Re: More md5 checksums, please

[Drew Einhorn <drew.einhorn@starband.net>]

At 02:47 PM 6/14/2002, you wrote:

>Drew Einhorn wrote:
> > It randomly corrupts about 1 or 2 bytes per 100MB.
>You've realized the *real* problem? Random corruption at that
>point will always be a thorn on your side. That accelerator
>isn't. Either change that, or live with the consequences.

The accelerator is broken, it causes the corruption.  It's part of
starband.  It's not something I have control over.
You don't want to hear the details.

The alternative where I live a substandard dialup.
Closer to 5K than to 56K.

>Besides, MD5 won't stop your "accelerator" from wasting your
>time and network resources, right? The real solution lies
>elsewhere, and unless you get out of that transparent
>proxy, you'll always have a problem.

That's like telling the folks in Australia to get out from
behind their transparent proxies.  It's not easy and
can be extremely expensive.

The md5s will tell me whether the download was successful.
And whether I need to try again.  For small to midsize files
it's usually successful.  But I need to be able to know whether
it was successful, not just guess.

For larger files, it's hard to guess.

Right now I download them all at least twice into
two separate directories, and do a diff -r when I've got them all.

md5's would make it easier.  It would take, what, a half dozen
or so lines of shell script for someone with an account with
the necessary privilege on ftp.openbsd.org.  If somebody can
do it, then: Great!  Thank you, thank you, thank you!!

If not.  Oh well!

For really big iso's, I don't even try, I hop in the car and take a 45
minute ride to use a friend's cable modem.

Getting ready to pack up my laptop and go to town.