Allin Cottrell schrieb:
On Wed, 14 Jan 2009, Sven Schreiber wrote:
> given that at least two new loop bugs have been reported in very
> recent cvs versions (January 2009, one by me and the second one
> in this thread), it seems to me that the recent
> under-the-hood-changes in that area have the potential to
> produce other related bugs in the near future.
>
> So maybe the loop stuff should be systematically tested before a
> new release is made.
Thanks, Sven. I've tested the new code on a fairly wide variety
of scripts. The bug that you discovered was a localized parsing
issue, now fixed. The "bug" that Yuniarto found was, as I said
subsequently, a bug fix. Under the old scheme the loop index
variable was sui generis, but now it's a proper scalar variable,
which staves off various potential problems. A side effect of
that is that you can't use as loop index an identifier that is
already claimed as the name of a series (or matrix). But this is
already strongly implicit in gretl's variable-type system.
I certainly don't want to discourage further testing; but I'm
fairly confident that the current loop code is "right".
Fair enough, especially given that the assumption (two bugs in a short
period) turned out to be wrong...
I still plan to make minor adjustments to de.po quite soon (I wanted to
do it last night, but had networking problems), but those are certainly
not release stoppers.
cheers,
sven