[Spambayes] reporet on trying 1.0a6 on SuSE

[Meyer, Tony]

> I'm probably doing this all wrong, but I don't understand CVS or
> sourceforge, and someone might like some feedback on what I've found,
> particularly as 1.0a6 bills itself as getting the install 
> sorted (I think I saw that somewhere, but I can't find it now).

I'm not sure that that claim has ever been made ;)

> setup.py won't run.  Try it several times, dies with horrible looking
> errors.  On offchance, try using the .zip rather than the 
> .tar.gz.  When I do that, setup.py gets much further - is the .tar.gz
corrupt?

The tar.gz of 1.0a6 is missing the entire scripts directory, so isn't
all that usable :/.  There's now a 1.0a6.1 available, which has the
missing stuff.

> However, setup.py still won't complete.  It now moans that 
> "name 'False' is not defined".  Some poking around teh web leads me to

> believe that True and False were added in python 2.2.2.  If that is
the
> case, the README statement that you need 2.2 or later is not true -
you
> need 2.2.2 or later.

That's a bug in setup.py.  All the actual scripts should work fine in
2.2.  I've checked in a fix for the setup bug.

> sb_mboxtrain won't run, complains "bad interpreter: no such file or
> directory".  Aha - I know that one, and a quick peak in teh 
> scripts - yes indeed, all have a ^M at teh end of every line - they're
dos
> format. Is this because I worked from teh .zip rather than the
.tar.gz?

Yes.

> Next I run sb_mboxtrain.  That fails with teh same complaint about
> FALSE.  Reading teh traceback, it seemed the actual complaint 
> was related to site-packages/spambayes/Options.py, so I put True=1 and

> False=0 lines in tehre, just above import sys, os.

This shouldn't happen - sb_mboxtrain should define True and False before
it tries to import Options.py.  I don't suppose you still have a copy of
the traceback?

> Now sb_mboxtrain runs without crashing type error, but comes up with a
> usage error, apparently I must specify -d or -D.  There 
> doesn't seem to be anything about that in teh README.

There's a bug open about that.  That section of the README needs to be
updated.  I'm not really the right person to do so, though, because I
don't use it...

> So, I try sb_imapfilter.py.  That doesn't run either - 
> apparently "-b, -c or -t operands must be specified".  No mention of
> that in teh README file, however.

Fixed, thanks.

> I note that sb_mboxtrain -h says lower case -d is DBM and 
> upper case -D is pickle.  However, sb_imapfilter -h says lower case -d
> is pickled and uppercase -D is dbm.  Is this really so?  It might be
easier
> if there was consistency in the options.

Yes, unfortunately that is the case.  There was some discussion a while
back (on spambayes-dev) about fixing this up.  I imagine that it'll get
done for 1.1 (probably "-p" for pickle and "-d" for DBM), but the best
option is really to put all of this in a config file (.spambayesrc, for
example) rather than on the command line.

> sb_imapfilter.py -b runs up lynx (the machine it's running on 
> doesn't have X, so that's the best browser available), but lynx hangs
for 
> a long, long time at "HTTP request sent; waiting for response".  Not
sure
> where the hang is occurring, and it's never given up (or at least,
I've always
> interrupted it first).

Hmm.  If you run sb_server.py and go to http://localhost:8880, does it
hang then, too?  Using "Lynx Version 2.8.4rel.1 (17 Jul 2001)" works
fine for me.

> "cat sb_test_email | sb_filter.py -d
> ~/.hammiedb".  That dies with teh complaint about False again.  
> Not sure I fancy editing every script, and I'm also not certain that 
> that's actually working - I might be breaking something do that,
> so I postpone that approach for now.

Again, if you had the traceback, that would be helpful.  There's a lot
of code scattered about that's meant to provide False/True if it doesn't
exist, which should be working...

> Dunno if any of that is helpful to anyone.

Absolutely, thanks.

=Tony Meyer