> Just a thought... > > THE APPLICATION: I'm trying to use nntpcache to improve reading news over a > dial up line (intermittant connection). It works very well, a substantial > improvement to running a news reader locally with nntp to the server or > using suck + inn. I still run suck on a handful of favourite newsgroups to > pre-load the cache and speed up reading. > > THE PROBLEM: I have diald disconnect the ppp connection after a minute of > inactivity, with nntpcache this gives annoying delays as diald > reconnects. I have increased the uptime to 2 minutes but it is still > quite clumsy. > > WHAT WOULD HELP: If nntp could assume that all data it has cached > locally is "good" until some (configurable) time out is reached. There's two issues I can think of here. One is for 'group' commands to return data based on the cached active file rather than asking the server(s). This is planned anyway, mostly to reduce time spent in some of the nasty conversational stuff that NNTP seems to require sometimes which really slows things down if there's a bit of round trip time between nntpcache and the server(s). The other is to lose the idea of the 'default' server, which should be possible. It's really a hangover from the days before nntpcache was multiserver. Both of these are currently in process, as part of a major reorganisation of the code I'm doing. I've been normalising data structures and doing better encapsulation of things that need it and lots of other long words I learnt in first year comp sci ;). > HOW I WOULD USE IT: I would set the time out to an hour or so. Then a > crond job would "read" my favourite groups every so often using suck. > When I sit down to read news - no waiting to dial out, nice fast news > reading. If I decide to skim groups outside my usual focus I get the > delays while nntpcache pulls down what it needs - but its still way > faster than setting NNTPSERVER to my providers address and reading news. > > CLOSING: Thanks for reading this far, sorry if this has already been > hashed out here before. Also sorry if this is really just a verbose > restatement of the todo bullet "cache GROUP command for small time period > (say 5-60 minutes)". But most of all thanks for nntpcache. You're welcome. Thanks for helping test the thing. > > Matt > -- -- Luke Bowker, puke@suburbia.net, puke@deakin.edu.au Suburbia Public Access Network Site Sysadmin "Don't try to understand. Knowing you, I'm probably wrong" - D. Mustaine
