[rc5] Mac Client bug

Tristan Horn tristan at camel.ethereal.net
Sun Jun 22 22:20:35 EDT 1997


On Sun, 22 Jun 1997, Colin L. Hildinger wrote:
> No one ever answered my questions about the Mac bug and whether it
> could be fixed in the server's database.  Maybe someone's on #rc5 that
> can answer it now...

This is not possible at the moment, as far as I understand.  It should be
made possible sometime though...

[snip]
> but it needed more reliability testing before release.
[snip]

I spent about half an hour testing the latest release... it's not a small
task, considering these factors:

multitasking or no multitasking (-w 1+ or -w 0)
status bar or no status bar (-s or no -s)
-m or -p 3056
68k or ppc (or fat, I guess)

4**2 (16) tests.  Admittedly, I only did -p 3056 on the 68k client, since
it's so slow... (6Kk/s to 8Kk/s on my machines)

Normally I only use -p 3056 for checking new releases.  I wouldn't blame
TimC if that's all he does, considering all the different options in his
client.

I think having the proxy servers hand out blocks with solutions is a
perfect secondary test.  If it was done just 1% of the time, it would
(theoretically) find all bogus clients which have completed 100 or more
blocks.  And if someone was completing one block every 5 seconds, they'd
get found in only 8m20s.

Now, off to go camping... if my iostream library's eof() wasn't broken, I
might have had Andrew Meggs' v2 client out by now (which looks very nice 
btw, and yes, it's MP-savvy).

I'll be back on Wednesday...

Tris

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list