On Thu, 14 Feb 2013, Pindar wrote:

> I also encountered some bugs (at least on W8) [...]

Some of these I can replicate, some not. What version of gretl 
are you running (if a snapshot, what's the date?). Anyway, 
here's a run-through of the points you made:

> - having 'opened' a bundle and then using the '= null' command gretl 
> crashes

I can't replicate this. Here the following works fine:

<hansl>
bundle b
b["foo"] = 6
print b
b = null
print b
</hansl>

> - 'smpl full' does 'sometimes' not work in functions - I think if it's 
> outside a loop.

Inside a function, smpl --full only restores the sample range 
that was given by the caller. That is, it undoes any 
restriction on the sample range that was set inside the 
function, but not one set outside. The following illustrates 
the actual (and intended) behavior:

<hansl>
function void foo (series y)
   print y -o
   smpl +3 ;
   print y -o
   smpl full
   print y -o
end function

open data4-1
smpl 3 14
foo(price)
</hansl>

> - 'delete --type=' does not update the symbols window automatically

Confirmed. There's an attempt at fixing that in CVS and 
today's snapshot. Should work for matrices and bundles. The 
scalars window may still need fixing.

> - delete --type=series is not working.

It's not supposed to work: "series" is not mentioned among the 
supported types for "delete --type=<typename>" in the help, 
and that is by design; you're not allowed to empty a dataset.

But note that you can delete many series at once by using 
wildcards:

<hansl>
nulldata 20
loop i=1..10
   series x$i = normal()
endloop
varlist
delete x*
varlist
</hansl>

> deleting a nonexisting lists leads to deleting the last 
> series of the dataset!

Confirmed, and now fixed in CVS and snapshots.

Allin Cottrell

Reply via email to