AW: net.sf.saxon.exslt.Random

> What does the signature random:random-sequence(number?, 
> number?) mean? Does it mean that the function can have 0, 1, 
> or 2 arguments, or does it mean that the arguments can be ()? 
> I guess the former, since () wasn't invented in 1.0.

The EXSLT spec says that both parameters are optional, so the function
can have 0, 1 or 2 arguments. The default values should be 1 for
numberOfItems and seed should be equal to current-time (I think it means
fn:current-time()).
 
> So if it has one argument, is this the number of items to be 
> returned, or the seed? The spec doesn't seem to say. Your 
> implementation makes it the seed.

For me it makes more sense that it should be the numberOfItems, but how
can I achieve this with Saxon? Do I have to remove randomSequence(long
seed) from the Random class?

> Your implementation accepts a seed that's a long. This means 
> that supplying 0.1 and 0.2 will give you the same sequence. 
> This doesn't feel right; I think it should be a double.

Ok, what do you suggest to use: Double.doubleToLongBits or
.doubleToRawLongBits? You'll code that?

> I've done some internal changes to make it return values in 
> Saxon's internal data formats to avoid the need for 
> subsequent conversion. Ideally I would make it return an 
> iterator, in which case the sequence could be infinite, but 
> then I'd need somewhere to keep the current state, which is tricky.

Ok, that's why random:random-sequence(1,1) doesn't work well with the
current implementation.

Martin

> 
> Michael Kay
> 
> > -----Original Message-----
> > From: Martin Szugat [mailto:[email protected]]
> > Sent: 30 June 2004 19:28
> > To: [email protected]
> > Subject: net.sf.saxon.exslt.Random
> > 
> > Attached you'll find a ZIP archive with an Eclipse 3.0
> > project including
> > the source code for the net.sf.saxon.exslt.Random class and a
> > corresponding test case called RandomTest. The test is repeated 400
> > times and ensure that
> > 
> > - the random number lies between 0 (inclusive) and 1 (exclusive)
> > - for the same seed the same random numbers are generated
> > - for different seeds, different random numbers are generated
> > 
> > I'll hope you'll find it useful! If so, please notify me when you 
> > integrated it, so I can mention it in my article.
> > 
> > Greetings,
> > 
> > Martin
> > 
> > PS: If you have some suggestions how to improve the code or 
> something 
> > isn't good enough, feel free to contact me! I'll fix it :)
> > 
> > > -----Ursprüngliche Nachricht-----
> > > Von: SourceForge.net [mailto:[email protected]]
> > > Gesendet: Mittwoch, 30. Juni 2004 19:37
> > > An: [email protected]
> > > Betreff: [saxon - Help] RE: random in loop returns same 
> > number each t
> > > 
> > > 
> > > 
> > > Read and respond to this message at:
> > > https://sourceforge.net/forum/message.php?msg_id=2643233
> > > By: mhkay
> > > 
> > > Just send it in when it's tested and I'll integrate it.
> > > Remember to send the test cases too. If you can devise a way 
> > > of making the test repeatable (so it gives the same results 
> > > each time) that will help the regression testing no end.
> > > 
> > > Michael Kay
> > > 
> > > 
> > 
> ______________________________________________________________________
> > > You are receiving this email because you elected to monitor
> > > this forum. To stop monitoring this forum, login to 
> > > SourceForge.net and visit: 
> > > https://sourceforge.net/forum/unmonitor.php?forum_id=94027
> > > 
> > 
>