Is the Error Building Cuba-2.1 Causing Faulty Outputs?

  • Thread starter Thread starter Angelos K
  • Start date Start date
  • Tags Tags
    Building Error
Click For Summary

Discussion Overview

The discussion centers around the potential issues encountered while building the Cuba-2.1 library on an Ubuntu system. Participants explore the implications of error messages received during the compilation process, particularly concerning the Fortran components, and whether these errors affect the integrity of the outputs needed for C programming.

Discussion Character

  • Technical explanation
  • Debate/contested
  • Experimental/applied

Main Points Raised

  • One participant reports an error during the build process, specifically a conflicting type error related to the Fortran file, and expresses concern about the validity of the outputs produced.
  • Another participant suggests that the presence of the library and executable indicates that the build may be successful despite the error, referencing general behavior of 'make' and error codes.
  • A different participant challenges this view, emphasizing that the error indicates a failure in the test executed during the build, which could imply further issues with the library.
  • Suggestions are made to check the Makefile for specific references to the error and to potentially comment out the Fortran module if it is not needed.

Areas of Agreement / Disagreement

There is disagreement among participants regarding the significance of the error encountered during the build process. Some believe it can be safely ignored, while others argue that it indicates a failure that could affect the library's functionality.

Contextual Notes

The discussion highlights uncertainty regarding the impact of the Fortran-related error on the overall build, as well as the specific behavior of the 'make' process in this context. There is also a lack of consensus on whether the outputs can be trusted without further investigation.

Angelos K
Messages
43
Reaction score
0
Hi, all!

Unfortunately I'm not much of a computer guy, so I'm sorry if I leave out anything important.

I'm attempting to build Cuba-2.1. The distribution I'm using is:

DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=10.04
DISTRIB_CODENAME=lucid
DISTRIB_DESCRIPTION="Ubuntu 10.04.2 LTS"

I go inside the directory containing all Cuba files and do:

./configure

whereupon everything looks fine.

Then I do

make all

whereupon a huge amount of output is produced, ending:

ranlib libcuba.a
gcc -O0 -fomit-frame-pointer -ffast-math -DHAVE_CONFIG_H -I./src/common -I. -I. -o demo-c ./demo/demo-c.c libcuba.a -lm
f77 -g -O2 -o demo-fortran ./demo/demo-fortran.F libcuba.a -lm
MAIN cubatest:
integrand:
<stdin>:119: error: conflicting types for ‘integrand_’
./demo/demo-fortran.F:34: note: previous declaration of ‘integrand_’ was here
/usr/bin/f77: aborting compilation
make: *** [demo-fortran] Error 1


The outputs I need - cuba.h and libcuba.a - are produced, buit of course I fear they are faulty. I am unsure whether the error concerns demo files or anything vital, but I need to do everything cleanly...

By the way: the output talks a lot about fortran. I won't use this in fortran, but only in pure C. In addition, I have tried the same procedure with the 3.0 beta release. The same (or very similar) things occurred.

Thank you for your time and any help,
Angelos
 
Technology news on Phys.org
It's probably not anything that you need to worry about. If it produced the library and the executable that you need, then it's fine.

Taken from a website describing the 'gcc make' Error codes:

[foo] Error NN’ ‘[foo] signal description’ These errors are not really make errors at all. They mean that a program that make invoked as part of a recipe returned a non-0 error code (‘Error NN’), which make interprets as failure, or it exited in some other abnormal fashion (with a signal of some type). See Errors in Recipes. If no * is attached to the message, then the subprocess failed but the rule in the makefile was prefixed with the - special character, so make ignored the error.

I've had that happen with several different libraries during the make process. It's usually not anything to worry about.
 
FourierFaux said:
It's probably not anything that you need to worry about. If it produced the library and the executable that you need, then it's fine.

Taken from a website describing the 'gcc make' Error codes:

[foo] Error NN’ ‘[foo] signal description’ These errors are not really make errors at all. They mean that a program that make invoked as part of a recipe returned a non-0 error code (‘Error NN’), which make interprets as failure, or it exited in some other abnormal fashion (with a signal of some type). See Errors in Recipes. If no * is attached to the message, then the subprocess failed but the rule in the makefile was prefixed with the - special character, so make ignored the error.

I've had that happen with several different libraries during the make process. It's usually not anything to worry about.

Uhh :rolleyes: the error you mention is not given.
Moreover the build output says 'aborting compilation', so make really did not ignore the error.

Apparently a test is executed as part of the build to verify the correct behavior of the library.
And that test failed! Ignore at your own peril.
Furthermore, the tests that came after were not executed, so there may be more that is wrong.



To the OP, perhaps you can check the content of your Makefile and find the references to:
1. MAIN cubatest:
2. integrand:
3. ./demo/demo-fortran.F line 34

If it's only a problem with the fortran file which you do not use, perhaps you can comment out the fortran module or something and redo the make.
 
make really did not ignore the error.

Oh... I didn't realize that it would do that. Sorry. :redface:
 

Similar threads

Fortran Fortran output has lots of errors that I don't understand
  • · Replies 2 ·
Replies
2
Views
3K