Troubleshooting CScout processing

When processing a project for the first time, Cscout may terminate its processing with a syntax error. The errors are typically occurring because of incompatibilities between CScout and C compiler extensions or features of a particular version of the C Standard used in the project. Here is advice for troubleshooting such errors.

• Examine the offending source code.
• If the issue is obvious consider the following alternatives.
• Add a workaround macro in the appropriate file in the source code include/template directory, and reinstall CScout. Look within the existing header files for examples of how diverse compiler extensions are handled. If the fix is general and widely applicable, please consider contributing it back via a GitHub pull request.
• Skip the offending code, by adding a #if !defined(__CSCOUT__) conditional compilation directive.
• Skip the offending file, by commeting out the corresponding #pragma process directive of the employed CScout processing script.
• Extend CScout to handle the offending construct. In this case, please consider contributing back your fix via a GitHub pull request.
• If the issue is non-obvious it is likely that it is hidden underneath code generated by the preprocessor. In this case proceed as follows.
• Run CScout to proprocess the offending C file, with a command such as cscout -E /path/to/file.c main.cs >file.i.
• If the CScout script part associated with the offending part, contains #pragma process directives for header files, as in the example below, then the above method will not work.

  #include "/usr/local/include/cscout/csmake-post-defs.h"
  #pragma process "[...]/include/linux/kconfig.h"
  #pragma process "[...]/init/do_mounts_rd.c"
        

  In this case, employ the cscut command to extract from the processing script the part associated with the offending file (e.g. into a file named file.cs), and then run CScout in preprocessing mode on that file: cscout -E . file.cs >file.i
• Create a CScout processing script for the preprocessed file, such as the following.

  #pragma block_enter
  #pragma block_enter
  #pragma process "file.i"
  #pragma block_exit
  #pragma block_exit
        

• Process the new script with CScoutto see the line in which the error occurs in the preprocessed code and handle it as in the first case.
• To identify the correspondence between the preprocessed code and the original source code, try to match identifiers that are likely to be specific to the offending code part, such as the name of a function being defined.
• In all cases, optimize your work process as follows.
  • Use a cscout -P command to process only the parts of the processing script associated with the offending file. This will allow you to iterate through possible fixes and test them, without waiting for CScout to process all other files. As described above, if the processing script contains #pragma process directives for header files use cscut to extract the script part for the offending file, and only process this.
  • As you advance through the CScout processing script, generate a subset of unprocessed parts with a cscut -e command or by deleting the processed parts with your editor. When modifying the processing script by hand, be careful not to violate the project and file block nestings. You can remove the processing of entire projects (an outward #pragma block_enter / block_exit block) or entire files (an inward #pragma block_enter / block_exit), but you should never leave incomplete blocks.