Mercurial > hg
view mercurial/cext/parsers.pyi @ 51424:3cee8706f53b
phases: directly update the phase sets in advanceboundary
This is similar to what we do in retractboundary. There is no need to invalidate
the cache if we have everything at hand to update it.
author | Pierre-Yves David <pierre-yves.david@octobus.net> |
---|---|
date | Fri, 23 Feb 2024 06:37:25 +0100 |
parents | f8bf1a8e9181 |
children | e58f02e2f6a9 |
line wrap: on
line source
from typing import ( Callable, Dict, Iterator, List, Optional, Set, Tuple, Union, ) version: int versionerrortext: str class DirstateItem: __doc__: str def __len__(self) -> int: ... def __getitem__(self, key: int) -> Union[bytes, int]: ... # From dirs.c class dirs: __doc__: str def __init__(self, source, skipchar: bytes): ... def __iter__(self) -> Iterator[bytes]: ... def addpath(self, path: bytes) -> None: ... def delpath(self, path: bytes) -> None: ... # From manifest.c class lazymanifest: def __init__(self, nodelen: int, data: bytes): ... def __iter__(self) -> Iterator[bytes]: ... def __len__(self) -> int: ... def __getitem__(self, item: bytes) -> Optional[Tuple[bytes, bytes]]: ... def __setitem__(self, key: bytes, value: Tuple[bytes, bytes]) -> None: ... def __delitem__(self, key: bytes) -> None: ... def iterkeys(self) -> Iterator[bytes]: ... def iterentries(self) -> Iterator[Tuple[bytes, bytes, bytes]]: ... def copy(self) -> lazymanifest: ... def filtercopy(self, matchfn: Callable[[bytes], bool]) -> lazymanifest: ... def diff(self, other: lazymanifest, clean: Optional[bool]) -> Dict[bytes, Tuple[bytes, Tuple]]: ... def text(self) -> bytes: ... # From revlog.c class index: __doc__: str nodemap: Dict[bytes, int] def ancestors(self, *args: int) -> Iterator[int]: ... def commonancestorsheads(self, *args: int) -> List[int]: ... def clearcaches(self) -> None: ... def get(self, value: bytes) -> Optional[int]: ... def get_rev(self, value: bytes) -> Optional[int]: ... def has_node(self, value: Union[int, bytes]) -> bool: ... def rev(self, node: bytes) -> int: ... def computephasesmapsets(self, root: Dict[int, Set[int]]) -> Tuple[int, Dict[int, Set[bytes]]]: ... def reachableroots2(self, minroot: int, heads: List[int], roots: List[int], includepath: bool) -> List[int]: ... def headrevs(self, filteredrevs: Optional[List[int]]) -> List[int]: ... def headrevsfiltered(self, filteredrevs: Optional[List[int]]) -> List[int]: ... def issnapshot(self, value: int) -> bool: ... def findsnapshots(self, cache: Dict[int, List[int]], start_rev: int) -> None: ... def deltachain(self, rev: int, stop: int, generaldelta: bool) -> Tuple[List[int], bool]: ... def slicechunktodensity(self, revs: List[int], targetdensity: float, mingapsize: int) -> List[List[int]]: ... def append(self, value: Tuple[int, int, int, int, int, int, int, bytes]) -> None: ... def partialmatch(self, node: bytes) -> bytes: ... def shortest(self, value: bytes) -> int: ... def stats(self) -> Dict[bytes, int]: ... class nodetree: __doc__: str def insert(self, rev: int) -> None: ... def shortest(self, node: bytes) -> int: ... # The IndexObject type here is defined in C, and there's no type for a buffer # return, as of py3.11. https://github.com/python/typing/issues/593 def parse_index2(data: object, inline: object, format: int = ...) -> Tuple[object, Optional[Tuple[int, object]]]: ...