Re: [certi-dev] Building Certi on Windows from CVS Head

[Eric Noulard]

2011/9/9 Jean-Philippe Lebel <address@hidden>:
> Good morning!
>
> Ok, now I'm able to fully compile Certi but I get some linker errors

That's better, but obviously not enough :-]

> 1 - file certiCheckHostAndIP.c, includes file netdb.h (not defined on
> windows).  I suggest something like in socket.hh

[...]

> Function get_myaddress is undefined on windows. I've found not workaround. I
> don't know if the file certiCheckHostAndIp.c is revelant anyway.
> Also, the projet CertiCheckHostAndIP shoud link with ws2_32.lib.

This executable is not mandatory, I suppress its compilation on Windows
from CMakeLists.txt.

> 2 - LibFedTime1516 has unresolved external for
> rti1516::VariableLengthData(void const *, unsigned long) and
> rti1516::VariableLengthData::data(void) const
> Shoud libFedTime1516 link with RTI1516?
> 3 - LibFedTime has unresolved external for ~RTIfedTime and RTIfedTime()
> 4 - RTI has unresolved external ~RTIfedTime and RTIfedTime()
> 5 - RTI1516 has unresolved external for rti1516::VariableLengthData(void
> const *, unsigned long) and rti1516::VariableLengthData::data(void) const

Those 4 link errors are annoying see:
https://savannah.nongnu.org/patch/?7566
https://savannah.nongnu.org/patch/?7569
https://savannah.nongnu.org/bugs/?30003

I cannot fix this just now.
May be you can try the patch from Christoph
http://lists.nongnu.org/archive/html/certi-devel/2011-06/msg00018.html

even if I'm not sure [yet] that this will be the final solution to this issue.

> 6 - Federate.cc line 193. Not sure the string is correct, shoudn't it be
> fom_paths.insert(fom_paths.end(),string(temp)+"\\share\\federations\\");
> instead? (two \\ before "share")

Yes it is fixed and commited.

> 7 - libcerti\objectclass.cc(1017) : warning C4700: uninitialized local
> variable 'oca' used
> Will likely crash ;)

You bet it will....
http://lists.nongnu.org/archive/html/certi-devel/2011-08/msg00006.html

fixed too, but be warned that ownership management has been unused (by us)
for some time now so I expect to discover some regression when someone
will try to use
it again.
We will fix it but I cannot ensure deadlline for that.

> I'm not familiar enough with the CERTI source code to propose patches, but
> I'll be pleased to test whatever you want.

If you want to help read the referenced bug report concerning the
Windows link trouble
and may be you'll propose something that works "clean" for all of us.

I may easilly review (read) Windows specific patches but
I won't be able to test very soon so may be other Windows user can help
to test and I'll integrate the proposed code when it looks good.


-- 
Erk
Membre de l'April - « promouvoir et défendre le logiciel libre » -
http://www.april.org