g++ include path
Hello,
I trying to compile a rather large project initially written and compiled with MSVC++ using g++ in Linux. Within the project I have a local include folder, which I specify in my Makefile as a "CFLAG=-I/path/to/include".
This works well, as long as the header files within the include folder don't use other header files living in the same include folder. When a header file from the include folder includes another header file from this same folder, I get this compile error (../../Include is the local include folder):
../../Include/headers/header_1.h:9:35: fatal error: headers/header_2.h: No such file or directory
If it found header_1.h why doesn't it find header_2.h, given that they are in the same folder?
Any ideas?
Thanks!
I trying to compile a rather large project initially written and compiled with MSVC++ using g++ in Linux. Within the project I have a local include folder, which I specify in my Makefile as a "CFLAG=-I/path/to/include".
This works well, as long as the header files within the include folder don't use other header files living in the same include folder. When a header file from the include folder includes another header file from this same folder, I get this compile error (../../Include is the local include folder):
../../Include/headers/header_1.h:9:35: fatal error: headers/header_2.h: No such file or directory
If it found header_1.h why doesn't it find header_2.h, given that they are in the same folder?
Any ideas?
Thanks!
| This works well, as long as the header files within the include folder don't use other header files living in the same include folder. |
@ne555: funny thing: I can't reproduce it with a minimal example...that works just fine. The problem seems to lie somewhere else. I will keep trying.
@kbw: I have tried either way...
Anyway, I found out that if I include header_2.h as "header_2.h" without path, it works fine. I don't like that solution, but if it works......
Thanks for the help.
@kbw: I have tried either way...
Anyway, I found out that if I include header_2.h as "header_2.h" without path, it works fine. I don't like that solution, but if it works......
Thanks for the help.
Why would you use hard-coded paths in any of your source-files?
...usually you would pass the "include" folder to g++ via compiler flag:
g++ -c -o test.o -I ../include/ test.cpp
This way you can use either file anywhere and won't get path problems...
-I == capital i
...usually you would pass the "include" folder to g++ via compiler flag:
g++ -c -o test.o -I ../include/ test.cpp
This way you can use either file anywhere and won't get path problems...
-I == capital i
Last edited on
It's not that I want to....the code is not written by me originally. I am just trying to port it from MSVC to gcc, and I'd wanted to avoid changing all those headers.
But it seems there is no way around it ;)
But it seems there is no way around it ;)
| But it seems there is no way around it ;) |
Sure there is, you just have to do it right.
| If it found header_1.h why doesn't it find header_2.h, given that they are in the same folder? |
You're not including header_2.h, you're including headers/header_2.h, which apparently does not exist in any of the include paths.
Mhm...since you're under *nix maybe you could symlink the /header folder into itself ?? All header files should find themself then in a pseudo subfolder?!
e.g. ln -s . include
(replace include with whatever path your headers included)
e.g. ln -s . include
(replace include with whatever path your headers included)
Topic archived. No new replies allowed.