Show More
@@ -1,843 +1,846 | |||
|
1 | 1 | # obsolete.py - obsolete markers handling |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2012 Pierre-Yves David <pierre-yves.david@ens-lyon.org> |
|
4 | 4 | # Logilab SA <contact@logilab.fr> |
|
5 | 5 | # |
|
6 | 6 | # This software may be used and distributed according to the terms of the |
|
7 | 7 | # GNU General Public License version 2 or any later version. |
|
8 | 8 | |
|
9 | 9 | """Obsolete markers handling |
|
10 | 10 | |
|
11 | 11 | An obsolete marker maps an old changeset to a list of new |
|
12 | 12 | changesets. If the list of new changesets is empty, the old changeset |
|
13 | 13 | is said to be "killed". Otherwise, the old changeset is being |
|
14 | 14 | "replaced" by the new changesets. |
|
15 | 15 | |
|
16 | 16 | Obsolete markers can be used to record and distribute changeset graph |
|
17 | 17 | transformations performed by history rewriting operations, and help |
|
18 | 18 | building new tools to reconciliate conflicting rewriting actions. To |
|
19 | 19 | facilitate conflicts resolution, markers include various annotations |
|
20 | 20 | besides old and news changeset identifiers, such as creation date or |
|
21 | 21 | author name. |
|
22 | 22 | |
|
23 | 23 | The old obsoleted changeset is called "precursor" and possible replacements are |
|
24 | 24 | called "successors". Markers that used changeset X as a precursors are called |
|
25 | 25 | "successor markers of X" because they hold information about the successors of |
|
26 | 26 | X. Markers that use changeset Y as a successors are call "precursor markers of |
|
27 | 27 | Y" because they hold information about the precursors of Y. |
|
28 | 28 | |
|
29 | 29 | Examples: |
|
30 | 30 | |
|
31 | 31 | - When changeset A is replacement by a changeset A', one marker is stored: |
|
32 | 32 | |
|
33 | 33 | (A, (A')) |
|
34 | 34 | |
|
35 | 35 | - When changesets A and B are folded into a new changeset C two markers are |
|
36 | 36 | stored: |
|
37 | 37 | |
|
38 | 38 | (A, (C,)) and (B, (C,)) |
|
39 | 39 | |
|
40 | 40 | - When changeset A is simply "pruned" from the graph, a marker in create: |
|
41 | 41 | |
|
42 | 42 | (A, ()) |
|
43 | 43 | |
|
44 | 44 | - When changeset A is split into B and C, a single marker are used: |
|
45 | 45 | |
|
46 | 46 | (A, (C, C)) |
|
47 | 47 | |
|
48 | 48 | We use a single marker to distinct the "split" case from the "divergence" |
|
49 | 49 | case. If two independents operation rewrite the same changeset A in to A' and |
|
50 | 50 | A'' when have an error case: divergent rewriting. We can detect it because |
|
51 | 51 | two markers will be created independently: |
|
52 | 52 | |
|
53 | 53 | (A, (B,)) and (A, (C,)) |
|
54 | 54 | |
|
55 | 55 | Format |
|
56 | 56 | ------ |
|
57 | 57 | |
|
58 | 58 | Markers are stored in an append-only file stored in |
|
59 | 59 | '.hg/store/obsstore'. |
|
60 | 60 | |
|
61 | 61 | The file starts with a version header: |
|
62 | 62 | |
|
63 | 63 | - 1 unsigned byte: version number, starting at zero. |
|
64 | 64 | |
|
65 | 65 | |
|
66 | 66 | The header is followed by the markers. Each marker is made of: |
|
67 | 67 | |
|
68 | 68 | - 1 unsigned byte: number of new changesets "R", could be zero. |
|
69 | 69 | |
|
70 | 70 | - 1 unsigned 32-bits integer: metadata size "M" in bytes. |
|
71 | 71 | |
|
72 | 72 | - 1 byte: a bit field. It is reserved for flags used in obsolete |
|
73 | 73 | markers common operations, to avoid repeated decoding of metadata |
|
74 | 74 | entries. |
|
75 | 75 | |
|
76 | 76 | - 20 bytes: obsoleted changeset identifier. |
|
77 | 77 | |
|
78 | 78 | - N*20 bytes: new changesets identifiers. |
|
79 | 79 | |
|
80 | 80 | - M bytes: metadata as a sequence of nul-terminated strings. Each |
|
81 | 81 | string contains a key and a value, separated by a color ':', without |
|
82 | 82 | additional encoding. Keys cannot contain '\0' or ':' and values |
|
83 | 83 | cannot contain '\0'. |
|
84 | 84 | """ |
|
85 | 85 | import struct |
|
86 | 86 | import util, base85, node |
|
87 | 87 | import phases |
|
88 | 88 | from i18n import _ |
|
89 | 89 | |
|
90 | 90 | _pack = struct.pack |
|
91 | 91 | _unpack = struct.unpack |
|
92 | 92 | |
|
93 | 93 | _SEEK_END = 2 # os.SEEK_END was introduced in Python 2.5 |
|
94 | 94 | |
|
95 | 95 | # the obsolete feature is not mature enough to be enabled by default. |
|
96 | 96 | # you have to rely on third party extension extension to enable this. |
|
97 | 97 | _enabled = False |
|
98 | 98 | |
|
99 | 99 | # data used for parsing and writing |
|
100 | 100 | _fmversion = 0 |
|
101 | 101 | _fmfixed = '>BIB20s' |
|
102 | 102 | _fmnode = '20s' |
|
103 | 103 | _fmfsize = struct.calcsize(_fmfixed) |
|
104 | 104 | _fnodesize = struct.calcsize(_fmnode) |
|
105 | 105 | |
|
106 | 106 | ### obsolescence marker flag |
|
107 | 107 | |
|
108 | 108 | ## bumpedfix flag |
|
109 | 109 | # |
|
110 | 110 | # When a changeset A' succeed to a changeset A which became public, we call A' |
|
111 | 111 | # "bumped" because it's a successors of a public changesets |
|
112 | 112 | # |
|
113 | 113 | # o A' (bumped) |
|
114 | 114 | # |`: |
|
115 | 115 | # | o A |
|
116 | 116 | # |/ |
|
117 | 117 | # o Z |
|
118 | 118 | # |
|
119 | 119 | # The way to solve this situation is to create a new changeset Ad as children |
|
120 | 120 | # of A. This changeset have the same content than A'. So the diff from A to A' |
|
121 | 121 | # is the same than the diff from A to Ad. Ad is marked as a successors of A' |
|
122 | 122 | # |
|
123 | 123 | # o Ad |
|
124 | 124 | # |`: |
|
125 | 125 | # | x A' |
|
126 | 126 | # |'| |
|
127 | 127 | # o | A |
|
128 | 128 | # |/ |
|
129 | 129 | # o Z |
|
130 | 130 | # |
|
131 | 131 | # But by transitivity Ad is also a successors of A. To avoid having Ad marked |
|
132 | 132 | # as bumped too, we add the `bumpedfix` flag to the marker. <A', (Ad,)>. |
|
133 | 133 | # This flag mean that the successors express the changes between the public and |
|
134 | 134 | # bumped version and fix the situation, breaking the transitivity of |
|
135 | 135 | # "bumped" here. |
|
136 | 136 | bumpedfix = 1 |
|
137 | 137 | |
|
138 | 138 | def _readmarkers(data): |
|
139 | 139 | """Read and enumerate markers from raw data""" |
|
140 | 140 | off = 0 |
|
141 | 141 | diskversion = _unpack('>B', data[off:off + 1])[0] |
|
142 | 142 | off += 1 |
|
143 | 143 | if diskversion != _fmversion: |
|
144 | 144 | raise util.Abort(_('parsing obsolete marker: unknown version %r') |
|
145 | 145 | % diskversion) |
|
146 | 146 | |
|
147 | 147 | # Loop on markers |
|
148 | 148 | l = len(data) |
|
149 | 149 | while off + _fmfsize <= l: |
|
150 | 150 | # read fixed part |
|
151 | 151 | cur = data[off:off + _fmfsize] |
|
152 | 152 | off += _fmfsize |
|
153 | 153 | nbsuc, mdsize, flags, pre = _unpack(_fmfixed, cur) |
|
154 | 154 | # read replacement |
|
155 | 155 | sucs = () |
|
156 | 156 | if nbsuc: |
|
157 | 157 | s = (_fnodesize * nbsuc) |
|
158 | 158 | cur = data[off:off + s] |
|
159 | 159 | sucs = _unpack(_fmnode * nbsuc, cur) |
|
160 | 160 | off += s |
|
161 | 161 | # read metadata |
|
162 | 162 | # (metadata will be decoded on demand) |
|
163 | 163 | metadata = data[off:off + mdsize] |
|
164 | 164 | if len(metadata) != mdsize: |
|
165 | 165 | raise util.Abort(_('parsing obsolete marker: metadata is too ' |
|
166 | 166 | 'short, %d bytes expected, got %d') |
|
167 | 167 | % (mdsize, len(metadata))) |
|
168 | 168 | off += mdsize |
|
169 | 169 | yield (pre, sucs, flags, metadata) |
|
170 | 170 | |
|
171 | 171 | def encodemeta(meta): |
|
172 | 172 | """Return encoded metadata string to string mapping. |
|
173 | 173 | |
|
174 | 174 | Assume no ':' in key and no '\0' in both key and value.""" |
|
175 | 175 | for key, value in meta.iteritems(): |
|
176 | 176 | if ':' in key or '\0' in key: |
|
177 | 177 | raise ValueError("':' and '\0' are forbidden in metadata key'") |
|
178 | 178 | if '\0' in value: |
|
179 | 179 | raise ValueError("':' are forbidden in metadata value'") |
|
180 | 180 | return '\0'.join(['%s:%s' % (k, meta[k]) for k in sorted(meta)]) |
|
181 | 181 | |
|
182 | 182 | def decodemeta(data): |
|
183 | 183 | """Return string to string dictionary from encoded version.""" |
|
184 | 184 | d = {} |
|
185 | 185 | for l in data.split('\0'): |
|
186 | 186 | if l: |
|
187 | 187 | key, value = l.split(':') |
|
188 | 188 | d[key] = value |
|
189 | 189 | return d |
|
190 | 190 | |
|
191 | 191 | class marker(object): |
|
192 | 192 | """Wrap obsolete marker raw data""" |
|
193 | 193 | |
|
194 | 194 | def __init__(self, repo, data): |
|
195 | 195 | # the repo argument will be used to create changectx in later version |
|
196 | 196 | self._repo = repo |
|
197 | 197 | self._data = data |
|
198 | 198 | self._decodedmeta = None |
|
199 | 199 | |
|
200 | 200 | def __hash__(self): |
|
201 | 201 | return hash(self._data) |
|
202 | 202 | |
|
203 | 203 | def __eq__(self, other): |
|
204 | 204 | if type(other) != type(self): |
|
205 | 205 | return False |
|
206 | 206 | return self._data == other._data |
|
207 | 207 | |
|
208 | 208 | def precnode(self): |
|
209 | 209 | """Precursor changeset node identifier""" |
|
210 | 210 | return self._data[0] |
|
211 | 211 | |
|
212 | 212 | def succnodes(self): |
|
213 | 213 | """List of successor changesets node identifiers""" |
|
214 | 214 | return self._data[1] |
|
215 | 215 | |
|
216 | 216 | def metadata(self): |
|
217 | 217 | """Decoded metadata dictionary""" |
|
218 | 218 | if self._decodedmeta is None: |
|
219 | 219 | self._decodedmeta = decodemeta(self._data[3]) |
|
220 | 220 | return self._decodedmeta |
|
221 | 221 | |
|
222 | 222 | def date(self): |
|
223 | 223 | """Creation date as (unixtime, offset)""" |
|
224 | 224 | parts = self.metadata()['date'].split(' ') |
|
225 | 225 | return (float(parts[0]), int(parts[1])) |
|
226 | 226 | |
|
227 | 227 | class obsstore(object): |
|
228 | 228 | """Store obsolete markers |
|
229 | 229 | |
|
230 | 230 | Markers can be accessed with two mappings: |
|
231 | 231 | - precursors[x] -> set(markers on precursors edges of x) |
|
232 | 232 | - successors[x] -> set(markers on successors edges of x) |
|
233 | 233 | """ |
|
234 | 234 | |
|
235 | 235 | def __init__(self, sopener): |
|
236 | 236 | # caches for various obsolescence related cache |
|
237 | 237 | self.caches = {} |
|
238 | 238 | self._all = [] |
|
239 | 239 | # new markers to serialize |
|
240 | 240 | self.precursors = {} |
|
241 | 241 | self.successors = {} |
|
242 | 242 | self.sopener = sopener |
|
243 | 243 | data = sopener.tryread('obsstore') |
|
244 | 244 | if data: |
|
245 | 245 | self._load(_readmarkers(data)) |
|
246 | 246 | |
|
247 | 247 | def __iter__(self): |
|
248 | 248 | return iter(self._all) |
|
249 | 249 | |
|
250 | 250 | def __nonzero__(self): |
|
251 | 251 | return bool(self._all) |
|
252 | 252 | |
|
253 | 253 | def create(self, transaction, prec, succs=(), flag=0, metadata=None): |
|
254 | 254 | """obsolete: add a new obsolete marker |
|
255 | 255 | |
|
256 | 256 | * ensuring it is hashable |
|
257 | 257 | * check mandatory metadata |
|
258 | 258 | * encode metadata |
|
259 | ||
|
260 | If you are a human writing code creating marker you want to use the | |
|
261 | `createmarkers` function in this module instead. | |
|
259 | 262 | """ |
|
260 | 263 | if metadata is None: |
|
261 | 264 | metadata = {} |
|
262 | 265 | if 'date' not in metadata: |
|
263 | 266 | metadata['date'] = "%d %d" % util.makedate() |
|
264 | 267 | if len(prec) != 20: |
|
265 | 268 | raise ValueError(prec) |
|
266 | 269 | for succ in succs: |
|
267 | 270 | if len(succ) != 20: |
|
268 | 271 | raise ValueError(succ) |
|
269 | 272 | marker = (str(prec), tuple(succs), int(flag), encodemeta(metadata)) |
|
270 | 273 | self.add(transaction, [marker]) |
|
271 | 274 | |
|
272 | 275 | def add(self, transaction, markers): |
|
273 | 276 | """Add new markers to the store |
|
274 | 277 | |
|
275 | 278 | Take care of filtering duplicate. |
|
276 | 279 | Return the number of new marker.""" |
|
277 | 280 | if not _enabled: |
|
278 | 281 | raise util.Abort('obsolete feature is not enabled on this repo') |
|
279 | 282 | known = set(self._all) |
|
280 | 283 | new = [] |
|
281 | 284 | for m in markers: |
|
282 | 285 | if m not in known: |
|
283 | 286 | known.add(m) |
|
284 | 287 | new.append(m) |
|
285 | 288 | if new: |
|
286 | 289 | f = self.sopener('obsstore', 'ab') |
|
287 | 290 | try: |
|
288 | 291 | # Whether the file's current position is at the begin or at |
|
289 | 292 | # the end after opening a file for appending is implementation |
|
290 | 293 | # defined. So we must seek to the end before calling tell(), |
|
291 | 294 | # or we may get a zero offset for non-zero sized files on |
|
292 | 295 | # some platforms (issue3543). |
|
293 | 296 | f.seek(0, _SEEK_END) |
|
294 | 297 | offset = f.tell() |
|
295 | 298 | transaction.add('obsstore', offset) |
|
296 | 299 | # offset == 0: new file - add the version header |
|
297 | 300 | for bytes in _encodemarkers(new, offset == 0): |
|
298 | 301 | f.write(bytes) |
|
299 | 302 | finally: |
|
300 | 303 | # XXX: f.close() == filecache invalidation == obsstore rebuilt. |
|
301 | 304 | # call 'filecacheentry.refresh()' here |
|
302 | 305 | f.close() |
|
303 | 306 | self._load(new) |
|
304 | 307 | # new marker *may* have changed several set. invalidate the cache. |
|
305 | 308 | self.caches.clear() |
|
306 | 309 | return len(new) |
|
307 | 310 | |
|
308 | 311 | def mergemarkers(self, transaction, data): |
|
309 | 312 | markers = _readmarkers(data) |
|
310 | 313 | self.add(transaction, markers) |
|
311 | 314 | |
|
312 | 315 | def _load(self, markers): |
|
313 | 316 | for mark in markers: |
|
314 | 317 | self._all.append(mark) |
|
315 | 318 | pre, sucs = mark[:2] |
|
316 | 319 | self.successors.setdefault(pre, set()).add(mark) |
|
317 | 320 | for suc in sucs: |
|
318 | 321 | self.precursors.setdefault(suc, set()).add(mark) |
|
319 | 322 | if node.nullid in self.precursors: |
|
320 | 323 | raise util.Abort(_('bad obsolescence marker detected: ' |
|
321 | 324 | 'invalid successors nullid')) |
|
322 | 325 | |
|
323 | 326 | def _encodemarkers(markers, addheader=False): |
|
324 | 327 | # Kept separate from flushmarkers(), it will be reused for |
|
325 | 328 | # markers exchange. |
|
326 | 329 | if addheader: |
|
327 | 330 | yield _pack('>B', _fmversion) |
|
328 | 331 | for marker in markers: |
|
329 | 332 | yield _encodeonemarker(marker) |
|
330 | 333 | |
|
331 | 334 | |
|
332 | 335 | def _encodeonemarker(marker): |
|
333 | 336 | pre, sucs, flags, metadata = marker |
|
334 | 337 | nbsuc = len(sucs) |
|
335 | 338 | format = _fmfixed + (_fmnode * nbsuc) |
|
336 | 339 | data = [nbsuc, len(metadata), flags, pre] |
|
337 | 340 | data.extend(sucs) |
|
338 | 341 | return _pack(format, *data) + metadata |
|
339 | 342 | |
|
340 | 343 | # arbitrary picked to fit into 8K limit from HTTP server |
|
341 | 344 | # you have to take in account: |
|
342 | 345 | # - the version header |
|
343 | 346 | # - the base85 encoding |
|
344 | 347 | _maxpayload = 5300 |
|
345 | 348 | |
|
346 | 349 | def listmarkers(repo): |
|
347 | 350 | """List markers over pushkey""" |
|
348 | 351 | if not repo.obsstore: |
|
349 | 352 | return {} |
|
350 | 353 | keys = {} |
|
351 | 354 | parts = [] |
|
352 | 355 | currentlen = _maxpayload * 2 # ensure we create a new part |
|
353 | 356 | for marker in repo.obsstore: |
|
354 | 357 | nextdata = _encodeonemarker(marker) |
|
355 | 358 | if (len(nextdata) + currentlen > _maxpayload): |
|
356 | 359 | currentpart = [] |
|
357 | 360 | currentlen = 0 |
|
358 | 361 | parts.append(currentpart) |
|
359 | 362 | currentpart.append(nextdata) |
|
360 | 363 | currentlen += len(nextdata) |
|
361 | 364 | for idx, part in enumerate(reversed(parts)): |
|
362 | 365 | data = ''.join([_pack('>B', _fmversion)] + part) |
|
363 | 366 | keys['dump%i' % idx] = base85.b85encode(data) |
|
364 | 367 | return keys |
|
365 | 368 | |
|
366 | 369 | def pushmarker(repo, key, old, new): |
|
367 | 370 | """Push markers over pushkey""" |
|
368 | 371 | if not key.startswith('dump'): |
|
369 | 372 | repo.ui.warn(_('unknown key: %r') % key) |
|
370 | 373 | return 0 |
|
371 | 374 | if old: |
|
372 | 375 | repo.ui.warn(_('unexpected old value') % key) |
|
373 | 376 | return 0 |
|
374 | 377 | data = base85.b85decode(new) |
|
375 | 378 | lock = repo.lock() |
|
376 | 379 | try: |
|
377 | 380 | tr = repo.transaction('pushkey: obsolete markers') |
|
378 | 381 | try: |
|
379 | 382 | repo.obsstore.mergemarkers(tr, data) |
|
380 | 383 | tr.close() |
|
381 | 384 | return 1 |
|
382 | 385 | finally: |
|
383 | 386 | tr.release() |
|
384 | 387 | finally: |
|
385 | 388 | lock.release() |
|
386 | 389 | |
|
387 | 390 | def allmarkers(repo): |
|
388 | 391 | """all obsolete markers known in a repository""" |
|
389 | 392 | for markerdata in repo.obsstore: |
|
390 | 393 | yield marker(repo, markerdata) |
|
391 | 394 | |
|
392 | 395 | def precursormarkers(ctx): |
|
393 | 396 | """obsolete marker marking this changeset as a successors""" |
|
394 | 397 | for data in ctx._repo.obsstore.precursors.get(ctx.node(), ()): |
|
395 | 398 | yield marker(ctx._repo, data) |
|
396 | 399 | |
|
397 | 400 | def successormarkers(ctx): |
|
398 | 401 | """obsolete marker making this changeset obsolete""" |
|
399 | 402 | for data in ctx._repo.obsstore.successors.get(ctx.node(), ()): |
|
400 | 403 | yield marker(ctx._repo, data) |
|
401 | 404 | |
|
402 | 405 | def allsuccessors(obsstore, nodes, ignoreflags=0): |
|
403 | 406 | """Yield node for every successor of <nodes>. |
|
404 | 407 | |
|
405 | 408 | Some successors may be unknown locally. |
|
406 | 409 | |
|
407 | 410 | This is a linear yield unsuited to detecting split changesets. It includes |
|
408 | 411 | initial nodes too.""" |
|
409 | 412 | remaining = set(nodes) |
|
410 | 413 | seen = set(remaining) |
|
411 | 414 | while remaining: |
|
412 | 415 | current = remaining.pop() |
|
413 | 416 | yield current |
|
414 | 417 | for mark in obsstore.successors.get(current, ()): |
|
415 | 418 | # ignore marker flagged with specified flag |
|
416 | 419 | if mark[2] & ignoreflags: |
|
417 | 420 | continue |
|
418 | 421 | for suc in mark[1]: |
|
419 | 422 | if suc not in seen: |
|
420 | 423 | seen.add(suc) |
|
421 | 424 | remaining.add(suc) |
|
422 | 425 | |
|
423 | 426 | def allprecursors(obsstore, nodes, ignoreflags=0): |
|
424 | 427 | """Yield node for every precursors of <nodes>. |
|
425 | 428 | |
|
426 | 429 | Some precursors may be unknown locally. |
|
427 | 430 | |
|
428 | 431 | This is a linear yield unsuited to detecting folded changesets. It includes |
|
429 | 432 | initial nodes too.""" |
|
430 | 433 | |
|
431 | 434 | remaining = set(nodes) |
|
432 | 435 | seen = set(remaining) |
|
433 | 436 | while remaining: |
|
434 | 437 | current = remaining.pop() |
|
435 | 438 | yield current |
|
436 | 439 | for mark in obsstore.precursors.get(current, ()): |
|
437 | 440 | # ignore marker flagged with specified flag |
|
438 | 441 | if mark[2] & ignoreflags: |
|
439 | 442 | continue |
|
440 | 443 | suc = mark[0] |
|
441 | 444 | if suc not in seen: |
|
442 | 445 | seen.add(suc) |
|
443 | 446 | remaining.add(suc) |
|
444 | 447 | |
|
445 | 448 | def foreground(repo, nodes): |
|
446 | 449 | """return all nodes in the "foreground" of other node |
|
447 | 450 | |
|
448 | 451 | The foreground of a revision is anything reachable using parent -> children |
|
449 | 452 | or precursor -> successor relation. It is very similar to "descendant" but |
|
450 | 453 | augmented with obsolescence information. |
|
451 | 454 | |
|
452 | 455 | Beware that possible obsolescence cycle may result if complex situation. |
|
453 | 456 | """ |
|
454 | 457 | repo = repo.unfiltered() |
|
455 | 458 | foreground = set(repo.set('%ln::', nodes)) |
|
456 | 459 | if repo.obsstore: |
|
457 | 460 | # We only need this complicated logic if there is obsolescence |
|
458 | 461 | # XXX will probably deserve an optimised revset. |
|
459 | 462 | nm = repo.changelog.nodemap |
|
460 | 463 | plen = -1 |
|
461 | 464 | # compute the whole set of successors or descendants |
|
462 | 465 | while len(foreground) != plen: |
|
463 | 466 | plen = len(foreground) |
|
464 | 467 | succs = set(c.node() for c in foreground) |
|
465 | 468 | mutable = [c.node() for c in foreground if c.mutable()] |
|
466 | 469 | succs.update(allsuccessors(repo.obsstore, mutable)) |
|
467 | 470 | known = (n for n in succs if n in nm) |
|
468 | 471 | foreground = set(repo.set('%ln::', known)) |
|
469 | 472 | return set(c.node() for c in foreground) |
|
470 | 473 | |
|
471 | 474 | |
|
472 | 475 | def successorssets(repo, initialnode, cache=None): |
|
473 | 476 | """Return all set of successors of initial nodes |
|
474 | 477 | |
|
475 | 478 | The successors set of a changeset A are a group of revisions that succeed |
|
476 | 479 | A. It succeeds A as a consistent whole, each revision being only a partial |
|
477 | 480 | replacement. The successors set contains non-obsolete changesets only. |
|
478 | 481 | |
|
479 | 482 | This function returns the full list of successor sets which is why it |
|
480 | 483 | returns a list of tuples and not just a single tuple. Each tuple is a valid |
|
481 | 484 | successors set. Not that (A,) may be a valid successors set for changeset A |
|
482 | 485 | (see below). |
|
483 | 486 | |
|
484 | 487 | In most cases, a changeset A will have a single element (e.g. the changeset |
|
485 | 488 | A is replaced by A') in its successors set. Though, it is also common for a |
|
486 | 489 | changeset A to have no elements in its successor set (e.g. the changeset |
|
487 | 490 | has been pruned). Therefore, the returned list of successors sets will be |
|
488 | 491 | [(A',)] or [], respectively. |
|
489 | 492 | |
|
490 | 493 | When a changeset A is split into A' and B', however, it will result in a |
|
491 | 494 | successors set containing more than a single element, i.e. [(A',B')]. |
|
492 | 495 | Divergent changesets will result in multiple successors sets, i.e. [(A',), |
|
493 | 496 | (A'')]. |
|
494 | 497 | |
|
495 | 498 | If a changeset A is not obsolete, then it will conceptually have no |
|
496 | 499 | successors set. To distinguish this from a pruned changeset, the successor |
|
497 | 500 | set will only contain itself, i.e. [(A,)]. |
|
498 | 501 | |
|
499 | 502 | Finally, successors unknown locally are considered to be pruned (obsoleted |
|
500 | 503 | without any successors). |
|
501 | 504 | |
|
502 | 505 | The optional `cache` parameter is a dictionary that may contain precomputed |
|
503 | 506 | successors sets. It is meant to reuse the computation of a previous call to |
|
504 | 507 | `successorssets` when multiple calls are made at the same time. The cache |
|
505 | 508 | dictionary is updated in place. The caller is responsible for its live |
|
506 | 509 | spawn. Code that makes multiple calls to `successorssets` *must* use this |
|
507 | 510 | cache mechanism or suffer terrible performances. |
|
508 | 511 | |
|
509 | 512 | """ |
|
510 | 513 | |
|
511 | 514 | succmarkers = repo.obsstore.successors |
|
512 | 515 | |
|
513 | 516 | # Stack of nodes we search successors sets for |
|
514 | 517 | toproceed = [initialnode] |
|
515 | 518 | # set version of above list for fast loop detection |
|
516 | 519 | # element added to "toproceed" must be added here |
|
517 | 520 | stackedset = set(toproceed) |
|
518 | 521 | if cache is None: |
|
519 | 522 | cache = {} |
|
520 | 523 | |
|
521 | 524 | # This while loop is the flattened version of a recursive search for |
|
522 | 525 | # successors sets |
|
523 | 526 | # |
|
524 | 527 | # def successorssets(x): |
|
525 | 528 | # successors = directsuccessors(x) |
|
526 | 529 | # ss = [[]] |
|
527 | 530 | # for succ in directsuccessors(x): |
|
528 | 531 | # # product as in itertools cartesian product |
|
529 | 532 | # ss = product(ss, successorssets(succ)) |
|
530 | 533 | # return ss |
|
531 | 534 | # |
|
532 | 535 | # But we can not use plain recursive calls here: |
|
533 | 536 | # - that would blow the python call stack |
|
534 | 537 | # - obsolescence markers may have cycles, we need to handle them. |
|
535 | 538 | # |
|
536 | 539 | # The `toproceed` list act as our call stack. Every node we search |
|
537 | 540 | # successors set for are stacked there. |
|
538 | 541 | # |
|
539 | 542 | # The `stackedset` is set version of this stack used to check if a node is |
|
540 | 543 | # already stacked. This check is used to detect cycles and prevent infinite |
|
541 | 544 | # loop. |
|
542 | 545 | # |
|
543 | 546 | # successors set of all nodes are stored in the `cache` dictionary. |
|
544 | 547 | # |
|
545 | 548 | # After this while loop ends we use the cache to return the successors sets |
|
546 | 549 | # for the node requested by the caller. |
|
547 | 550 | while toproceed: |
|
548 | 551 | # Every iteration tries to compute the successors sets of the topmost |
|
549 | 552 | # node of the stack: CURRENT. |
|
550 | 553 | # |
|
551 | 554 | # There are four possible outcomes: |
|
552 | 555 | # |
|
553 | 556 | # 1) We already know the successors sets of CURRENT: |
|
554 | 557 | # -> mission accomplished, pop it from the stack. |
|
555 | 558 | # 2) Node is not obsolete: |
|
556 | 559 | # -> the node is its own successors sets. Add it to the cache. |
|
557 | 560 | # 3) We do not know successors set of direct successors of CURRENT: |
|
558 | 561 | # -> We add those successors to the stack. |
|
559 | 562 | # 4) We know successors sets of all direct successors of CURRENT: |
|
560 | 563 | # -> We can compute CURRENT successors set and add it to the |
|
561 | 564 | # cache. |
|
562 | 565 | # |
|
563 | 566 | current = toproceed[-1] |
|
564 | 567 | if current in cache: |
|
565 | 568 | # case (1): We already know the successors sets |
|
566 | 569 | stackedset.remove(toproceed.pop()) |
|
567 | 570 | elif current not in succmarkers: |
|
568 | 571 | # case (2): The node is not obsolete. |
|
569 | 572 | if current in repo: |
|
570 | 573 | # We have a valid last successors. |
|
571 | 574 | cache[current] = [(current,)] |
|
572 | 575 | else: |
|
573 | 576 | # Final obsolete version is unknown locally. |
|
574 | 577 | # Do not count that as a valid successors |
|
575 | 578 | cache[current] = [] |
|
576 | 579 | else: |
|
577 | 580 | # cases (3) and (4) |
|
578 | 581 | # |
|
579 | 582 | # We proceed in two phases. Phase 1 aims to distinguish case (3) |
|
580 | 583 | # from case (4): |
|
581 | 584 | # |
|
582 | 585 | # For each direct successors of CURRENT, we check whether its |
|
583 | 586 | # successors sets are known. If they are not, we stack the |
|
584 | 587 | # unknown node and proceed to the next iteration of the while |
|
585 | 588 | # loop. (case 3) |
|
586 | 589 | # |
|
587 | 590 | # During this step, we may detect obsolescence cycles: a node |
|
588 | 591 | # with unknown successors sets but already in the call stack. |
|
589 | 592 | # In such a situation, we arbitrary set the successors sets of |
|
590 | 593 | # the node to nothing (node pruned) to break the cycle. |
|
591 | 594 | # |
|
592 | 595 | # If no break was encountered we proceed to phase 2. |
|
593 | 596 | # |
|
594 | 597 | # Phase 2 computes successors sets of CURRENT (case 4); see details |
|
595 | 598 | # in phase 2 itself. |
|
596 | 599 | # |
|
597 | 600 | # Note the two levels of iteration in each phase. |
|
598 | 601 | # - The first one handles obsolescence markers using CURRENT as |
|
599 | 602 | # precursor (successors markers of CURRENT). |
|
600 | 603 | # |
|
601 | 604 | # Having multiple entry here means divergence. |
|
602 | 605 | # |
|
603 | 606 | # - The second one handles successors defined in each marker. |
|
604 | 607 | # |
|
605 | 608 | # Having none means pruned node, multiple successors means split, |
|
606 | 609 | # single successors are standard replacement. |
|
607 | 610 | # |
|
608 | 611 | for mark in sorted(succmarkers[current]): |
|
609 | 612 | for suc in mark[1]: |
|
610 | 613 | if suc not in cache: |
|
611 | 614 | if suc in stackedset: |
|
612 | 615 | # cycle breaking |
|
613 | 616 | cache[suc] = [] |
|
614 | 617 | else: |
|
615 | 618 | # case (3) If we have not computed successors sets |
|
616 | 619 | # of one of those successors we add it to the |
|
617 | 620 | # `toproceed` stack and stop all work for this |
|
618 | 621 | # iteration. |
|
619 | 622 | toproceed.append(suc) |
|
620 | 623 | stackedset.add(suc) |
|
621 | 624 | break |
|
622 | 625 | else: |
|
623 | 626 | continue |
|
624 | 627 | break |
|
625 | 628 | else: |
|
626 | 629 | # case (4): we know all successors sets of all direct |
|
627 | 630 | # successors |
|
628 | 631 | # |
|
629 | 632 | # Successors set contributed by each marker depends on the |
|
630 | 633 | # successors sets of all its "successors" node. |
|
631 | 634 | # |
|
632 | 635 | # Each different marker is a divergence in the obsolescence |
|
633 | 636 | # history. It contributes successors sets distinct from other |
|
634 | 637 | # markers. |
|
635 | 638 | # |
|
636 | 639 | # Within a marker, a successor may have divergent successors |
|
637 | 640 | # sets. In such a case, the marker will contribute multiple |
|
638 | 641 | # divergent successors sets. If multiple successors have |
|
639 | 642 | # divergent successors sets, a cartesian product is used. |
|
640 | 643 | # |
|
641 | 644 | # At the end we post-process successors sets to remove |
|
642 | 645 | # duplicated entry and successors set that are strict subset of |
|
643 | 646 | # another one. |
|
644 | 647 | succssets = [] |
|
645 | 648 | for mark in sorted(succmarkers[current]): |
|
646 | 649 | # successors sets contributed by this marker |
|
647 | 650 | markss = [[]] |
|
648 | 651 | for suc in mark[1]: |
|
649 | 652 | # cardinal product with previous successors |
|
650 | 653 | productresult = [] |
|
651 | 654 | for prefix in markss: |
|
652 | 655 | for suffix in cache[suc]: |
|
653 | 656 | newss = list(prefix) |
|
654 | 657 | for part in suffix: |
|
655 | 658 | # do not duplicated entry in successors set |
|
656 | 659 | # first entry wins. |
|
657 | 660 | if part not in newss: |
|
658 | 661 | newss.append(part) |
|
659 | 662 | productresult.append(newss) |
|
660 | 663 | markss = productresult |
|
661 | 664 | succssets.extend(markss) |
|
662 | 665 | # remove duplicated and subset |
|
663 | 666 | seen = [] |
|
664 | 667 | final = [] |
|
665 | 668 | candidate = sorted(((set(s), s) for s in succssets if s), |
|
666 | 669 | key=lambda x: len(x[1]), reverse=True) |
|
667 | 670 | for setversion, listversion in candidate: |
|
668 | 671 | for seenset in seen: |
|
669 | 672 | if setversion.issubset(seenset): |
|
670 | 673 | break |
|
671 | 674 | else: |
|
672 | 675 | final.append(listversion) |
|
673 | 676 | seen.append(setversion) |
|
674 | 677 | final.reverse() # put small successors set first |
|
675 | 678 | cache[current] = final |
|
676 | 679 | return cache[initialnode] |
|
677 | 680 | |
|
678 | 681 | def _knownrevs(repo, nodes): |
|
679 | 682 | """yield revision numbers of known nodes passed in parameters |
|
680 | 683 | |
|
681 | 684 | Unknown revisions are silently ignored.""" |
|
682 | 685 | torev = repo.changelog.nodemap.get |
|
683 | 686 | for n in nodes: |
|
684 | 687 | rev = torev(n) |
|
685 | 688 | if rev is not None: |
|
686 | 689 | yield rev |
|
687 | 690 | |
|
688 | 691 | # mapping of 'set-name' -> <function to compute this set> |
|
689 | 692 | cachefuncs = {} |
|
690 | 693 | def cachefor(name): |
|
691 | 694 | """Decorator to register a function as computing the cache for a set""" |
|
692 | 695 | def decorator(func): |
|
693 | 696 | assert name not in cachefuncs |
|
694 | 697 | cachefuncs[name] = func |
|
695 | 698 | return func |
|
696 | 699 | return decorator |
|
697 | 700 | |
|
698 | 701 | def getrevs(repo, name): |
|
699 | 702 | """Return the set of revision that belong to the <name> set |
|
700 | 703 | |
|
701 | 704 | Such access may compute the set and cache it for future use""" |
|
702 | 705 | repo = repo.unfiltered() |
|
703 | 706 | if not repo.obsstore: |
|
704 | 707 | return () |
|
705 | 708 | if name not in repo.obsstore.caches: |
|
706 | 709 | repo.obsstore.caches[name] = cachefuncs[name](repo) |
|
707 | 710 | return repo.obsstore.caches[name] |
|
708 | 711 | |
|
709 | 712 | # To be simple we need to invalidate obsolescence cache when: |
|
710 | 713 | # |
|
711 | 714 | # - new changeset is added: |
|
712 | 715 | # - public phase is changed |
|
713 | 716 | # - obsolescence marker are added |
|
714 | 717 | # - strip is used a repo |
|
715 | 718 | def clearobscaches(repo): |
|
716 | 719 | """Remove all obsolescence related cache from a repo |
|
717 | 720 | |
|
718 | 721 | This remove all cache in obsstore is the obsstore already exist on the |
|
719 | 722 | repo. |
|
720 | 723 | |
|
721 | 724 | (We could be smarter here given the exact event that trigger the cache |
|
722 | 725 | clearing)""" |
|
723 | 726 | # only clear cache is there is obsstore data in this repo |
|
724 | 727 | if 'obsstore' in repo._filecache: |
|
725 | 728 | repo.obsstore.caches.clear() |
|
726 | 729 | |
|
727 | 730 | @cachefor('obsolete') |
|
728 | 731 | def _computeobsoleteset(repo): |
|
729 | 732 | """the set of obsolete revisions""" |
|
730 | 733 | obs = set() |
|
731 | 734 | getrev = repo.changelog.nodemap.get |
|
732 | 735 | getphase = repo._phasecache.phase |
|
733 | 736 | for node in repo.obsstore.successors: |
|
734 | 737 | rev = getrev(node) |
|
735 | 738 | if rev is not None and getphase(repo, rev): |
|
736 | 739 | obs.add(rev) |
|
737 | 740 | return obs |
|
738 | 741 | |
|
739 | 742 | @cachefor('unstable') |
|
740 | 743 | def _computeunstableset(repo): |
|
741 | 744 | """the set of non obsolete revisions with obsolete parents""" |
|
742 | 745 | # revset is not efficient enough here |
|
743 | 746 | # we do (obsolete()::) - obsolete() by hand |
|
744 | 747 | obs = getrevs(repo, 'obsolete') |
|
745 | 748 | if not obs: |
|
746 | 749 | return set() |
|
747 | 750 | cl = repo.changelog |
|
748 | 751 | return set(r for r in cl.descendants(obs) if r not in obs) |
|
749 | 752 | |
|
750 | 753 | @cachefor('suspended') |
|
751 | 754 | def _computesuspendedset(repo): |
|
752 | 755 | """the set of obsolete parents with non obsolete descendants""" |
|
753 | 756 | suspended = repo.changelog.ancestors(getrevs(repo, 'unstable')) |
|
754 | 757 | return set(r for r in getrevs(repo, 'obsolete') if r in suspended) |
|
755 | 758 | |
|
756 | 759 | @cachefor('extinct') |
|
757 | 760 | def _computeextinctset(repo): |
|
758 | 761 | """the set of obsolete parents without non obsolete descendants""" |
|
759 | 762 | return getrevs(repo, 'obsolete') - getrevs(repo, 'suspended') |
|
760 | 763 | |
|
761 | 764 | |
|
762 | 765 | @cachefor('bumped') |
|
763 | 766 | def _computebumpedset(repo): |
|
764 | 767 | """the set of revs trying to obsolete public revisions""" |
|
765 | 768 | bumped = set() |
|
766 | 769 | # utils function (avoid attribut lookup in the loop) |
|
767 | 770 | phase = repo._phasecache.phase # would be faster to grab the full list |
|
768 | 771 | public = phases.public |
|
769 | 772 | cl = repo.changelog |
|
770 | 773 | torev = cl.nodemap.get |
|
771 | 774 | obs = getrevs(repo, 'obsolete') |
|
772 | 775 | for rev in repo: |
|
773 | 776 | # We only evaluate mutable, non-obsolete revision |
|
774 | 777 | if (public < phase(repo, rev)) and (rev not in obs): |
|
775 | 778 | node = cl.node(rev) |
|
776 | 779 | # (future) A cache of precursors may worth if split is very common |
|
777 | 780 | for pnode in allprecursors(repo.obsstore, [node], |
|
778 | 781 | ignoreflags=bumpedfix): |
|
779 | 782 | prev = torev(pnode) # unfiltered! but so is phasecache |
|
780 | 783 | if (prev is not None) and (phase(repo, prev) <= public): |
|
781 | 784 | # we have a public precursors |
|
782 | 785 | bumped.add(rev) |
|
783 | 786 | break # Next draft! |
|
784 | 787 | return bumped |
|
785 | 788 | |
|
786 | 789 | @cachefor('divergent') |
|
787 | 790 | def _computedivergentset(repo): |
|
788 | 791 | """the set of rev that compete to be the final successors of some revision. |
|
789 | 792 | """ |
|
790 | 793 | divergent = set() |
|
791 | 794 | obsstore = repo.obsstore |
|
792 | 795 | newermap = {} |
|
793 | 796 | for ctx in repo.set('(not public()) - obsolete()'): |
|
794 | 797 | mark = obsstore.precursors.get(ctx.node(), ()) |
|
795 | 798 | toprocess = set(mark) |
|
796 | 799 | while toprocess: |
|
797 | 800 | prec = toprocess.pop()[0] |
|
798 | 801 | if prec not in newermap: |
|
799 | 802 | successorssets(repo, prec, newermap) |
|
800 | 803 | newer = [n for n in newermap[prec] if n] |
|
801 | 804 | if len(newer) > 1: |
|
802 | 805 | divergent.add(ctx.rev()) |
|
803 | 806 | break |
|
804 | 807 | toprocess.update(obsstore.precursors.get(prec, ())) |
|
805 | 808 | return divergent |
|
806 | 809 | |
|
807 | 810 | |
|
808 | 811 | def createmarkers(repo, relations, flag=0, metadata=None): |
|
809 | 812 | """Add obsolete markers between changesets in a repo |
|
810 | 813 | |
|
811 | 814 | <relations> must be an iterable of (<old>, (<new>, ...)) tuple. |
|
812 | 815 | `old` and `news` are changectx. |
|
813 | 816 | |
|
814 | 817 | Trying to obsolete a public changeset will raise an exception. |
|
815 | 818 | |
|
816 | 819 | Current user and date are used except if specified otherwise in the |
|
817 | 820 | metadata attribute. |
|
818 | 821 | |
|
819 | 822 | This function operates within a transaction of its own, but does |
|
820 | 823 | not take any lock on the repo. |
|
821 | 824 | """ |
|
822 | 825 | # prepare metadata |
|
823 | 826 | if metadata is None: |
|
824 | 827 | metadata = {} |
|
825 | 828 | if 'date' not in metadata: |
|
826 | 829 | metadata['date'] = '%i %i' % util.makedate() |
|
827 | 830 | if 'user' not in metadata: |
|
828 | 831 | metadata['user'] = repo.ui.username() |
|
829 | 832 | tr = repo.transaction('add-obsolescence-marker') |
|
830 | 833 | try: |
|
831 | 834 | for prec, sucs in relations: |
|
832 | 835 | if not prec.mutable(): |
|
833 | 836 | raise util.Abort("cannot obsolete immutable changeset: %s" |
|
834 | 837 | % prec) |
|
835 | 838 | nprec = prec.node() |
|
836 | 839 | nsucs = tuple(s.node() for s in sucs) |
|
837 | 840 | if nprec in nsucs: |
|
838 | 841 | raise util.Abort("changeset %s cannot obsolete itself" % prec) |
|
839 | 842 | repo.obsstore.create(tr, nprec, nsucs, flag, metadata) |
|
840 | 843 | repo.filteredrevcache.clear() |
|
841 | 844 | tr.close() |
|
842 | 845 | finally: |
|
843 | 846 | tr.release() |
General Comments 0
You need to be logged in to leave comments.
Login now