Branch, automatic test about igraphhpp HOT 6 CLOSED

But I usually do it manually :p

If you find a revision sufficiently stable in all 3 tests, you could create the 
stable branch as described in 
http://svnbook.red-bean.com/en/1.1/ch04s06.html (the Simple Tag part, assuming 
you've checked out the 
whole repository and not just the trunk).

Thanks, I will take a look about how to make a branch. But the problem for the 
moment
is that I don't find a compilable revision.

I've got some idea on how to make the test automatically. I will work it out 
tmr :)

I have made the regression test on my server. You can visit my server
http://phz548.phys to have a look about the compiling results of different 
compilers.

It is scheduled to svn update from the google code each 10 mins (now). And 
compile
the test cases periodically if the revision is newer. Current testing compilers 
are:
g++-4.4 with gnu++0x, g++-4.4, g++-4.3 with gnu++0x, g++-4.3, g++-4.2, g++-4.1,
llvm-g++-4.2
Almost all test are pass (includes my computer and mosix g++-4.4).

I am plan to mark a tags now, but do you think that the directory structure 
need some
modification? I mean that to flatten the directory structure by moving all 
*.hpp to
the same level as igraph.hpp

What does the 256 in g++-4.3 (GNU++0x) mean? Warnings? Errors? Wrong test case 
results?

I put the *.hpp in cpp/ because /usr/include/igraph/ already has many *.h. 

/usr/include/igraph/*.h (official headers)
/usr/include/igraph/cpp/*.hpp (C++ wrapper interfaces)
/usr/include/igraph/cpp/impl/*.cpp (C++ wrapper implementations)

Do you know if WINE can run command line applications? If yes we can get a copy 
of cl.exe (MSVC's compiler) 
and add it to the test too :)

The 256 is just the error code returned by calling the python os.system(). 256 
just
means fail to compile, and you can see the error in the compiled result page. I 
have
changed it already.

I understand your idea of placing those file. So how about the tmpobj.cpp? 
Anyway, I
just make the first tag of this project.

Running cl.exe by wine maybe a bit difficult. The important point is that it is 
not
the usual environment running the MSVC, so it may not reflect the actual 
problem. If
you find out a way to use the command like 'wine "cl.exe ..."', then I can add 
it to
the list of regression test.

OK, both problem have been solved