Mercurial > public > mercurial-scm > hg-stable
view mercurial/base85.c @ 26117:4dc5b51f38fe
revlog: change generaldelta delta parent heuristic
The old generaldelta heuristic was "if p1 (or p2) was closer than the last full text,
use it, otherwise use prev". This was problematic when a repo contained multiple
branches that were very different. If commits to branch A were pushed, and the
last full text was branch B, it would generate a fulltext. Then if branch B was
pushed, it would generate another fulltext. The problem is that the last
fulltext (and delta'ing against `prev` in general) has no correlation with the
contents of the incoming revision, and therefore will always have degenerate
cases.
According to the blame, that algorithm was chosen to minimize the chain length.
Since there is already code that protects against that (the delta-vs-fulltext
code), and since it has been improved since the original generaldelta algorithm
went in (2011), I believe the chain length criteria will still be preserved.
The new algorithm always diffs against p1 (or p2 if it's closer), unless the
resulting delta will fail the delta-vs-fulltext check, in which case we delta
against prev.
Some before and after stats on manifest.d size.
internal large repo
old heuristic - 2.0 GB
new heuristic - 1.2 GB
mozilla-central
old heuristic - 242 MB
new heuristic - 261 MB
The regression in mozilla central is due to the new heuristic choosing p2r as
the delta when it's closer to the tip. Switching the algorithm to always prefer
p1r brings the size back down (242 MB). This is result of the way in which
mozilla does merges and pushes, and the result could easily swing the other
direction in other repos (depending on if they merge X into Y or Y into X), but
will never be as degenerate as before.
I future patch will address the regression by introducing an optional, even more
aggressive delta heuristic which will knock the mozilla manifest size down
dramatically.
author | Durham Goode <durham@fb.com> |
---|---|
date | Sun, 30 Aug 2015 13:58:11 -0700 |
parents | c1aefe57cf4e |
children | 4613a89bea42 |
line wrap: on
line source
/* base85 codec Copyright 2006 Brendan Cully <brendan@kublai.com> This software may be used and distributed according to the terms of the GNU General Public License, incorporated herein by reference. Largely based on git's implementation */ #define PY_SSIZE_T_CLEAN #include <Python.h> #include "util.h" static const char b85chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ" "abcdefghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~"; static char b85dec[256]; static void b85prep(void) { unsigned i; memset(b85dec, 0, sizeof(b85dec)); for (i = 0; i < sizeof(b85chars); i++) b85dec[(int)(b85chars[i])] = i + 1; } static PyObject * b85encode(PyObject *self, PyObject *args) { const unsigned char *text; PyObject *out; char *dst; Py_ssize_t len, olen, i; unsigned int acc, val, ch; int pad = 0; if (!PyArg_ParseTuple(args, "s#|i", &text, &len, &pad)) return NULL; if (pad) olen = ((len + 3) / 4 * 5) - 3; else { olen = len % 4; if (olen) olen++; olen += len / 4 * 5; } if (!(out = PyBytes_FromStringAndSize(NULL, olen + 3))) return NULL; dst = PyBytes_AsString(out); while (len) { acc = 0; for (i = 24; i >= 0; i -= 8) { ch = *text++; acc |= ch << i; if (--len == 0) break; } for (i = 4; i >= 0; i--) { val = acc % 85; acc /= 85; dst[i] = b85chars[val]; } dst += 5; } if (!pad) _PyBytes_Resize(&out, olen); return out; } static PyObject * b85decode(PyObject *self, PyObject *args) { PyObject *out; const char *text; char *dst; Py_ssize_t len, i, j, olen, cap; int c; unsigned int acc; if (!PyArg_ParseTuple(args, "s#", &text, &len)) return NULL; olen = len / 5 * 4; i = len % 5; if (i) olen += i - 1; if (!(out = PyBytes_FromStringAndSize(NULL, olen))) return NULL; dst = PyBytes_AsString(out); i = 0; while (i < len) { acc = 0; cap = len - i - 1; if (cap > 4) cap = 4; for (j = 0; j < cap; i++, j++) { c = b85dec[(int)*text++] - 1; if (c < 0) return PyErr_Format( PyExc_ValueError, "bad base85 character at position %d", (int)i); acc = acc * 85 + c; } if (i++ < len) { c = b85dec[(int)*text++] - 1; if (c < 0) return PyErr_Format( PyExc_ValueError, "bad base85 character at position %d", (int)i); /* overflow detection: 0xffffffff == "|NsC0", * "|NsC" == 0x03030303 */ if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c) return PyErr_Format( PyExc_ValueError, "bad base85 sequence at position %d", (int)i); acc += c; } cap = olen < 4 ? olen : 4; olen -= cap; for (j = 0; j < 4 - cap; j++) acc *= 85; if (cap && cap < 4) acc += 0xffffff >> (cap - 1) * 8; for (j = 0; j < cap; j++) { acc = (acc << 8) | (acc >> 24); *dst++ = acc; } } return out; } static char base85_doc[] = "Base85 Data Encoding"; static PyMethodDef methods[] = { {"b85encode", b85encode, METH_VARARGS, "Encode text in base85.\n\n" "If the second parameter is true, pad the result to a multiple of " "five characters.\n"}, {"b85decode", b85decode, METH_VARARGS, "Decode base85 text.\n"}, {NULL, NULL} }; #ifdef IS_PY3K static struct PyModuleDef base85_module = { PyModuleDef_HEAD_INIT, "base85", base85_doc, -1, methods }; PyMODINIT_FUNC PyInit_base85(void) { b85prep(); return PyModule_Create(&base85_module); } #else PyMODINIT_FUNC initbase85(void) { Py_InitModule3("base85", methods, base85_doc); b85prep(); } #endif