19990413: building mcidasx (cont.)

>From: address@hidden
>Organization: St. Cloud State
>Keywords: 199904091934.NAA23884 McIDAS-X make

Alan,

>My assistant (Todd) was working on Waldo this am, but he did not
>mention any activity that would have corresponded to your attempts.

I saw that he was logged on as 'mcidas'.  I tried to contact him with
"talk", but I got no response.  Your machine is probably fast enough that
he didn't notice anything unusual in terms of performance as compilations
were proceeding.

>He also tried another make this am, and left a note that it failed.

This was before I made my changes to .cshrc.

>I came on at about 11 am our time and read your email of friday, then
>checked to see the changes Todd had made.  I noticed some discussion
>about the .mcenv file and how it corresponds to the .cshrc file.  Found
>that what we had was a copy of our earlier .cshrc file, so I made the
>changes called for as best I could interpret.

.mcenv only affects the McIDAS remote ADDE server.  It has nothing
to do with your build failures.  We will, however, need to get this
right before the remote ADDE server can be used successfully.  I will
take a look at .mcenv before I logoff this time. (see below)

>Then, went back to ~mcidas/mcidas7.5/src and did a make clobber, then
>make all

OK, good.  The 'make clobber' insures that you are starting from scratch.

>Lots and lots of stuff seemed to all get done with no errors, but it
>did fail.  I don't know how far it got, must have been a ways since the
>makelog file has a size of 190k + before it stopped.

I logged on and note that it failed when it was attempting to run
configure for the netCDF component of the distribution.

>As far as I can tell, our compiler(s) should be ok as so many other
>files were compiled successfully.

The compilers are OK.

>I am not sure about the reference to netcdf.

McIDAS will have more and more ADDE servers for data in netCDF form.
SSEC's approach to supporting this with a minimum of headache for its
users was to bundle the netCDF with McIDAS.  In order to link against
netCDF libraries, the netCDF has to be built during the McIDAS build.

>I looked in your html files, and see a comments in the Notes &
>Warning section.  Did not see anything specific for solaris, unless the
>very last section is the one I need to follow (don't see a reference to
>Solaris7).

The last entry is the operative one.  The key is the 'i86pc' entry on
the line:

SunOS lenny 5.7 Generic_106542-01 i86pc i386 i86pc

This says that the operating system is Solaris 5.7 (which is the same as
Solaris 7) running on Intel (i86pc).

>I will check on this part some more (Notes & Warning re
>NetCDF environment variables)

I added the appropriate entries into /home/mcidas/.cshrc.

>Anyway, that is how I left Waldo at about 12:10 our time.

I found that you had downloaded the McIDAS distribution before I had put
out a fix to the mcfc script (makes gcc and f2c look like a Fortran
compiler).  I FTPed the new mcfc.sh source for this and restarted the
build:

cd mcidas7.5/update
ftp ftp.unidata.ucar.edu
<I just grabbed mcfc.sh from our mcidas account; you would login as if
 you were going to regrab the distribution and do the following:
  cd unix/750/bugfix
  bin
  get mcupdate.tar.Z
  quit
./mcunpack
cd ../src
make clobber
make all

This updates your source distibution with all of the latest bugfixes.
Please see the Addenda web page for McIDAS-X:

http://www.unidata.ucar.edu/packages/mcidas/mcx/addenda.html

for more information on bugfixes for the release.  You may want to grab
the updates as there have been a couple since my original release of
McIDAS-X.  In fact, I will go ahead and update your distribution
myself so that you don't have to worry about this step right now.

I retried the build and ran into a snag with the C++ part.  I'll have to
ask Steve Emmerson about this since he is the one that created the netCDF
configure stuff.  In the interim, I believe that if one comments out the
g++ entry in .cshrc (setenv CXX /usr/local/bin/g++), logs off and then
logs back on the build should continue.  I will try this myself to make
sure.

<few minutes later>
Yup, commenting out the definition for CXX worked.  The build is proceeding
as I write this.  I will let it run for awhile to make sure it gets to the
point of building McIDAS-X program executables.

As I am writing this, I am watching McIDAS-X build on waldo.  The make
process is almost finished with the SSEC McIDAS-X program executables.
After that, the routines included by Unidata will be built followed by
McIDAS-XCD (the 'all' in 'make all' means to build McIDAS-X and then
McIDAS-XCD).

Hey, your machine is pretty fast.  This build is zooming by...

While waiting for the build to finish, I decided to look at .mcenv.
It is, in fact, just a copy of .cshrc.  Since this is not what is
needed, I modified it for you.  For reference, the entries I put in
.mcenv came straight from the Configuring the mcidas and mcadde accounts
web page:

http://www.unidata.ucar.edu/packages/mcidas/mcx/mcxacct.html

It is now 12:56 MDT.  The McIDAS build ('make all') just finished
successfully.  I will now bring over all of the updated routines
(the reference to mcupdate.tar.Z above) and make your distribution
as current as it can get.

Ok, everything is uptodate. I will now install the distribution:

cd mcidas7.5/src
make install.all

Hmm... I had a little problem with the executable 'mcwish'.  I will
look into this when I can get to a machine that supports X windows
(I am currently working from home from a Windows NT machine).  Other
than that, the installation is complete and McIDAS _should_ be fully
functional.  The next step is for someone there to run the tests
outlined in the web documentation.

Talk to you later...

Tom