Mercurial > hg > mercurial-crew-with-dirclash
comparison doc/FAQ.txt @ 455:8d43dfdfb514
More FAQ updates
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
More FAQ updates
manifest hash: 98447c3da5aefcc6c4071d03d8014944cf4cbb79
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQFCu3TCywK+sNU5EO8RArRjAJ0ZtMHztUL1cQw7FC0C3uJ0YIfKjwCfWfSe
JndrQxPs1QeCPK/RbfYiKjE=
=aMHP
-----END PGP SIGNATURE-----
author | mpm@selenic.com |
---|---|
date | Thu, 23 Jun 2005 18:49:38 -0800 |
parents | df83b2c306ac |
children | d6ac88a738c4 |
comparison
equal
deleted
inserted
replaced
454:58d57594b802 | 455:8d43dfdfb514 |
---|---|
120 effectively the sum of all the contents of the repository for that | 120 effectively the sum of all the contents of the repository for that |
121 change, it is impossible in Mercurial to simultaneously commit and add | 121 change, it is impossible in Mercurial to simultaneously commit and add |
122 a tag. Thus tagging a revision must be done as a second step. | 122 a tag. Thus tagging a revision must be done as a second step. |
123 | 123 |
124 | 124 |
125 .Q. What if I want to just keep local tags? | |
126 | |
127 You can add a section called "[tags]" to your .hg/hgrc which contains | |
128 a list of tag = changeset ID pairs. Unlike traditional tags, these are | |
129 only visible in the local repository, but otherwise act just like | |
130 normal tags. | |
131 | |
132 | |
125 .Q. How do tags work with multiple heads? | 133 .Q. How do tags work with multiple heads? |
126 | 134 |
127 The tags that are in effect at any given time are the tags specified | 135 The tags that are in effect at any given time are the tags specified |
128 in each head, with heads closer to the tip taking precedence. | 136 in each head, with heads closer to the tip taking precedence. Local |
137 tags override all other tags. | |
129 | 138 |
130 | 139 |
131 .Q. What are some best practices for distributed development with Mercurial? | 140 .Q. What are some best practices for distributed development with Mercurial? |
132 | 141 |
133 First, merge often! This makes merging easier for everyone and you | 142 First, merge often! This makes merging easier for everyone and you |
185 | 194 |
186 Mercurial is primarily developed for UNIX systems, so some UNIXisms | 195 Mercurial is primarily developed for UNIX systems, so some UNIXisms |
187 may be present in ports. | 196 may be present in ports. |
188 | 197 |
189 | 198 |
190 .Q. How does signing work? | 199 .Q. How does Mercurial store its data? |
191 | 200 |
192 Take a look at the hgeditor script for an example. The basic idea | 201 The fundamental storage type in Mercurial is a "revlog". A revlog is |
193 is to sign the manifest ID inside that changelog entry. The manifest | 202 the set of all revisions of a named object. Each revision is either |
194 ID is a recursive hash of all of the files in the system and their | 203 stored compressed in its entirety or as a compressed binary delta |
195 complete history, and thus signing the manifest hash signs the entire | 204 against the previous version. The decision of when to store a full |
196 project to that point. | 205 version is made based on how much data would be needed to reconstruct |
197 | 206 the file. This lets us ensure that we never need to read huge amounts |
198 More precisely: each file hash is an SHA1 hash of the contents of that | 207 of data to reconstruct a object, regardless of how many revisions of it |
199 file and the hashes of its parent revisions. The manifest contains a | 208 we store. |
200 list of each file in the project along with its current file hash. | 209 |
201 This manifest is hashed similarly to the file hashes, incorporating | 210 In fact, we should always be able to do it with a single read, |
202 the hashes of the parent revisions. | 211 provided we know when and where to read. This is where the index comes |
212 in. Each revlog has an index containing a special hash (nodeid) of the | |
213 text, hashes for its parents, and where and how much of the revlog | |
214 data we need to read to reconstruct it. Thus, with one read of the | |
215 index and one read of the data, we can reconstruct any version in time | |
216 proportional to the object size. | |
217 | |
218 Similarly, revlogs and their indices are append-only. This means that | |
219 adding a new version is also O(1) seeks. | |
220 | |
221 Revlogs are used to represent all revisions of files, manifests, and | |
222 changesets. Compression for typical objects with lots of revisions can | |
223 range from 100 to 1 for things like project makefiles to over 2000 to | |
224 1 for objects like the manifest. | |
225 | |
226 | |
227 .Q. How are manifests and changesets stored? | |
228 | |
229 A manifest is simply a list of all files in a given revision of a | |
230 project along with the nodeids of the corresponding file revisions. So | |
231 grabbing a given version of the project means simply looking up its | |
232 manifest and reconstruction all the file revisions pointed to by it. | |
233 | |
234 A changeset is a list of all files changed in a check-in along with a | |
235 change description and some metadata like user and date. It also | |
236 contains a nodeid to the relevent revision of the manifest. | |
237 | |
238 | |
239 .Q. How do Mercurial hashes get calculated? | |
240 | |
241 Mercurial hashes both the contents of an object and the hash of its | |
242 parents to create an identifier that uniquely identifies an object's | |
243 contents and history. This greatly simplifies merging of histories | |
244 because it avoid graph cycles that can occur when a object is reverted | |
245 to an earlier state. | |
246 | |
247 All file revisions have an associated hash value. These are listed in | |
248 the manifest of a given project revision, and the manifest hash is | |
249 listed in the changeset. The changeset hash is again a hash of the | |
250 changeset contents and its parents, so it uniquely identifies the | |
251 entire history of the project to that point. | |
252 | |
253 | |
254 .Q. What checks are there on repository integrity? | |
255 | |
256 Every time a revlog object is retrieved, it is checked against its | |
257 hash for integrity. It is also incidentally doublechecked by the | |
258 Adler32 checksum used by the underlying zlib compression. | |
259 | |
260 Running 'hg verify' decompresses and reconstitutes each revision of | |
261 each object in the repository and cross-checks all of the index | |
262 metadata with those contents. | |
263 | |
264 But this alone is not enough to ensure that someone hasn't tampered | |
265 with a repository. For that, you need cryptographic signing. | |
266 | |
267 | |
268 .Q. How does signing work with Mercurial? | |
269 | |
270 Take a look at the hgeditor script for an example. The basic idea is | |
271 to use GPG to sign the manifest ID inside that changelog entry. The | |
272 manifest ID is a recursive hash of all of the files in the system and | |
273 their complete history, and thus signing the manifest hash signs the | |
274 entire project contents. | |
203 | 275 |
204 | 276 |
205 .Q. What about hash collisions? What about weaknesses in SHA1? | 277 .Q. What about hash collisions? What about weaknesses in SHA1? |
206 | 278 |
207 The SHA1 hashes are large enough that the odds of accidental hash collision | 279 The SHA1 hashes are large enough that the odds of accidental hash collision |
211 becomes a realistic concern. | 283 becomes a realistic concern. |
212 | 284 |
213 Collisions with the "short hashes" are not a concern as they're always | 285 Collisions with the "short hashes" are not a concern as they're always |
214 checked for ambiguity and are still long enough that they're not | 286 checked for ambiguity and are still long enough that they're not |
215 likely to happen for reasonably-sized projects (< 1M changes). | 287 likely to happen for reasonably-sized projects (< 1M changes). |
288 | |
289 | |
290 |