reformatting controlchan

Fabien Tassin fta at sofaraway.org
Wed Jan 24 17:25:58 UTC 2001


According to Marco d'Itri:
> 
> On Jan 24, Russ Allbery <rra at stanford.edu> wrote:
> 
>  >I almost did that myself the other day; I don't see any reason at all to
>  >object.  Please, feel free.  We should probably use roughly the same
> I finished the grunt work of reformatting the source code and now I'm
> checking I did not broke anything and doing general optimization.
> I also plan to rewrite docheckgroups in perl.
> I wrote a "gpgverify" program and I'm going to post it soon.
> 
> Can I remove from controlchan support for these things?
> 
> - perl < 5.00403 (5.004 was required anyway)
> - not using syslog

Then it should shout loudly if the syslog module is not available.
I still have an old Sun that is using this code (Sys::Syslog has never been
installed on that box and controlchan is falling back to errlog).
Note that logmsg() with $use_syslog == 0 produces strange things under
control msg storms. Each new entry repeat the previous ones, producing
things like this :

line 1
line 2
line 3  <= beginning of the storm
line 3
line 4
line 3
line 4
line 5
line 3
...
line 6
etc.

no clue why :(

> (BTW, does anybody know a way to use soft references like $a='string'; &$a
> without using "no strict"?)

by using hard links, for example with a scalar or a hash..

====
#!/usr/bin/perl -w

use strict;

my $a = 'foo';
my $b = \&$a;
my $c = { 'foo' => sub { print shift, "\n"; } };

$b->("ok");
$$c{$a}->("ok");

sub foo {
  print shift, "\n";
}
====

soft links are a bad idea because they aren't reference counted. They can't
access lexical variables either and they lack of module information.

-- 
Fabien Tassin -+- fta at sofaraway.org



More information about the inn-workers mailing list