Mercurial > public > mercurial-scm > hg
comparison mercurial/verify.py @ 31761:b044c339c06d
verify: document corner cases
It seems a good idea to list all kinds of "surprises" and expected behavior
to make the upcoming changes easier to understand.
Note: the comment added does not reflect the actual behavior of the current
code.
| author | Jun Wu <quark@fb.com> |
|---|---|
| date | Wed, 29 Mar 2017 14:45:01 -0700 |
| parents | 5249b6470de9 |
| children | dff03f68ef11 |
comparison
equal
deleted
inserted
replaced
| 31758:04ec317b8128 | 31761:b044c339c06d |
|---|---|
| 377 if havemf and n not in filenodes[f]: | 377 if havemf and n not in filenodes[f]: |
| 378 self.err(lr, _("%s not in manifests") % (short(n)), f) | 378 self.err(lr, _("%s not in manifests") % (short(n)), f) |
| 379 else: | 379 else: |
| 380 del filenodes[f][n] | 380 del filenodes[f][n] |
| 381 | 381 |
| 382 # verify contents | 382 # Verify contents. 4 cases to care about: |
| 383 # | |
| 384 # common: the most common case | |
| 385 # rename: with a rename | |
| 386 # meta: file content starts with b'\1\n', the metadata | |
| 387 # header defined in filelog.py, but without a rename | |
| 388 # ext: content stored externally | |
| 389 # | |
| 390 # More formally, their differences are shown below: | |
| 391 # | |
| 392 # | common | rename | meta | ext | |
| 393 # ------------------------------------------------------- | |
| 394 # flags() | 0 | 0 | 0 | not 0 | |
| 395 # renamed() | False | True | False | ? | |
| 396 # rawtext[0:2]=='\1\n'| False | True | True | ? | |
| 397 # | |
| 398 # "rawtext" means the raw text stored in revlog data, which | |
| 399 # could be retrieved by "revision(rev, raw=True)". "text" | |
| 400 # mentioned below is "revision(rev, raw=False)". | |
| 401 # | |
| 402 # There are 3 different lengths stored physically: | |
| 403 # 1. L1: rawsize, stored in revlog index | |
| 404 # 2. L2: len(rawtext), stored in revlog data | |
| 405 # 3. L3: len(text), stored in revlog data if flags==0, or | |
| 406 # possibly somewhere else if flags!=0 | |
| 407 # | |
| 408 # L1 should be equal to L2. L3 could be different from them. | |
| 409 # "text" may or may not affect commit hash depending on flag | |
| 410 # processors (see revlog.addflagprocessor). | |
| 411 # | |
| 412 # | common | rename | meta | ext | |
| 413 # ------------------------------------------------- | |
| 414 # rawsize() | L1 | L1 | L1 | L1 | |
| 415 # size() | L1 | L2-LM | L1(*) | L1 (?) | |
| 416 # len(rawtext) | L2 | L2 | L2 | L2 | |
| 417 # len(text) | L2 | L2 | L2 | L3 | |
| 418 # len(read()) | L2 | L2-LM | L2-LM | L3 (?) | |
| 419 # | |
| 420 # LM: length of metadata, depending on rawtext | |
| 421 # (*): not ideal, see comment in filelog.size | |
| 422 # (?): could be "- len(meta)" if the resolved content has | |
| 423 # rename metadata | |
| 424 # | |
| 425 # Checks needed to be done: | |
| 426 # 1. length check: L1 == L2, in all cases. | |
| 427 # 2. hash check: depending on flag processor, we may need to | |
| 428 # use either "text" (external), or "rawtext" (in revlog). | |
| 383 try: | 429 try: |
| 384 l = len(fl.read(n)) | 430 l = len(fl.read(n)) |
| 385 rp = fl.renamed(n) | 431 rp = fl.renamed(n) |
| 386 if l != fl.size(i): | 432 if l != fl.size(i): |
| 387 if len(fl.revision(n)) != fl.size(i): | 433 if len(fl.revision(n)) != fl.size(i): |