20000623: McIDAS updates, ldm-mcidas7.6.3 on RH 6.2, question

>From: "James D. Marco" <address@hidden>
>Organization: Cornell
>Keywords: 200006231803.e5NI3JT07000 McIDAS-X addenda ldm-mcidas decoders

James,

>       I've got the new decoders installed on my RH 6.2 box.  But there were
>2 issues with the 7.6.xx update for mcidas.
>
>       1) mcunpack fails.  It appears that the FSF/RH people did some bad 
>       things to tar.  It fails when a directory is specified as "..".
>       The old RH 6.0 tar unpacked fine after moving the files to the old
>       McIDAS machine, hence this was non-fatal. (I could have installed the
>       old tar on the new machine also.)

OK, thanks for the update.  I have RedHat 6.1 and will be upgrading to 6.2
in the near future.  I will check this out as soon as I get a 6.2 development
platform.

>       2) After installing the update, xcd stuff failed to compile. I 
>       downloaded the full McIDASX version and it compiled OK.

Very strange.

>       I used the binary ldm-mcidas7.6.3 to avoid compiling issues. 

OK.  This is the easy thing to do in any case.

>       Unfortunatly, I overwrote the old directory with the new stuff
>       and lost the makelog...sorry...but, I think this is reproducible 
>       'cuz I tried twice from backup copies before rebuilding McIDAS. 
>       If anyone else does this, please forward the log to Unidata.

I see from your later note that you had a backup copy of makelog.  More
on that below.

>       3) Good news.  The LDM appears happy with the new decoders.

I got as similar report back from Gilbert Sebenste of NIU (after an
initial error which appears to have been caused by bad timing of changing
a link to the new profiler decoder binary.  After the first failure, he
saw no others.

>Query:
>       I didn't try to update the new Mcidas, downloaded 20000622.
>       Is this OK?; can I assume it has the Latest-and-Greatest code?  

Yes.  The addenda that are produced are cumulative, and I always try to
make a new distribution compressed tar file available at the same time
as an addenda is released.  The distribution tar file should, therefore,
always contain the latest  code.

By the way, I had running emails with the one user that noted potential
timing problems with conversion to PNG compressed imagery during which
I provided him with information on the PNG compressed AREA file format.
I think that he will be able to update his code well in advance of the
datastream change at the end of July.

>From address@hidden Fri Jun 23 13:00:38 2000

>I found a backup copy of the makelog from the update attempt.  

OK, I have deleted everything in the log file up until the point of
failure.

****snip*** ****snip*** ****snip*** ****snip*** ****snip*** ****snip*** 
 ...
compile wtxgserv.cp:            done
./mccomp -O -s -vendor -o wtxgserv wtxgserv.o -L. -lxcd -lmcidas -lX11
mcfc -s -O -o wtxgserv wtxgserv.o -L. -L/usr/local/lib -L/usr/X11R6/lib
-lxcd -lmcidas -lX11 -lf2c -ldl -lm
link    wtxgserv:               done
./mccomp -O -vendor -I. -I../netcdf/libsrc -c statdisp.cp
gcc -c -O -ansi -D_GNU_SOURCE -I/usr/X11R6/include -O -I.
-I../netcdf/libsrc statdisp.c
statdisp.c: In function `main':
statdisp.c:100: warning: return type of `main' is not `int'
statdisp.c: In function `draw_text':
statdisp.c:1178: virtual memory exhausted
compile statdisp.cp:            FAILED

##########################################
Wed Jun 21 16:24:01 EDT 2000: BUILD BEGIN
##########################################


##########################################
Wed Jun 21 16:24:02 EDT 2000: BUILD END
##########################################


##########################################
Wed Jun 21 16:24:11 EDT 2000: BUILD BEGIN
##########################################


##########################################
Wed Jun 21 16:24:12 EDT 2000: BUILD END
##########################################

./mccomp -O -vendor -I. -I../netcdf/libsrc -c statdisp.cp
gcc -c -O -ansi -D_GNU_SOURCE -I/usr/X11R6/include -O -I.
-I../netcdf/libsrc statdisp.c
statdisp.c: In function `main':
statdisp.c:100: warning: return type of `main' is not `int'
statdisp.c: In function `draw_text':
statdisp.c:1178: virtual memory exhausted
compile statdisp.cp:            FAILED

##########################################
Wed Jun 21 16:25:27 EDT 2000: BUILD BEGIN
##########################################


##########################################
Wed Jun 21 16:25:28 EDT 2000: BUILD END
##########################################


##########################################
Wed Jun 21 16:25:34 EDT 2000: BUILD BEGIN
##########################################


##########################################
Wed Jun 21 16:25:36 EDT 2000: BUILD END
##########################################

./mccomp -O -vendor -I. -I../netcdf/libsrc -c statdisp.cp
gcc -c -O -ansi -D_GNU_SOURCE -I/usr/X11R6/include -O -I.
-I../netcdf/libsrc statdisp.c
statdisp.c: In function `main':
statdisp.c:100: warning: return type of `main' is not `int'
statdisp.c: In function `draw_text':
statdisp.c:1178: virtual memory exhausted
compile statdisp.cp:            FAILED

You can see that the failure was caused by your system running out of
virtual memory.  The fact that statdisp.cp compiles during a full
rebuild is puzzling since there was not change to this module in any
addendum.  Is it possible that statdisp.cp somehow got munged on your
disk?  Starting fresh with a new distribution would get around this
by writing a new copy, of course.

Tom