🚀 go-pugleaf

On Saturday, in article <830566222snz@wbriscoe.demon.co.uk>
     walter@wbriscoe.demon.co.uk "walter briscoe" wrote:

wb> Although the current code does not depend on this, I reckon that
wb> adjacent index entries should refer to adjacent articles with neither
wb> gaps between nor overlaps of articles.

I'm [gulp, touch wood] sure this is true.  I've not
referred back to Prune source, but I think that assumes so.

wb> As the sizes of articles are determinable from the index, copy
wb> operations can be done in units of a large buffer rather than lines.

I also think Prune does that, in 16K chunks.  But I've yet
to do any performance tuning to see if this resulted in a
worthwhile speed increase.

wb> I was thinking of direct access to a read file buffer passed in by
wb> setvbuf but the standard says: "The contents of the array at any time
wb> are indeterminate."

Use read binary instead?

wb> BTW, I read an earlier article where Philib McNab suggested that expire
wb> should take note of prune files.  I would not have thought there was a
wb> problem with the current expire program because it does not renumber
wb> articles but merely beheads the list in each newsgroup.

This used to result in fatal Prune errors (Prune 0.1 was a rather
over-cautious young virgin).  I think (hope) it would be ok with
later Prunes.

--
John Washington