Mercurial > hg > xemacs-beta
view netinstall/hash.cc @ 1298:1b4bc72f433e
[xemacs-hg @ 2003-02-14 12:05:06 by ben]
speedups to build process
autoload.el: Factor out common code in generate-{c-,}file-autoloads-1 into new
function generate-autoload-ish-1. \(I was originally going to use
this for custom as well but ended up thinking better of it.)
cus-dep.el: Cache the old computed values in custom-load.el and reuse them as
necessary, to speed up running cus-dep (which would take 25-30
seconds to do all files in lisp/*, lisp/*/* on my Pentium III
700). Use `message' not `princ' to get correct newline behavior.
Output messages showing each file we do actually process.
update-elc-2.el: Rewrite algorithm to be much faster -- cache calls to
directory-files and don't make needless calls to file-exists-p,
file-directory-p because they're way way slow.
Autoload early and only when update-elc has told us to.
update-elc.el: If no files need byte compilation, signal to update-elc-2 to do
any necessary autoload updating (using the file REBUILD_AUTOLOADS)
rather than doing it ourselves, which would be way slow. Ignore
updates to custom-load.el and auto-autoloads.el when checking to
see whether autoloads need updating. Optimize out many
unnecessary calls to file-exists-p to speed it up somewhat. (####
The remaining time is 50% or more in locate-file; this is
presumably because, even though it has a cache, it's still
statting each file to determine it's actually there. By calling
directory-files ourselves, building a tree, and then looking in
that tree, we could drastically shorten the time needed to do the
locate operation.)
author | ben |
---|---|
date | Fri, 14 Feb 2003 12:05:07 +0000 |
parents | 3078fd1074e8 |
children |
line wrap: on
line source
/* * Copyright (c) 2000, Red Hat, Inc. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation; either version 2 of the License, or * (at your option) any later version. * * A copy of the GNU General Public License can be found at * http://www.gnu.org/ * * Written by DJ Delorie <dj@cygnus.com> * */ /* Simple hash class for install.cc */ #include "win32.h" #include <stdlib.h> #include <string.h> #include "hash.h" class hash_internals { public: char **keys; int numkeys; int maxkeys; int prev_index; }; hash::hash () { h = new hash_internals; h->numkeys = 0; h->maxkeys = 10; h->keys = (char **) malloc (h->maxkeys * sizeof (char *)); h->prev_index = 0; } hash::~hash () { free (h->keys); free (h); } void hash::add (char *string) { int i; for (i=0; i<h->numkeys; i++) if (strcmp (h->keys[i], string) == 0) return; if (h->numkeys >= h->maxkeys) { h->maxkeys += 10; h->keys = (char **) realloc (h->keys, h->maxkeys * sizeof (char *)); } h->keys[h->numkeys] = _strdup (string); h->numkeys ++; } int hash::has (char *string) { int i; for (i=0; i<h->numkeys; i++) if (strcmp (h->keys[i], string) == 0) return 1; return 0; } char * hash::enumerate (char *prev) { if (prev == 0) h->prev_index = -1; h->prev_index ++; if (h->prev_index >= h->numkeys) return 0; return h->keys[h->prev_index]; } static int CDECL rev_len (const void *va, const void *vb) { char *a = *(char **)va; char *b = *(char **)vb; return strcmp (b, a); } void hash::reverse_sort () { qsort (h->keys, h->numkeys, sizeof (h->keys[0]), rev_len); }