On Jul 3, 2010, at 13:03, Richard Troth wrote:
> I'd recommend nothing less than 32M these days for a user v-machine.
> Memory is cheap. (And bandwidth is cheaper, to repeat an olde quote.)
>
> Seriously, while you might justifiably want to constrain some service
> machines, let the users have more. No telling what they might
> legitimately need to do. See if you are already allowed:
>
> def stor 16m
> ipl cms
>
> and then retry the ftp.
>
I asked for 100M. CP told me I was allowed 32. I took that and it
worked. Since I'm carrying 2G in my pocket, 32M sounds puny.
In fact, this is a service machine. I just logged on to update
the RLD.
Clearly the FTP designers didn't have pipethink. FTP may be older;
that's irrelevant. If so, it dates from a time of severe storage
constraints when it would have been wiser to stream the response to
LIST than to buffer it and format it subsequently. I suppose they
never considered the possibility of a directory with 2342 entries.
Again, as a sporadic lurker, I wish that announcements of library
updates would cite the URL:
ftp://vm.marist.edu/academ:pipeline/eweb./
... so I needn't chase it down each time.
Thanks,
gil
