Mercurial > public > mercurial-scm > hg-stable
comparison mercurial/revlog.py @ 34898:1bde8e8e5de0
sparse-read: ignore trailing empty revs in each read chunk
An empty entry in the revlog may happen for two reasons:
- when the file is empty, and the revlog stores a snapshot;
- when there is a merge and both parents were identical.
`hg debugindex -m | awk '$3=="0"{print}' | wc -l` gives 1917 of such entries
in my clone of pypy, and 113 on my clone of mercurial.
These empty revision may be located at the end of a sparse chain, and in some
special cases may lead to read relatively large amounts of data for nothing.
| author | Paul Morelle <paul.morelle@octobus.net> |
|---|---|
| date | Wed, 18 Oct 2017 15:28:19 +0200 |
| parents | 8c9b08a0c48c |
| children | 6226668a7169 |
comparison
equal
deleted
inserted
replaced
| 34897:2e350d2a0eca | 34898:1bde8e8e5de0 |
|---|---|
| 160 s = hashlib.sha1(a) | 160 s = hashlib.sha1(a) |
| 161 s.update(b) | 161 s.update(b) |
| 162 s.update(text) | 162 s.update(text) |
| 163 return s.digest() | 163 return s.digest() |
| 164 | 164 |
| 165 def _trimchunk(revlog, revs, startidx, endidx=None): | |
| 166 """returns revs[startidx:endidx] without empty trailing revs | |
| 167 """ | |
| 168 length = revlog.length | |
| 169 | |
| 170 if endidx is None: | |
| 171 endidx = len(revs) | |
| 172 | |
| 173 # Trim empty revs at the end, but never the very first revision of a chain | |
| 174 while endidx > 1 and endidx > startidx and length(revs[endidx - 1]) == 0: | |
| 175 endidx -= 1 | |
| 176 | |
| 177 return revs[startidx:endidx] | |
| 178 | |
| 165 def _slicechunk(revlog, revs): | 179 def _slicechunk(revlog, revs): |
| 166 """slice revs to reduce the amount of unrelated data to be read from disk. | 180 """slice revs to reduce the amount of unrelated data to be read from disk. |
| 167 | 181 |
| 168 ``revs`` is sliced into groups that should be read in one time. | 182 ``revs`` is sliced into groups that should be read in one time. |
| 169 Assume that revs are sorted. | 183 Assume that revs are sorted. |
| 192 prevend = None | 206 prevend = None |
| 193 for i, rev in enumerate(revs): | 207 for i, rev in enumerate(revs): |
| 194 revstart = start(rev) | 208 revstart = start(rev) |
| 195 revlen = length(rev) | 209 revlen = length(rev) |
| 196 | 210 |
| 211 # Skip empty revisions to form larger holes | |
| 212 if revlen == 0: | |
| 213 continue | |
| 214 | |
| 197 if prevend is not None: | 215 if prevend is not None: |
| 198 gapsize = revstart - prevend | 216 gapsize = revstart - prevend |
| 199 # only consider holes that are large enough | 217 # only consider holes that are large enough |
| 200 if gapsize > revlog._srmingapsize: | 218 if gapsize > revlog._srmingapsize: |
| 201 heapq.heappush(gapsheap, (-gapsize, i)) | 219 heapq.heappush(gapsheap, (-gapsize, i)) |
| 220 | 238 |
| 221 # Cut the revs at collected indices | 239 # Cut the revs at collected indices |
| 222 previdx = 0 | 240 previdx = 0 |
| 223 while indicesheap: | 241 while indicesheap: |
| 224 idx = heapq.heappop(indicesheap) | 242 idx = heapq.heappop(indicesheap) |
| 225 yield revs[previdx:idx] | 243 |
| 244 chunk = _trimchunk(revlog, revs, previdx, idx) | |
| 245 if chunk: | |
| 246 yield chunk | |
| 247 | |
| 226 previdx = idx | 248 previdx = idx |
| 227 yield revs[previdx:] | 249 |
| 250 chunk = _trimchunk(revlog, revs, previdx) | |
| 251 if chunk: | |
| 252 yield chunk | |
| 228 | 253 |
| 229 # index v0: | 254 # index v0: |
| 230 # 4 bytes: offset | 255 # 4 bytes: offset |
| 231 # 4 bytes: compressed length | 256 # 4 bytes: compressed length |
| 232 # 4 bytes: base rev | 257 # 4 bytes: base rev |