Bug 370 - ipv4-interface.h is installed
: ipv4-interface.h is installed
Status: RESOLVED FIXED
: ns-3
internet-stack
: pre-release
: All All
: P2 normal
Assigned To:
:
:
:
:
  Show dependency treegraph
 
Reported: 2008-10-02 10:30 EDT by
Modified: 2009-03-02 10:48 EDT (History)


Attachments
patch (29.05 KB, patch)
2009-02-28 10:03 EDT, Gustavo J. A. M. Carneiro
Details | Diff


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2008-10-02 10:30:03 EDT
it clearly should not be installed: this is a private header. Removing it
breaks ./waf --python-scan on my system though. It looks like a compiler bug.
------- Comment #1 From 2008-10-02 10:36:54 EDT -------
we seem to have the same problem with ipv4-l4-protocol.h which is not installed
but which is needed to allow --python-scan to work with the latest gccxml. 

Basically, --python-scan is broken with latest gccxml :/
------- Comment #2 From 2008-10-02 10:49:52 EDT -------
damn ! There is no reason to install ipv4-l3-protocol and ipv4-static-routing
either :/

This thing is a mess and it looks like joe is the culprit who made these
additions (hg annotate says so).

So:
1) why do we install ipv4-l3-protocol, ipv4-interface and ipv4-static-routing ?
is it becacuse of the  --python-scan breakage ?
2) If it is because of this breakage:
  - first, must figure out whether this is not a gccxml bug. I think that it is
  - second, must implement a _different_ workaround. We can't install every
header because of _one_ stupid tool.

I am really annoyed to have to discover this now: that kind of change should
not go in without discussion. The list of installed headers is _public API_.
------- Comment #3 From 2008-10-02 11:49:19 EDT -------
I don't get that gccxml bug, with my old gccxml from around 2008-04-20.

What I do get is a WAF script bug: ./waf --python-scan generates new
bindings/python/ns3_module_*.py files, but WAF does not notice the new files,
and so does not run pybindgen with them to generate new .cc module files.  This
should be fixable, but I've been short on time...

Is the build bug I described what you've been having, or is it really a gccxml
bug?  If it's the gccxml bug, try downgrading (cvs up -D 2008-04-20) it.
------- Comment #4 From 2008-10-02 12:18:08 EDT -------
Well in fact fixing the build problem with --python-scan is hard, due to the
WAF architecture.  As work around I at least made WAF exit right after
scanning.
------- Comment #5 From 2008-10-08 04:41:37 EDT -------
I installed the latest gccxml (cvs HEAD) on fedora9 and and ns-3-dev
--python-scan fails. I spent a good while looking at the errors and I am fairly
certain that these are correct c++ code. (Can't copy/paste errors here since
the fedora9 box is not connected to the internet.
------- Comment #6 From 2008-10-08 05:59:00 EDT -------
As noted here,
http://www.nsnam.org/wiki/index.php/NS-3_Python_Bindings#The_semi-automatic_way

The version that is tested to work is not latest CVS but 2008-04-20.  A few
weeks ago I also tested gccxml CVS HEAD, but it only worked with pygccxml SVN
trunk, not with 0.9.5.
------- Comment #7 From 2009-02-28 10:03:52 EDT -------
Created an attachment (id=392) [details]
patch

I made the mentioned headers private and re-scanned python bindings, compiled,
regression testing, all appears to work with no problem.