[DGD] File access question

Noah Gibbs noah_gibbs at yahoo.com
Sat Jul 14 00:34:02 CEST 2007


  The trick to more quickly read in just part of a file is to know where to
seek to.  Two simple-ish ways to do that:

  * Store an index of line starts so that you can quickly seek to the beginning
of a line (N).  This requires keeping an index up to date, and you have to load
it before you can read the file.
  * Have fixed-length lines.  Then you don't need an index, you can just do a
multiply to find the start offset.

--- Kurt Nordstrom <kurt at blar.net> wrote:

> Yes, the file format is totally up to me.  Do you have some suggestions in
> that regard?
> 
> -Kurt
> 
> >   Do you control the file format?  This could make reading 200 random
> > entries
> > easier.
> >
> >   Yes, you'll probably want to do this over multiple invocations to avoid
> > running out of ticks.  If you have enough privilege it's also possible to
> > do
> > this within a tick-unlimited rlimits() statement, but that's only possible
> > in
> > certain cases.
> >
> ___________________________________________
> https://mail.dworkin.nl/mailman/listinfo/dgd
> 





      ____________________________________________________________________________________
Shape Yahoo! in your own image.  Join our Network Research Panel today!   http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7 





More information about the DGD mailing list