view mercurial/pure/base85.py @ 40445:634b45317459 stable

changegroup: restore default node ordering (issue6001)

Changeset db5501d9 changed the default node ordering from "storage" to
"linearize".

While the new API is more explicit and cleaner, the "linearize" order is
problematic on certain repositories like netbeans where it makes bundling
slower the more nodes we bundle.

Pushing and pulling 100 changesets was ~20% slower and pushing and pulling
1000 changesets was ~600% slower.

A very quick analysis of profile traces showed that the pull operation was
taking more time creating the delta.

Putting back the old default order seems to be the safe option. With more time
during the next cycle, we can understand better the impact of sorting with the
DAG order by default, the source of the regression and how to mitigate it.

/!\ We are still waiting for the full performance impact but with this patch,
bundling and pulling locally (not on the performance workstation) 1000
changesets on the netbeans repository is as fast as before the regression.

Differential Revision: https://phab.mercurial-scm.org/D5196

# base85.py: pure python base85 codec
#
# Copyright (C) 2009 Brendan Cully <[e-mail]>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.

from __future__ import absolute_import

import struct

from .. import pycompat

_b85chars = pycompat.bytestr("0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdef"
                             "ghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~")
_b85chars2 = [(a + b) for a in _b85chars for b in _b85chars]
_b85dec = {}

def _mkb85dec():
    for i, c in enumerate(_b85chars):
        _b85dec[c] = i

def b85encode(text, pad=False):
    """encode text in base85 format"""
    l = len(text)
    r = l % 4
    if r:
        text += '\0' * (4 - r)
    longs = len(text) >> 2
    words = struct.unpack('>%dL' % (longs), text)

    out = ''.join(_b85chars[(word // 52200625) % 85] +
                  _b85chars2[(word // 7225) % 7225] +
                  _b85chars2[word % 7225]
                  for word in words)

    if pad:
        return out

    # Trim padding
    olen = l % 4
    if olen:
        olen += 1
    olen += l // 4 * 5
    return out[:olen]

def b85decode(text):
    """decode base85-encoded text"""
    if not _b85dec:
        _mkb85dec()

    l = len(text)
    out = []
    for i in range(0, len(text), 5):
        chunk = text[i:i + 5]
        chunk = pycompat.bytestr(chunk)
        acc = 0
        for j, c in enumerate(chunk):
            try:
                acc = acc * 85 + _b85dec[c]
            except KeyError:
                raise ValueError('bad base85 character at position %d'
                                 % (i + j))
        if acc > 4294967295:
            raise ValueError('Base85 overflow in hunk starting at byte %d' % i)
        out.append(acc)

    # Pad final chunk if necessary
    cl = l % 5
    if cl:
        acc *= 85 ** (5 - cl)
        if cl > 1:
            acc += 0xffffff >> (cl - 2) * 8
        out[-1] = acc

    out = struct.pack('>%dL' % (len(out)), *out)
    if cl:
        out = out[:-(5 - cl)]

    return out