>       Still struggling with being unable to configure or compile lots of 
> macports.  Using port -d hasn't really changed much. it clearly is related to 
> clang, but not clear exactly how.

-d potentially provides us helpful debugging information.

>       python27 dies in the first call to clang in configure, python33 dies in 
> the second.  
> 
>       I can build both by running configure from a shell with the same call 
> as macports.  (using the call line from the top of configure.log). 

>       This same approach has worked for other standard ports (using GNU 
> configure scripts).    So the problem has something to do with how clang 
> behaves when called as a subprocess from macports (tcl).

Do you set the same environment variables and call it with the same user and 
flags that MacPorts would?

>       Some ports configure and build fine.  Have not checked in detail what 
> their configure scripts test or don't test for compared with ones that fail.  
>  
> 
>       Now my question is after building and installing from a shell, but into 
> /opt/local, how do I alter the macports db to indicate I have dependencies 
> installed?

Not sure if that's easily accomplished, since MacPorts will then try to manage 
those files (might run into permissions problems, missing files, etc). Perhaps 
this script is still valid and can help:
https://trac.macports.org/browser/contrib/repairfilemap

However, I feel you'd be better off doing that manual configure/compiling in 
the work directory of the various ports you're installing, and modifying the 
work/.macports.PORTNAME.state file to indicate build phase is complete. Then 
MacPorts can destroot, archive and install it properly without having to tweak 
the db in unsupported ways. If you'd like more thorough instructions for this 
please let me know :-)

_______________________________________________
macports-users mailing list
[email protected]
http://lists.macosforge.org/mailman/listinfo/macports-users

Reply via email to