Menu
▾
▴
#27 remove duplicate list due to letter case in filename
In windows platform, the filename is case in-sensitive, such that "io.h" and "IO.h" is the same file. But cscope seems to treat them as different files and will generate two file entries in the database. This will result in a symbol will match in two places and produce some inconvenience.
Below is the screen shot: All these four items are exactly the same.
http://www.flickr.com/photos/18351815@N00/2365060111/
Logged In: YES
user_id=827328
Originator: NO
No, you're misunderstanding this. Dos/Windows will treat io.h and IO.h as the same file in the same directory. As such there is no way for cscope to recognize one file as two different files. What you're most likely seeing is the result of different case file names in different directories (but with the same or simmilar content) getting picked up. This is working exactly as it should. If you manually search your source tree, you should find all the variants that cscope has in its list. Note also that if there isn't sufficient space to display it, cscope will truncate the path names on files.
example files for filename case problem
Logged In: YES
user_id=2047017
Originator: YES
I have attached a simple example to illustrate my problem.
I have main.c and main.h in the same directory. In main.c, I include main.h as:
#include "main.h", cscope.out is good, only main.c and main.h are listed. However, if I change as:
#include "MAIN.H", cscope.out will produce 3 files in it: main.c, main.h and MAIN.H
File Added: cs.zip
Logged In: YES
user_id=27517
Originator: NO
> #include "MAIN.H", cscope.out will produce 3 files in it: main.c, main.h
> and MAIN.H
Well, as the saying goes, don't do that, then.
cscope is a UNIX-based program, so it has some UNIX concepts deeply engrained in it. One of those is that filenames are supposed to be case-sensitive. Your source would fail to compile on Unix platforms because of this, i.e. it's a portability problem of the source at least as it one of cscope.
We could add an --ignore-file-name-case option (implementation would rely on a case-insensitive compare in dir.c:infilelist()). But for a traditional UNIX tool like cscope, I don't think that's a good idea. When in Rome, do as the Romans do.
Logged In: YES
user_id=2047017
Originator: YES
Thanks broeker, I agree with you.
But I have to work with many other people who may not concern the letter case of filenames.
So I have made a simple utility to trim the cscope database.
http://code.google.com/p/cscopetrim/