[RC5] Really useful feature request
root at jennifer-unix.dyn.ml.org
Thu Apr 2 22:07:20 EST 1998
On Wed, 1 Apr 1998, Peter Hugosson-Miller wrote:
> So here comes idea #1:
> For clients running in offline mode, give the option to specify a local
> fetch directory. Let the client check for the existence of a buff-in.rc5
> file in that directory each time the working buff-in.rc5 file is empty,
> and if it exists, move it to the client directory, and start processing
> those blocks. Otherwise, start generating random blocks as before.
> Simple, easy to implement, and _very very_ useful.
Hmm... if you can do this, then why not just point the client to this
directory in the first place?
> And, as a logical spin-off from that idea, here comes idea #2:
> When the buff-out.rc5 file is full, do a local flush to a buff-out.rc5
> file in the directory mentioned in idea #1. I guess it would then be
> called a local fetch/flush directory.
Now this is a useful concept, since buff-outs have a hard limit on size.
For clients that might acumulate over this limit of blocks, there should be
some clean way of keeping the blocks ad infinitum -- I would suguest
extending the format, but not knowing the current format, I can't offer
informed ideas on how to do so withought creating incompatablity. (Hint: GPL)
> Now, if the developers wanted to be a bit more ambitious, and building
> on idea #2, here comes idea #3:
> Make an option to specify an e-mail address to which completed blocks
> can be sent. After performing a local flush, that file could then be
> sent as an attachment to an e-mail message and sent off for flushing.
> The file could then be renamed to mailxxx.rc5 (where xxx is a serial
> number, increasing for each already existing mailxxx.rc5 file), and one
> would then delete it manually once it had arrived safely at its
> destination. Not that I'm lazy or anything ;-).
That seems like a good idea... but are there really all that many machines
that can email out where the client can't flush via the normal methods? (It
shouldn't be an incredable amount of work -- the STMP code is already there
for the log mailings. We would need to implement base64 encoding.)
-=- James Mastros
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest
More information about the rc5