[Thread Prev][Thread Next][Index]

Re: [ferret_users] parsing in go-files fails

Hi Martin,

I'm not seeing this. What you show looks as though a command line "define symbol mv_r    = 3" is losing some text somewhere along the way as the command is read or as the parser is filling in the argument values to create the actual command that will run. I wonder if this is in the operation of the readline library or something of that sort.  Try this, to capture the commands as they run,

   yes? set redirect/tee/clobber/file=multi_view.out
   yes? set mode verify:always
   yes? go multi_view 3, 5, 0.15, 0.1, 0.03, 0.25, 0.10, 0.022, , 0.05,0.035

In the file multi_view.out, I'm seeing this, after all the introductory comment lines:

! define internal variables
    define symbol mv_r    = $1"5"      ! number of rows
 !-> define symbol mv_r    = 3
    define symbol mv_c    = $2"3"      ! number of columns
 !-> define symbol mv_c    = 5


On 5/22/2018 7:48 AM, Martin Schmidt wrote:
in the latest ferret version, v7.4,  an error found in v7.3 is still alive. A least a little bit.

        PyFerret v7.4 (optimized)
        Linux 4.4.126-48-default - 04/29/18
        22-May-18 15:49

yes? go multi_view 3, 5, 0.15, 0.1, 0.03, 0.25, 0.10, 0.022, , 0.05,0.035
 **ERROR: invalid subcommand: mv_r    = 3
defiymbol mv_r    = 3
Command file, command group, or REPEAT execution aborted

Opening data sets and rendering images works well otherwise.

I compiled pyferret by myself. The compiler is gcc-5, linux is OS42.3. Compiler flags are left as recommended. python is python2.7 coming with the system.  The code works well on one machine but the identical binary fails on other machines with the same OS42.3-system, all are 64-bit systems. - ldd applied to libpyferret.so gives identical results on both machines (after removing the address-information)
- the user environment is "similar"
The major difference is that the code is build on a E7-8893 v4 CPU. It works there. For many other software (including ferret v22) those binaries run on other platforms with E5 and i5-CPUs. No hardware specific optimisation flags were used. We are maintaining a common software pool this way, it works for many software packages.

I do not know, where in the code parameter parsing for go-scripts or go-script parsing is done. The error message "defiymbol mv_r" suggests a local memory problem or a problem with function parameter types?

All suggestions or help would be very welcome.

Martin Schmidt

[Thread Prev][Thread Next][Index]
Contact Us
Dept of Commerce / NOAA / OAR / PMEL / Ferret

Privacy Policy | Disclaimer | Accessibility Statement