Contents of /xml/htdocs/proj/en/glep/glep-0025.txt

Parent Directory Parent Directory | Revision Log Revision Log

Revision 1.1 - (show annotations) (download)
Sun Apr 4 22:56:06 2004 UTC (14 years, 2 months ago) by g2boojum
Branch: MAIN
File MIME type: text/plain
GLEP update + addition

1 GLEP: 25
2 Title: Distfile Patching Support
3 Version: $Revision: 1.0 $
4 Last-Modified: $Date: 2004/04/01 01:07:40 $
5 Author: Brian Harring <ferringb@gentoo.org>
6 Status: Draft
7 Type: Standards Track
8 Content-Type: text/x-rst
9 Created: 6-Mar-2004
10 Post-History: 4-Apr-2004
12 Abstract
13 ========
15 The intention of this GLEP is to propose the creation of patching support for
16 portage, and iron out the implementation details.
18 Motivation
19 ==========
21 Reduce the bandwidth load placed on our mirrors by decreasing the amount of
22 bytes transferred when upgrading between versions. Side benefit of this is to
23 significantly decrease the download requirements for users lacking broadband.
25 Binary patches vs GNUDiff patches
26 =================================
28 Most people are familiar with diff patches (unified diff for example)- this
29 glep is specifically proposing the use of an actual binary differencer. The
30 reason for this is that diff patches are line based- you change a single
31 character in a line, and the whole line must be included in the patch. Binary
32 differencers work at the byte level- it encodes just that byte. In that
33 respect binary patches are often much more efficient then diff patches.
35 Further, the ability to reverse a unified patch is due to the fact the diff
36 includes **both** the original line, and the modified line. The author isn't
37 aware of any binary differencer that is able to create patches the can be
38 reversed- basically they're unidirectional, the patch that is generated can
39 only be used to upgrade or downgrade the version, not both. The plus side of
40 this limitation is a significantly decreased patch size.
42 The choice of binary patches over diff patches pretty much comes down to the
43 fact they're smaller- example being a kdelibs binary patch for 3.1.4->3.1.5 is
44 75kb, the equivalent diff patch is 123kb, and is unable to result in a correct
45 md5 [1]_.
47 Currently, this glep is proposing only the usage of binary patches- that's not
48 to say (with a fair amount of work) it couldn't be extended to support
49 standard diffs.
51 Rationale
52 =========
54 The difference between source releases typically isn't very large, especially
55 for minor releases. As an example, kdelibs-3.1.4.tar.bz2 is 10.53 MB, and
56 kdelibs-3.1.5.tar.bz2 is 10.54 MB. A bzip2'ed patch between those versions is
57 75.6 kb [2]_, less then 1% the size of 3.1.5's tbz2.
59 Specification
60 =============
62 Quite a few sections of gentoo are affected- mirroring, the portage tree, and
63 portage itself.
65 Additions to the tree
66 ---------------------
68 For adding patch info into the tree, this glep proposes a global patch list
69 (stored in profiles as patches.global), and individual patch lists stored in
70 relevant package directories (named patches). Using the kernel packages as an
71 example, a global list of patches enables us to create a patch once, add an
72 entry, and have all kernel packages benefit from that single entry. Both
73 patches.global, and individual package patch files share the same format:
75 ::
77 MD5 md5-value patch-url size MD5 md5-value ref-file size UMD5 md5-value new-file size
79 For those familiar with digest file layout, this should look familiar.
80 Essentially, chksum type, value, filename, size. The UMD5 chksum type is just
81 the uncompressed md5/size of the file- so if the UMD5 were for a bzip2
82 compressed file, it would be the md5 value/size of the uncompressed file.
83 And an example:
85 ::
87 MD5 ccd5411b3558326cbce0306fcae32e26 http://dev.gentoo.org/~ferringb/patches/kdelibs-3.1.4-3.1.5.patch.bz2 75687 MD5 82c265de78d53c7060a09c5cb1a78942 kdelibs-3.1.4.tar.bz2 10537433 UMD5 0b1908a51e739c07ff5a88e189d2f7a9 kdelibs-3.1.5.tar.bz2 48056320
89 In the above example, the md5sum of
90 http://dev.gentoo.org/~ferringb/patches/kdelibs-3.1.4-3.1.5.patch.bz2 is
91 calculated, compared to the stored value, and then the file size is checked.
92 The one difference is the UMD5 checksum type- the md5 value and the size are
93 specific to the *uncompressed* file. Continuing, for cases where the patch
94 will reside on one of our mirrors, the patch filename would be sufficient.
96 Finally, note that this is a unidirectional patch- using the above example,
97 kdelibs-3.1.4-3.1.5 can **only** be used to upgrade from 3.1.4 to 3.1.5, not
98 in reverse (originally explained in `Binary patches vs GNUDiff patches`_).
100 Portage Implementation
101 ----------------------
103 This glep proposes the patching support should be (at this stage) optional-
104 specifically, enabled via FEATURES="patching".
106 Fetching
107 ''''''''
109 When patching is enabled, the global patch list is read, and the packages
110 patch list is read. From there, portage determines what files could be used
111 as a base for patching to the desired file- further, determining if it's
112 actually worth patching (case where it wouldn't be is when the target file is
113 less then the sum of the patches needed). Any patches to be used are fetched,
114 and md5 verified.
116 Reconstruction
117 ''''''''''''''
119 Upon fetching and md5 verification of patch(es), the desired file is
120 reconstructed. Assuming reconstruction didn't return any errors, the target
121 file has its uncompressed md5sum calculated and verified, then is recompressed
122 and the compressed md5sum calculated. At this point, if the compressed md5
123 matches the md5 stored in the tree, then portage transfers the file into
124 distfiles, and continues on it's merry way.
126 If the compressed md5 is different from the tree's value, then the (proposed)
127 md5 database is updated with new compressed md5. Details of this database
128 (and the issue it addresses) follow.
130 Compressed MD5sums:
131 '''''''''''''''''''
133 There will be instances where a file is reconstructed perfectly, recompressed,
134 and the recompressed md5sum differs from what is stored in the tree- the
135 problem is that the md5sum of a compressed file is inherently tied to the
136 compressor version/options used to compress the original source.
138 =====================
139 The Problem in Detail
140 =====================
142 A good example of this problem is related to bzip2 versions used for
143 compression. Between bzip2 0.9x and bzip2 1.x, there was a subtle change in
144 the compressor resulting in a slightly better compression result- end result
145 being a different file, eg a different md5sum. Assuming compressor versions
146 are the same, there also is the issue of what compression level the target
147 source was originally compressed at- was it compressed with -9, -8 or -7?
148 That's just a sampling of the various original settings that must be accounted
149 for, and that's limited to gzip/bzip2; other compressors will add to the
150 number of variables to be accounted for to produce an exact recreation of the
151 compressed md5sum.
153 Tracking the compressor version and options originally used isn't really a
154 valid option- assuming all options were accounted for, clients would still be
155 required to have multiple versions of the same compressor installed just for
156 the sake of recreating a compressed md5sum *even though* the uncompressed
157 source's md5 has already been verified.
159 =====================
160 The Proposed Solution
161 =====================
163 The creation of a clientside flatfile/db of valid alternate md5/size pairs
164 would enable portage to handle perfectly reconstructed files, that have a
165 different md5sum due to compression differences. The proposed format is thus:
167 ::
169 MD5 md5sum orig-file size MD5 md5sum [ optional new-name ] size
171 Example:
173 ::
175 MD5 984146931906a7d53300b29f58f6a899 OOo_1.0.3_source.tar.bz2 165475319 MD5 0733dd85ed44d88d1eabed704d579721 165444187
177 An alternate md5/size pair for a file would be added **only** when the
178 uncompressed source's md5/size has been verified, yet upon recompression the
179 md5 differs. For cleansing of older md5/size pairs from this db, a utility
180 would be required- the author suggests the addition of a distfiles-cleaning
181 utility to portage, with the ability to also cleanse old md5/size pairs when
182 the file the pair was created for no longer exists in distfiles.
184 Where to store the database is debatable- /etc/portage or /var/cache/edb are
185 definite options.
187 The reasoning for allowing for an optional new-name is that it provides needed
188 functionality should anyone attempt to extend portage to allow for clients to
189 change the compression used for a source (eg, recompress all gzip files as
190 bzip2). Granted, no such code or attempt has been made, but nothing is lost
191 by leaving the option open should the request/attempt be made.
193 A potential gotcha of adding this support is that in environments where the
194 distfiles directory is shared out to multiple systems, this db must be shared
195 also.
199 Distfile Mirror Additions
200 -------------------------
202 One issue of contention is where these files will actually be stored. As of
203 the writing of this glep, a full distfiles mirror is roughly around 40 gb- a
204 rough estimate by the author places the space requirements for patches for
205 each version at a total of around 4gb. Note this isn't even remotely a hard
206 figure yet, and a better figure is being checked into currently.
208 Regardless of the exact space figure, finding a place to store the patches
209 will be problematic. Expansion of the required mirror space (essentially just
210 swallowing the patches storage requirement) is unlikely, since it was one of
211 the main arguements against the now defunct glep9 attempt [2]_. A couple of
212 ideas that have been put forth to handle the additional space requirements are
213 as follows-
215 1) Identification of mirrors willing to handle the extra space requirements-
216 essentially create an additional patch mirror tier.
218 2) Mirroring only a patch for certain package versions, rather then full
219 source. Using kdelibs-3.1.5 as an example, only the patch would be mirrored
220 (rather then the full 10.53 MB source). Downside to this approach is that a
221 user who is downloading kdelibs for the first time would either need to pull
222 it from the original SRC_URI (placing the burden onto the upstream mirror), or
223 pull the 3.1.4 version, and the patch- pulling 63k more then if they had just
224 pulled the full version. The kdelibs 3.1.4/3.1.5 example is something of an
225 optimal case- not all versions will have such miniscule patches.
227 3) A variation on the idea above, essentially mirroring only the patch for
228 the oldest version(s) of a package; eg, kdelibs currently has version 3.05,
229 3.1.5, 3.2.0, and 3.2.1- the mirrors would only carry a patch for 3.05, not
230 full source (think RESTRICT="fetch"). One plus to this is that patches to
231 downgrade in version are smaller then the patches to upgrade in version- there
232 are exceptions to this, but they're hard to find. A major downside to this
233 approach is A) a user would have to sync up to get the patchlists for that
234 version, B) creation of a set of patches to go backwards in version (see
235 `Binary patches vs GNUDiff patches`_)..
237 Of the options listed above, the first is the easiest, although the second
238 could be made to work. Feedback and any possible alternatives would be
239 greatly appreciated.
241 Patch Creation
242 --------------
244 Maintenance of patch lists, and the actual patch creation ought to be managed
245 by a high level script- essentally a dev says "I want a patch between this
246 version, and that version: make it so", the script churns away
247 creating/updating the patch list, and generating the patch locally. The
248 utility next uploads the new patch to /space/distfiles-local on dev.gentoo.org
249 (exempting if it's not a locally generated patch), and repoman is used to
250 commit the updated patch list.
252 What would be preferable (although possibly wishful thinking), is if hardware
253 could be co-opted for automatic patch generation, rather then forcing it upon
254 the devs- something akin to how files are pulled onto the mirror automatically
255 for new ebuilds.
257 The initial bulk of patches to get will be generated by the author, to ease
258 the transition and offer patches for people to test out.
260 Backwards Compatability
261 =======================
263 As noted in `The Proposed Solution`_, a system using patching and sharing out
264 it's distfiles must share out it's alternate md5 db. Any system that uses the
265 distfiles share must support the alternate md5 db also. If this is considered
266 enough of an issue, it is conceivable to place reconstructed sources with an
267 alternate md5 into a subdirectory of distdir- portage only looks within
268 distdir, unwilling to descend into subdirectories.
270 Also note that `Distfile Mirror Additions`_ may add additional backwards
271 compatability issues, depending on what solution is accepted.
273 Reference Implementation
274 ========================
276 TODO
278 References
279 ==========
280 .. [1] http://dev.gentoo.org/~ferringb/patches/kdelibs-3.1.4-3.1.5.{patch,diff}.bz2.
281 .. [2] kdelibs-3.1.4-3.1.5.patch.bz2, switching format patch, created via diffball-0.4_pre4 (diffball is available at http://sourceforge.net/projects/diffball)
282 Bzip2 -9 compressed, the patch is 75,687 bytes, uncompressed it is 337,649 bytes. The patch is available at http://dev.gentoo.org/~ferringb/kdelibs-3.1.4-3.1.5.patch.bz2 for those curious.
283 .. [3] Glep9, 'Gentoo Package Update System'
284 (http://glep.gentoo.org/glep-0009.html)
286 Copyright
287 =========
289 This document has been placed in the public domain.

  ViewVC Help
Powered by ViewVC 1.1.20