Re: Bug#533934: pperl: FTBFS: tests failed directory
On Sun, Jun 12, 2011 at 11:20:07PM -0400, Daniel Kahn Gillmor wrote:
> On 06/12/2011 05:36 PM, Dominic Hargreaves wrote:
> > On Fri, May 13, 2011 at 07:38:41PM +0200, Dominique Dumont wrote:
> >> Le vendredi 13 mai 2011 00:31:52, Dominic Hargreaves a écrit :
> >>> This bug is caused by the fact that pperl names its socket files based
> >>> on the full path of the perl program being run under pperl, and there
> >>> is a fairly short limit on the full path of the socket (sun_path in
> >>> struct sockaddr_un from sys/un.h, 108 characters). The socket name is
> >>> being silently truncated on creation.
> >> How about hashing (à la MD5) the full path and use the result as the socket
> >> name ? See md5(3ssl)
> >> Although this would add a dependency on openssl...
> > I agree that adding a dependency on openssl would not be ideal. This
> > doesn't have to be cryptographically secure, so would crypt(3) or
> > similar suffice?
> If the concerns around OpenSSL are licensing issues, there are many
> better-licensed message digest toolkits, including:
> * nettle
> * gcrypt
> * Digest::MD5 (or other modules within the Digest:: namespace).
This would be implemented in the C part of pperl, so this one is out.
In general I would be looking to avoid any extra heavyweight dependency
if possible, but thank you for pointing out that there are viable
> > Donations of code welcome at this point...
> Does this really need any extra code?
Yes, pperl "needs" to be fixed. I wasn't suggesting the creation of a new
digest toolkit :)
I added the quotation marks because I'm starting to doubt that it is
worth spending time on. I use pperl a bit, but probably wouldn't miss
it hugely, and upstream appears to be dead.
Dominic Hargreaves | http://www.larted.org.uk/~dom/
PGP key 5178E2A5 from the.earth.li (keyserver,web,email)