0ct0pu5/wesnoth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

I made some performance related changes to wml.pm and wml_net.pm.

Browse source 
The change to wml.pm cut my memory use in half, but arrays or chars still
suck compared to strings (on the order of x50) when space is important.
This commit is contained in:
Bruno Wolff III 2005-10-18 06:55:14 +00:00
parent b02a37797e
commit 47d564d73a
3 changed files with 20 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
changelog
View file

@ -82,6 +82,13 @@ SVN trunk:
the cached campaign, to give perl a chance to release some memory.
* Fixed a bug in webtgz.pl where it wasn't handling escaped data in
downloaded files.
* Updated wml.pm and wml_net.pm to gain some efficiency. wml.pm was changed
not to use split, but instead use substr to convert a string to an array.
This saved about half the memory usage when using large strings. Arrays
are still a pretty bad way to store strings when space is a concern.
wml_net.pm was changed to insert into a preallocated string instead of
appending in order to avoid lots of reallocations. In theory this should
save memory and time, but I didn't benchmark it.
* fix untranslated unit create dialog (#4424)
* random map generator now uses island_size (#4458)
* documentation switch doxygen templates from CVS to SVN and from Savannah to Gna!

8
utils/wml.pm
View file

@ -9,7 +9,7 @@ $wml::max_schema_item_length = 20;
sub read_literal_word
{
my $word = "";
my $word = '';
my ($chars) = @_;
while(@$chars) {
my $char = shift @$chars;
@ -63,7 +63,11 @@ sub read_binary
my @stack = ($doc);
my $cur = $doc;
my @chars = split //, $input;
my $len = length $input;
my @chars;
for (my $i = 0; $i < $len; $i++) {
push @chars, substr($input, $i, 1);
}
while(@chars) {
my $char = shift @chars;

12
utils/wml_net.pm
View file

@ -42,15 +42,17 @@ sub read_packet
my $len = unpack('N',$buf);
my $res = '';
while($len != length $res) {
my $res = "\0" * $len;
my $count = 0;
while($len > $count) {
$buf = '';
my $bytes = $len - length $res;
my $bytes = $len - $count;
read $sock, $buf, $bytes or die "Error reading socket: $!";
$res .= $buf;
substr($res, $count, length $buf) = $buf;
$count += length $buf;
}
$res = substr($res,0,length($res)-1);
$res = substr($res,0,$len-1);
return &wml::read_binary($wml_net::incoming_schemas{$sock},$res);
}