[hashcash] Re: bugs/problems, complaint
- From: Adam Back <adam@xxxxxxxxxxxxxxx>
- To: hashcash@xxxxxxxxxxxxx
- Date: Fri, 10 Nov 2006 14:37:47 -0500
Sorry about the sluggish responses to the problems. I have been
rushed last week with some non-computer related stuff that has urgent
deadlines and stuff like that.
Heres a brain dump of the history of this problem, which is nagging,
and hasnt gone away fully despite several attempts.
There were problems reported in 1.21, and I spent quite a bit of
effort trying to debug and understand how the issue was triggered.
1.22 was my attempt to fix.
Part of the problem is the code that causes it as far as I can tell is
some of the specific cores which are part assembly and speed hacked.
The cores were written by Jonathan Morton and I was having problems
figuring out what was going wrong. I restructured things a bit and
hoped to have fixed a generic problem that could affect most of the
cores (which have similar code) in the 1.22 release. However problems
remain.
The other problem is to debug, you have to wait for a 2^32 bit stamp
as it is the wrap that causes problems. I do have some test code to
use smaller word sizes, but it may not fully exercise the problem
because the actual words obviously really are 32 bit, just I have some
& (2^8)-1 or whatever.
I am unclear whether I have introduced a new problem, or this is a 2nd
previously not-separately diagnosed problem from the previous version.
If Jonathan is reading, it'd be great if you (or anyone else) could
take a look at it. Maybe with the shorter word sizes if it repos with
that compilation option.
I might have a bit of time this weekend between this non-computer task
I mentioned, but this could likely be a thorny debugging problem
requiring more time than I will have at least for 3-4 days.
One thing that can help is to turn on the trace of biggest stamp seen
so far, until I fixed one of the bugs, it appeared the biggest stamp
seen was getting reset sometimes. (A variation of the same problem
potentially).
Adam
On Fri, Nov 10, 2006 at 01:59:41PM -0500, Hubert Chan wrote:
> On Fri, 10 Nov 2006 13:30:23 +0100, bas <beware@xxxxxxxxx> said:
>
> > i sent a mail to the list 2 weeks ago about some problems i had with
> > hashcash, and i got no reaction of any kind. i think if hashcash is
> > actually going to be used, it should be as stable and "just works" as
> > possible. what's wrong?
>
> I've had similar problems to what you described, on linux (and reported
> on the list several months ago). For me, version 1.21 "just works", but
> 1.22 has problems.
Other related posts:
