Tue 2003-02-04 10:08, Gerrit Kuehn (2:2411/12) wrote to andrew clarke:

 ac> This morning I got the Win32 version of FastLst 2.01 working to 
 ac> generate V7 indexes, so I'll probably have fidouser.lst & V7 reading 
 ac> code working in the next week or so, unless I get distracted.

GK> If you're interested you can check out the latest fastlst version
GK> from the Husky CVS. I did quite some changes during the last months
GK> and I would welcome some testers. :)

After looking at this for a while longer I've pretty much concluded that I
can't display multiple nodelist entries from a V7 nodelist without knowing
the V7 format intimately, which is a bit more than I'm prepared to do.  So
at best, if I search for "Gerrit Kuehn" it will return either
2:2411/12 or 2:246/4020, but not give the option of choosing between them,
which is presumably not what people want.  I put the fault more at V7's use
of a binary tree layout than anything else.  As far as I can work out,
b-trees are only really suitable for finding a single matching node, not
multiple nodes.  Which is presumably why Paul Edwards left the V7 code how
it is (which incidentally, doesn't appear to be endian-safe).

Either way, the V7 format is not my cup of tea, but maybe with your
experience at FastLst, you might have a better idea of how to work with
that format and have a stab at the getting timEd's V7 lookup code to work
the way it should.

Alternatively, maybe we can convince everyone to use a fidouser.lst, for
which there is working code (which is quite bizarre, if you've ever looked
at it), or come up with a simple nodelist compiler for timEd that just
stores sysop names along with an offset to the raw nodelist.  Or just parse
the raw nodelist each time the user does a lookup?

-- mail{at}ozzmosis.com

--- timEd/FreeBSD 1.11.a3