On Wed, Jul 28, 2004 at 10:14:27AM -0700, Marc Tooley wrote:
> On Saturday 24 July 2004 02:56, Martin Husemann wrote:
> >
> > In my experience this will only succeed in a minority of the cases
> > (aborted transfers happen to me if my network connection drops for
> > more than a few seconds or the remote server goes down), but of
> > course your milage may vary.
> 
> In the case of a transfer which is taking a long, long time, or is 
> user-aborted with a control-c, I find it tremendously annoying to see 
> pkgsrc complain about an incorrect MD5. I think it would be great if 
> pkgsrc noticed the file was short and tried to resume transfer.

I know that this is not what the resumption of transfer mods do, but
I think that some people may be overlooking the fact that pkgsrc has
had the FAILOVER_FETCH definition since 1998:

----------------------------
revision 1.154
date: 1998/09/01 13:15:29;  author: agc;  state: Exp;  lines: +63 -69
Amend the fetch:  target, so that the distfile's integrity is checked
with the additional checksum only if FAILOVER_FETCH is set (this does
not affect the checksum:  target at all).
        
Document FAILOVER_FETCH.
----------------------------

It's used all the time by the bulk build mechanism - it ensures that
the distfile is the correct one, and, if not, will fetch the distfile
from the next site until it gets the right one.

Regards,
Alistair
--
Alistair Crooks                    <agc@pkgsrc.org>