest2 verification and partial
265 generation has been posted at [#manifest2-patch]_, it will be reworked before
266 being considered for inclusion in portage. However it shows that adding support
267 for verification is quite simple, but generation is a bit tricky and will
268 therefore be implemented later.
271 Options
272 =======
274 Some things have been considered for this GLEP but aren't part of the proposal
275 yet for various reasons:
277 - timestamp field: the author has considered adding a timestamp field for
278 each entry to list the time the entry was created. However so far no practical
279 use for such a feature has been found.
281 - convert size field into checksum: Another idea was to treat the size field
282 like any other checksum. But so far no real benefit (other than a slightly
283 more modular implementation) for this has been seen while it has several
284 drawbacks: For once, unlike checksums, the size field is definitely required
285 for all ``DIST`` files, also it would slightly increase the length of
286 each entry by adding a ``SIZE`` keyword.
288 - removal of the ``MISC`` type: It has been suggested to completely drop
289 entries of type ``MISC``. This would result in a minor space reduction
290 (its rather unlikely to free any blocks) but completely remove the ability
291 to check these files for integrity. While they don't influence portage
292 or packages directly they can contain viable information for users, so
293 the author has the opinion that at least the option for integrity checks
294 should be kept.
296 Credits
297 =======
299 Thanks to the following persons for their input on or related to this GLEP
300 (even though they might not have known it):
301 Ned Ludd (solar), Brian Harring (ferringb), Jason Stubbs (jstubbs),
302 Robin H. Johnson (robbat2), Aron Griffis (agriffis)
304 Also thanks to Nicholas Jones (carpaski) to make the current Manifest system
305 resistent enough to be able to handle this change without too many transition
306 problems.
308 References
309 ==========
311 .. [#reorg-thread] http://thread.gmane.org/gmane.linux.gentoo.devel/21920
313 .. [#multi-hash-thread] http://thread.gmane.org/gmane.linux.gentoo.devel/33434
315 .. [#gpg-numbers] gentoo-core mailing list, topic "Gentoo key signing practices
316 and official Gentoo keyring", Message-ID <20051117075838.GB15734@curie-int.vc.shawcable.net>
318 .. [#manifest2-patch] http://thread.gmane.org/gmane.linux.gentoo.portage.devel/1374
320 .. [#manifest2-example] http://www.gentoo.org/proj/en/glep/glep-0044-extras/manifest2-example
322 Copyright
323 =========
325 This document has been placed in the public domain.

  ViewVC Help
Powered by ViewVC 1.1.20