[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20020417: Misc notes/fixes for v.5.6.e.1



Steven,

Still tarring/re-tarring parts for the 5.6.f release. Takes a while to
build the binaries and build the tarfiles, do the web page stuff, etc.
while machines bite the dust.

The .1 for past releases was a minor patch version number from NCEP.
This release is just 5.6.f. I think I've repacked 5.6.f for the last time on
the download site. I just need to formulate the announcement email.

Some of the gif driver files have changed/been renamed since 5.6.e.1,
so patching will be more hand work than applying the context diffs.

The biggest drawback of up'ing LLMDGG is that some users will have
stack size problems depending on how their kernels were built
(read HPUX generic is too small) where trying to statically allocate
that large of an array will force an allocation error (especially in gdvint).

Steve Chiswell
Unidata User Support



On Wed, 17 Apr 2002, Steven Danz wrote:

> Hello
>
> Couple things:
>     I'm working with a .5 degree avn right now, and while the LLMXGD is large
> enough for the grid (259,920pts)  the LLMDGG isn't large enough to compute
> divergence tendency (divt).   I used about the same ratio of LLMXGD/LLMDGG 
> that
> was in the original code and bumped LLMDGG to 9504000 and it seems to have 
> fixed
> my problem.
>
>     I've noticed that if you change the xsize or ysize when running the gif
> device driver that it results in a core dump from the gif process (it may even
> crash when you try to generate more than one image per use of the device, 
> can't
> remember now).  I've done some digging around and I have a set of diffs that
> seem to fix the problem that I thought I would pass them along in case you 
> were
> interested.
>
> I noticed that v.5.6.f was available for download, but I don't remember an 
> email
> about it.  Since it doesn't have a .1 on the end of the name, I was wondering 
> if
> it was 'ready for prime time' or just sort of a beta download.
>
> Thanks
>
> Steven Danz
>