Jeroen Asselman wrote:
> >Anyway, maybe it's not that good to reuse the download facility
> >to download the tigertree at all. It might be better to use
> >the lower level http stuff e.g., like gwebcache.c.

> I'd like to comment on this one. If you are going to use the lower http 
> stuff, you can't make use of queueing. There are clients out there that 
> can / will queue a request for a tigertree.

How big is a tigertree actually? If it's not too big it's probably
not necessary to persist partial tigertrees or queue them beyond a
single session. Queuing itself should not be that horribly difficult
then i.e., just use a callback queue.

Downloading the tigertree like a normal file is surely a possibility
(if there wasn't the filesize issue) but it's somewhat ugly considering
that it'll show up in GUI and appear as a file in your download directory
etc. So you'll have to make the download logic much more complex because
it'll have to treat TTs specially. This *might* be more difficult than
just rolling your own.

-- 
Christian

Attachment: pgpnykIBDoEex.pgp
Description: PGP signature

Reply via email to