I don't mean to sound negative, but at 5 employees there couldn't be too many issues to track, really? If you have more than two issues open at once, you need an issue tracker. Otherwise you'll drop things on the floor.
Cue hordes of arrests under the DMCA for circumventing a copy control measure, by holding a camera on its side. Just a minute while I fit a spirit level to my camera...
With all these suggestions of sanctions coming out of the woodwork, it's worth pointing out what powers the EU Competition Commission actually has.It can fine the company up to 10% of turnover. That's total group turnover, not just the business it does in the EU.
My understanding of SpamAssassin's "genetic algorithm" is that it's performed by SA Central before making the release, and therefore based on SA's own spam corpus. If their corpus differs from my own pattern of spam, doesn't that imply that their scoring will be less than optimal for my circumstances?
Naturally, this is rampant speculation based on my limited understanding of SA score generation - if the SA heuristics do train themselves on my email patterns in some way I've completely missed, then I take it all back.
Not quite the same thing, I know, but here's what I do:
I use SpamAssassin as a filter, using a trigger score of 4. If (and only if) SA flags it, I bounce it back saying:
\begin{quote} This message was flagged as spam so I've bounced it back. Sorry.
If I'm wrong (even Perl scripts are wrong sometimes) and you're a Real Person, look at
http://www.anchovy.durge.org/spam/
which will explain how to get your message through.
Iain's Perl script. \end{quote}
(Politeness costs nothing. Especially if it's a Perl script doing the talking;)
The method, FWIW, is to include a particular string in your subject line. It changes weekly.
Why do I add the burden of going to the web site? It's supposed to make it slightly harder for the Evil Spammers to work around it. I'm still considering whether it's a good idea.
Reading this leads me to wonder if SpamAssassin has its scoring inside out. Let me explain.
There's been a fair bit of talk on how to choose the tokens to feed to the Bayes engine. Paul Graham discusses using extended tokens like "Subject*free", and I can't help thinking that SpamAssassin has a ready-made body of tokens; RAZOR2_CHECK, FORGED_MUA_OUTLOOK and so on. So, why not turn SpamAssassin inside out, and instead of letting the Bayes engine report to the overall scoring, let each test become a token for the Bayes engine?
Perhaps it's the engineer in me, but I still feel slightly unhappy about SpamAssassin's scoring theory. The score doesn't fit an obvious model. With Bayes you can sum it up quickly with probability theory, but SpamAssassin's score is a complex heuristic.
Don't get me wrong - I use SA here and it's shit-hot. I just can't quite see it Taking Over The World[tm] in its current form.
I have seen any similar suggestions, but then I've only read brief reports of the conference. I suppose I ought to do a decent test on the scheme before waffling about it, but this is/. after all...
Slashdot Mirror
Focus-follows-mouse will have to do, until someone finally implements focus-follows-eyeball and stops me sending the root password to IRC *again*.
Iain.
Mine runs on a G4 Mac Mini, and it's perfectly snappy.
I realise that's a sample size of one, but I'm sure that's good enough for anyone.
Iain.
Obligatory joelonsoftware link.
Iain.
Hire a neighbour's kid to do it for you.
OK, I agree RT takes some investment to get going. As an RT administrator myself, I think it's well worth it. And once it's there, it runs itself.
But if that's really too hard for you, use mantis.
Or M-x todo-mode
Iain.In short, they're not. That's the BBC's attempt at explaining why the theoretical peak isn't practically achievable. See http://en.wikipedia.org/wiki/Teraflop#FLOPS_as_a_m easure_of_performance
Iain.
No.
They're not.
Even slightly.
This action is taking place in the UK. This means that US law doesn't actually apply. Shocking, I know, but true.
Iain.
You didn't check carefully enough, then. Format shifting is not explicitly allowed, and is therefore unlawful.
Imagine how much higher the bar would be if the USPTO were funded in relation to the number of patents they reject, instead by the number they grant.
Cue hordes of arrests under the DMCA for circumventing a copy control measure, by holding a camera on its side.
Just a minute while I fit a spirit level to my camera...
With all these suggestions of sanctions coming out of the woodwork, it's worth pointing out what powers the EU Competition Commission actually has.It can fine the company up to 10% of turnover. That's total group turnover, not just the business it does in the EU.
My understanding of SpamAssassin's "genetic algorithm" is that it's performed by SA Central before making the release, and therefore based on SA's own spam corpus. If their corpus differs from my own pattern of spam, doesn't that imply that their scoring will be less than optimal for my circumstances?
Naturally, this is rampant speculation based on my limited understanding of SA score generation - if the SA heuristics do train themselves on my email patterns in some way I've completely missed, then I take it all back.
Not quite the same thing, I know, but here's what I do:
;)
I use SpamAssassin as a filter, using a trigger score of 4. If (and only if) SA flags it, I bounce it back saying:
\begin{quote}
This message was flagged as spam so I've bounced it back. Sorry.
If I'm wrong (even Perl scripts are wrong sometimes) and you're a
Real Person, look at
http://www.anchovy.durge.org/spam/
which will explain how to get your message through.
Iain's Perl script.
\end{quote}
(Politeness costs nothing. Especially if it's a Perl script doing the talking
The method, FWIW, is to include a particular string in your subject line. It changes weekly.
Why do I add the burden of going to the web site? It's supposed to make it slightly harder for the Evil Spammers to work around it. I'm still considering whether it's a good idea.
Reading this leads me to wonder if SpamAssassin has its scoring inside out. Let me explain.
/. after all...
There's been a fair bit of talk on how to choose the tokens to feed to the Bayes engine. Paul Graham discusses using extended tokens like "Subject*free", and I can't help thinking that SpamAssassin has a ready-made body of tokens; RAZOR2_CHECK, FORGED_MUA_OUTLOOK and so on. So, why not turn SpamAssassin inside out, and instead of letting the Bayes engine report to the overall scoring, let each test become a token for the Bayes engine?
Perhaps it's the engineer in me, but I still feel slightly unhappy about SpamAssassin's scoring theory. The score doesn't fit an obvious model. With Bayes you can sum it up quickly with probability theory, but SpamAssassin's score is a complex heuristic.
Don't get me wrong - I use SA here and it's shit-hot. I just can't quite see it Taking Over The World[tm] in its current form.
I have seen any similar suggestions, but then I've only read brief reports of the conference. I suppose I ought to do a decent test on the scheme before waffling about it, but this is