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

20020527: McIDAS on Windows/VmWare



>From:  "Charles H. Murphy" <address@hidden>
>Organization:  Kean Uniersity
>Keywords:  200205271145.g4RBjKa02429 McIDAS-X VMWare Windows

Charlie,

>Here is the latest on using VmWare to run McIDAS.  When I got
>back I spent some more time on this with the following results:
>
>1. Host system - Windows 2000, guest system
>Red Hat 6.2, computer Compaq 450 Mhz PIII
>
>2. VmWare installed easily, Red Hat installed as the
>guest system without a hitch, McIDAS compiled with
>no errors.

(For the tracking system): this was using gcc and g77.  Do you happen
to know what version of gcc and g77 were used?  You can get the version
information from those compilers by using the '--version' flag:

gcc --version
g77 --version

>3. Mcidas started up with command window and image
>window.  It didn't start with the configuration widget that
>lets  you set # of frames, menu or gui, etc.

This is interesting.  The design of the startup procedure is to
run the GUI startup.  If the defaults file, ~/.mcidasrc, doesn't
exist, the GUI startup defaults to saving configuration parameters.
After ~/.mcidasrc exists, the GUI startup defaults to _not_ saving
configuration parameters you can set.  The GUI startup is a Tcl/Tk
script that gets linked to ~/bin/mcidas.  The fact that this did not
run seems to say that the link did not get created.  I have no
explanation for this (yet).

>4. There appear to be problems with connecting to the
>ADDE server at hurri.kean.edu ( could you check if you
>can connect - might be network gurus changing things?),
>so I exclusively used ADDE.UCAR.EDU

Right now, I am composing this reply on my machine at home running
in Windows 2000.  After I reboot into Linux, I will point at hurri
to see what kind of response I can get.  Pretty obvious why I am
interested in VMWare, huh :-)

>5. From command line data can be displayed OK.

This says a LOT.  Shared memory works correctly (a big thing) as do
interrupts (the image display screen gets interrupts from routines
that plot/display in shared memory segments).

>6. From the menu system everything seems to work OK
>(gine easts, goes east, radar loop, contour press and temp, wind barbs,
>wxtext, meteorogram, emmagram, fours, rtgrids/ngm. rtgrids/mrf-uw
>did fail because no grid matched search conditions).

Very good.  The non-working of the RTGRIDS/NGM-UW (I am supposing)
and RTGRIDS/MRF-UW is most likely caused by something on the server
end.

>7. The Unidata GUI works for point source data (plot T, contour T,
>contour pressures, meteorogram, SkweT, upper air contours,
>etc.)

Excellent.

>8. Unidata GUI does not work for images, the "error in TCl
>script" message box opens up and says "error: can't read
>"name": no such variable."  Tried this for gini east, rtimages,
>rtnexrad all give the same message.
>The stack trace is as follows:
>
>can't read "name": no such variable
>    while executing
>"lappend enhLists $name"
>    (procedure "addeImageLoad" line 90)
>    invoked from within
>"addeImageLoad [lindex $t 4]"
>    invoked from within
>".menus.disp.mnu.addeimag invoke active"
>    ("uplevel" body line 1)
>    invoked from within
>"uplevel #0 [list $w invoke active]"
>    (procedure "tkMenuInvoke" line 29)
>    invoked from within
>"tkMenuInvoke .menus.disp.mnu.addeimag 1"
>    (command bound to event)

OK.  I found the cause of this and corrected it.  if you load in a
new copy of upcguiprocs.tcl, this problem should go away:

cd mcidas7.8/src
ftp ftp.unidata.ucar.edu
  <user> anonymous
  <pass> your_email_address
  cd pub/mcidas/src
  binary
  get upcguiprocs.tcl
  quit

After downloading the corrected copy of upcguiprocs.tcl, you need to
make sure that the link between the copy in ~mcidas/mcidas7.8/src
and ~mcidas/bin is correct:

ls -alt ~mcidas/mcidas7.8/src/upcguiprocs.tcl ~mcidas/bin/upcguiprocs.tcl

If the inodes for these two files are the same, then everything is
good.  If they are different, then make a link from the copy in the
mcidas7.8/src directory to the bin directory:

<as 'mcidas'>
cd mcidas7.8/src
rm -f ~/bin/upcguiprocs.tcl
ln upcguiprocs.tcl ~/bin

After doing that, EXIT McIDAS and restart.

While you are mucking around with links, do the following:

<as 'mcidas'>
cd ~/mcidas7.8/src
rm ~/bin/mcidas
ln mcstart.gui ~/bin/mcidas

This will insure that the GUI startup is the thing that will be run
when you type 'mcidas' to start a McIDAS-X session.

>9. The loading of data and McIDAS performance such a
>looping images seems comparable to dual boot solaris
>on our machines at school - (also 450 Mhz)

This is very good news.

>10. Windows is working while VmWare is ruuing (I can play
>music CDs from Windows and listen while I'm working with
>mcidas,

This is good; it is always nice to have soothing sounds while working
with McIDAS.

>also the VmWare window can be minimized and
>you can work in wondows and then go right back to
>VmWare/Red Hat/McIDAS with a mouse click.

This is great!  I learned from Mike Wright that UCAR has a site license
for VMWare.  Right now, they only have the licenses that allow wone
to load Windows under Linux, but he has apparently gotten the ball rolling
on the license for running Linux under Windows.  For my home system,
I will want to run Linux (hopefully 7.1 is supported) under Windows 2000.

>11. I haven't had a chance to triy the binary for GEMPAK
>yet.

The key ingredients for GEMPAK will be the support of Motif and the
ability to run using an 8-bit X visual.  I, for one, will be very interested
in your findings!

>I think this is a pretty cool alternative to dual boot on
>Intel machines and fairly cost effective.

I agree completely.  The cost per license for the UCAR contract is
$99.  I think that this is very afordable especially since you don't
have to go out and buy an XWindows package.

>I know this is an unsupported platform combination - so
>I don't expect any support - but if you are interested ...

If this works cleanly (i.e., I don't have to start hacking code), I
will be recommending it to SSEC as an alternative to Interix.  I
am sure that they will be interested since they have made comments
about needing an alternative to Interix (comments made ever since
Microsoft bought Interix!).

>Take care,

Please let me know your results using the newly modified upcguiprocs.tcl.

Thanks Charlie!

Tom

>From address@hidden Tue May 28 06:24:23 2002
>Subject: Re: 20020527: McIDAS on Windows/VmWare

Tom,

Wow - I didn't expect a holiday response to my questions - wow.

Hey great the gui now works for images this is very good.
Additional comments interspersed below.

>The versions are:
>gcc --version
>egcs-2.91.66
>g77 --version
>GNU Fortran 0.5.24-19981002


>This is what I did - when setting and building McIDAS I moved .cshrc and 
>.mcidasrc over from another
>machine. So there was a .mcidasrc present when McIDAS started up - in 
>fact I edited it to start either the gui or menu.  However, I deleted it 
>and restarted McIDAS and it just starts the image windown and the 
>command window not the config widget.  By the way I did link bin/mcidas 
>to /mcidas7.8/src/mcstart.gui as you
>suggested.  I'm not sure why this doesn't work (it isn't a big deal 
>becasue I just edit .mcidasrc to start up
>the way I want).

re: using newly modified upcguiprocs.tcl

>Yes it works ad mentioned above - thanks.

re: link ~mcidas/mcidas7.8/src/mcstart.gui to ~mcidas/bin/mcidas

>Did this but doesn't start the gui-config  as mentioned above ??

re: will GEMPAK work?
>I'll let you know on this.

>Thanks a bunch on this - I'm really excited about this option for McIDAS.

>See ya,

>Charlie