Contents ----- Copyright Previous Next

Known Problems

The major problem of course is that if you have used the same name for different things in your source code, or you decide to vahunz a name you should not, it is easy to end up with vahunzed files that do not even compile. This program is not idiot proof, because this would required to implement huge parts of those things full compilers do.

When removing the indention, there are still some blanks or empty lines left. Use a language specific strip tools to avoid this.

The automatic creation of output directories is an AmigaOS-only feature. The simple reason for that: there is no ANSI C function for creating directories, and I don't know of any other C standard that would support it (POSIX is no standard - it's an illusion).

When you are doing very funky stuff with the C-preprocessor, it might not work any more after vahunzation. This especially refers to ## and such things. In my opinion, this is a bug in the design of the preprocessor, and a lack of brain if a programmer really uses such things.

You must not modify files during vahunzation. For example, you can change a source after the first pass, when all names have already been retrieved. If now an unknown name shows up in the second pass, Vahunz can no more ensure that all garbled name will not be the same as the new unknown word, and will abort with an error.

If carriage returns (\r) show up in the input, most of them will be gone in the output. If you use this tool on source files coming from Macintosh (with \r as line separator) or something worse (with \r\n), better use a cr/lf-converter before vahunzing them.

There is a limited line and name length when reading the lists for names and files. Currently this is 1023, which should be sufficient to read most identifiers and filenames. If a line is too long, the vahunz terminates with an error message. C-strings suck.

Non-Issues

The number of names and files is only limited by the available memory. The source files do not have any limitations concerning the line length or whole file size, as the are read character-by-character.

Stack usage should be very small, as most buffers are allocated dynamically, and the tree functions are none-recursive. Probably even the standard 4K are sufficient. Therefore, the program is compiled without stack check.