3.5 compilation woes

I'm having one of those days.

I want to upgrade to 3.5, since I'm starting to play with ies2rad and I know there were some major bugfixes since 3.4. I had problems with the makeall script back when 3.5 was released, so I just kept on using 3.4, but I recall Greg fixed a few more things in the script so I thought I'd try again. Here's what happenned:

Script seems to work, at least it no longer fails with an "if syntax" error when answering the questions. Rejoyce.

Compile process flies along, but spewed a few errors. Where is the compile log? Maybe I can find the answer to my problems there, but I don't know where it lives.

OK, so I figure I have no time to dilly-dally with this. Grab the binaries, install. Stuff doesn't work.

Reset permissions/ownership in the binaries, other stuff doesn't work. Specifically, rview complains it can't find rayinit.cal.

Reset permissions/ownership on the library files (including rayinit.cal), same stuff still don't work. ~8-\

So, now I have a busted system. Anyone got any clues for clueless over here?

···

----

      Rob Guglielmetti

e. [email protected]
w. www.rumblestrip.org

Rob Guglielmetti wrote:

Script seems to work, at least it no longer fails with an "if syntax"
error when answering the questions. Rejoyce.

Compile process flies along, but spewed a few errors. Where is the
compile log? Maybe I can find the answer to my problems there, but I
don't know where it lives.

On your screen...

Try this to capture it into a file (for [t]csh):

  $> makeall install >& log.txt

If you still want to see what happens, either use "tail -f" on
that file, or split the stream with "tee":

  $> makeall install |& tee log.txt

The "&" in both cases makes sure that both stdout and
stderr go to the file. As far as I know, there is no easy
way to have the csh redirect *only* stderr, which contains
the messages you're interested in.

OK, so I figure I have no time to dilly-dally with this. Grab the
binaries, install. Stuff doesn't work.

Reset permissions/ownership in the binaries, other stuff doesn't work.
Specifically, rview complains it can't find rayinit.cal.

Reset permissions/ownership on the library files (including
rayinit.cal), same stuff still don't work. ~8-\

Does your $RAYPATH still point to an old version of the library?

I'm not sure about the binary packages, but when you want to
compile yourself, then you also need a second "support" package,
which carries some stuff not in CVS, such as the tifflib and
possibly parts of the library.

-schorsch

···

--
Georg Mischler -- simulations developer -- schorsch at schorsch com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/

Hi Rob,

I hope you had a pleasant walk home after work yesterday...
Maybe the East Coast is more in need of Gouvernor Arnold Schwarzenegger
than California ...

back to the business:

compiling usually doesn't work at the first try, this is perfectly normal.
Equally usual, it's just some minor stuff like this or that header file not
found because of a wrong search path, so it pays off to be patient. The version mess nevertheless can be quite annoying, it's increasingly hard to keep track of all the different system specifics. Was it /usr/X11/include or /usr/include/X11 or /usr/f...it/X11R6/... (these are still the easy ones...)
Often links are set to redirect a lot of different namings, but just the one you need may be missing...

See it the positive way, you learn something about your system: where's what?
Do it like Schorsch proposed, and then look at the output file with all the compiler
messages. Look for 'file not found' items, then add the directory, in which these
headers reside on your system, to the rmake- "MACH" options with -I/your/specific/dir.
concerning rayinit.cal
When you do the compiling yourself, you know the directory where this file is moved
to, as you're asked by the script to give a name yourself (library-dir). This path then gets coded into the executables (but may be overriden by an environment variable setting) If you're using precompiled executables, look for some info what library path is set for these, so you can put all the cal files in the correct directory rightaway. I can imagine that this info is missing, so issue a fierce complaint at the binary package
distributor :slight_smile:

This rayninit.cal (i.e. all .cal-file) mess is certainly awkward, but on the other
hand this allows a great amount of flexibility by enabling you to load specific
stuff on startup, even complicated algorithms, which otherwise would have to be
compiled into the program.

-Carsten

P.S. I remember some other postings months ago dealing with installation problems, have a look at the mailing list archives.

Hi gang,

THanks for the swift replies, both on and offlist. I have tried your suggestions, but now I have a new problem: my screen is totally black. :wink:

The suggested PATH problem is not the cause, I don't believe, because I compiled to the same locations as I had 3.4. Everything is where is was before. So, I grabbed the .rpm installer that Mr. Anselmo set up, and was about 98% done with that when the power went out in my NYC office. Thanks for the tip on how to save the STDERR, Georg, I would really like to compile the source rather than using the precompiled binaries, so I will try your suggestions when the electrons begin flowing once again.

P.S.
It took my wife & I nine and a half hours to get home last night. But, with no electricity in the office even today, I get a free day off. =8-) And since about 80% of the people in Manhattan that were trying to get to New Jersey were all located at a single ferry terminal at 39th Street, the chances were good that I should run into a friend I used to work with at a bike shop ten years ago, which is exactly what happened. As an added bonus, I had several hours to kill, so we got all caught up on things. =8-)

Rob Guglielmetti
[email protected]
www.rumblestrip.org