Re: [las_users] LAS compare of two variables crashes or hangs server

[Jonathan Callahan <Jonathan.S.Callahan@xxxxxxxx>]

Penny,

Many thanks for tracking this one down and reporting back.  The Ferret 
visualization package used by Ferret has various MAX_LENGTHs and it 
appears you've found one that doesn't report an error when a string is 
too long.  I've forwarded your email on to the Ferret developers and 
asked for a list of places where you might have gone over the MAX_LENGTH 
of some predefined character array.  It could be any of:

   * file name
   * variable [D=...] qualifier
   * LET definitaion
   * something else entirely.

It would help if we knew what version of Ferret you are running so that 
we know whether the problem has already been fixed or not.


-- Jon


Penny Oots wrote:

> All,
>
> I don't know if this is a known LAS anomaly (fancy word for bug).  I 
> did not find it in the FAQ or LAS archives.  Matter of fact, it has 
> taken me weeks to stumble onto the solution to the crashes or 
> LASserver.pl hanging when comparing two months of the same parameter 
> variable.
>
> LAS had no problem displaying single months on one parameter but when 
> the 'compare two' tab was clicked on the same parameter, LAS would 
> display a blank popup and eventually timeout (if we did not run out of 
> swap space first and have to reboot).  If I selected the option to 
> cancel the job, I would get the message that the job was still active 
> and could not be canceled.  The debug_ferret.jnl file always stopped 
> at the same place with no error.
>
> While testing, I created the netcdf file with a name of 16 characters 
> plus ..cdf for the file extension.  This worked correctly.  When I 
> recreated the file with a file name of 46 or 55 characters plus the 
> .cdf extension, the 'compare two' hangs.
>
> I am in the process of recreating multiple files with shorter names.
>
> What is the maximum file name length for a netcdf file to be compared 
> in LAS?   A file with a name of 38 characters plus the .cdf extension 
> seems to work.
>
> Anyone else run into this problem?
>
> Penny