>> May I put your tarball to our public FTP archive?
Keith> Certainly. I hope it's useful to somebody. However, it still
Keith> needs a lot of work...
Ok. It's in ftp://ftp.ibr.cs.tu-bs.de/pub/local/libsmi/ now.
Keith> Question: how "complete" would you say libsmi is now? is it
Keith> really almost ready for 1.0? Do you anticipate any great
Keith> changes to the API?
I think the API is quite ok now, having the usual SMIv2 use cases in
mind. There are lots of error detections that could be added to the
parsers. Other things could be added like additional backends like an
`SMI daemon', or fast access to an SMI database. But these things are
not expected to change the API (significantly), and these changes are
not more than dreams right now.
On the other hand, we are research people. This means we have to
invent new things. ;-) Based on our work on SMIng we want to add
further information to the information model, which needs an interface
in the API. But these changes that we expect in future releases will
hopefully just *add* data structures and functions, so that existent
code will continue to work without changes. In theory. ;-)
Talking about the libsmi version, I'd say it's nothing more than a
number, that should be incremented from time to time. I don't believe
that a 1.0 release would stop us from any progress. ;-) I'm more
concerned about too many changes in the API that could get application
developers and users in trouble.
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
Keith> A little while ago I sent an addition to libsmi that provided Python
Keith> bindings using SWIG. I'm not sure it that actually went through as I was
Keith> having email problems. Did you see it?
Yes, thanks Keith! I (as the mailinglist owner) just had to resubmit
it to the list, since the list allows only members to post and your
submission was sent from your new email adress before you resubscribed
with your new address. You can check it yourself on the mailinglist
archive at http://www.ibr.cs.tu-bs.de/projects/libsmi/mail/.
Sorry, I did not yet find time to look at it and I'm not familiar with
Python (yet?). I know that Juergen took a look and now understands
your demand for an additional OID data structure. We think, this would
help to use SWIG, but we also think that it is more important to keep
the core C API not overloaded with nested data structures.
May I put your tarball to our public FTP archive?
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
Saurabh> The application for which I plan on using libsmi would
Saurabh> benefit from the convenience of being able to specify the
Saurabh> root of a subtree to which the search for the given object
Saurabh> should be restriced.
Did I get you right?... You wish to have a way to specify a node and
search for a specific node by a given name, e.g. search for sysDescr
underneath 1.3.6.1.2?
Saurabh> From my understanding of the API, there isn't a way to do
Saurabh> this? Am I right?
If I understand your wish correctly, you are right. This is not
possible.
Saurabh> If so, would the maintainers consider adding such an
Saurabh> interface in a future release?
I think, the only exact way to search for a node is by giving a module
and a name. A subtree root node is no good choice for limiting the
namespace of object identifiers, since (a) it is not capable to
preclude the possibility of a name duplicate and (b) it does not
(gernerally) allow to limit the set of nodes to semantic groups like a
module does.
So, I have a preference not to include such a function in the API.
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
I know from the description of the API in smi_node(3) that one can
retrieve information about an object given its name either by searching
the entire MIB tree or by containing the search to a specific module.
The application for which I plan on using libsmi would benefit from
the convenience of being able to specify the root of a subtree to
which the search for the given object should be restriced.
>From my understanding of the API, there isn't a way to do this? Am
I right? If so, would the maintainers consider adding such an
interface in a future release?
Saurabh
--
____________________________________________________________________________
Saurabh Jang iDEN OMC Software, NSS, Motorola Inc.
Lead Software Engineer Email : Saurabh_Jang(a)email.mot.com
1301 E. Algonquin Road, IL02-3346 Phone : +1-847-576-2630
Schaumburg, IL 60173, USA Fax : +1-847-538-5564
____________________________________________________________________________
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
Saurabh> I am trying to build libsmi-0.2 on a Solaris 2.5.1 machine
Saurabh> using gcc version 2.7.0 The "make" build terminates
Saurabh> prematurely while building data.c The problem seems to be
Saurabh> with a C++ style comment in data.h on lines 157-158. No
Saurabh> other header or source file seems to have such comments.
Saurabh> Is this a known problem ? Do later versions of gcc allow such
Saurabh> comments ?
Ouh. I'm ashamed for this. I must have been hacking to much Java when
I did this. I've changed it in the CVS code. It will be in the next
release.
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
Yes, later versions do accept C++ comments.
It is still bad style to use C++ comments in C.
Regards,
Andrew Hood
A distributed system is one in which the failure of a computer you didn't
even know existed can render your own computer unusable. -- Leslie Lamport,
as quoted in CACM, June 1992
"Any views or opinions presented are solely those of the author and do not
necessarily represent those of the Westpac Banking Corporation."
-----Original Message-----
From: jang(a)eis.comm.mot.com [mailto:jang@eis.comm.mot.com]
Sent: Tuesday, March 21, 2000 4:18 AM
To: libsmi(a)ibr.cs.tu-bs.de
Subject: [libsmi] C++ style comment in libsmi-0.2
I am trying to build libsmi-0.2 on a Solaris 2.5.1 machine using gcc version
2.7.0 The "make" build terminates prematurely while building data.c The
problem seems to be with a C++ style comment in data.h on lines 157-158.
No other header or source file seems to have such comments.
Is this a known problem ? Do later versions of gcc allow such comments ?
Saurabh
--
____________________________________________________________________________
___
Saurabh Jang iDEN OMC Software, NSS, Motorola
Inc.
Lead Software Engineer Email :
Saurabh_Jang(a)email.mot.com
1301 E. Algonquin Road, IL02-3346 Phone :
+1-847-576-2630
Schaumburg, IL 60173, USA Fax :
+1-847-538-5564
____________________________________________________________________________
___
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
I am trying to build libsmi-0.2 on a Solaris 2.5.1 machine using gcc version
2.7.0 The "make" build terminates prematurely while building data.c The
problem seems to be with a C++ style comment in data.h on lines 157-158.
No other header or source file seems to have such comments.
Is this a known problem ? Do later versions of gcc allow such comments ?
Saurabh
--
_______________________________________________________________________________
Saurabh Jang iDEN OMC Software, NSS, Motorola Inc.
Lead Software Engineer Email : Saurabh_Jang(a)email.mot.com
1301 E. Algonquin Road, IL02-3346 Phone : +1-847-576-2630
Schaumburg, IL 60173, USA Fax : +1-847-538-5564
_______________________________________________________________________________
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.
David Reeder writes:
David> It was so long ago you may have forgotten, but follows is the
David> beginnings of a C++ interface to the libsmi package. [...]
Great!
How should we proceed? My current preference is to let your C++
interface live besides libsmi as a separate package, because I'm not
familiar with it (and even not familiar with C++ ;-). When some time
went by and people would like to see both packages integrated, we
might decide to do so.
What do you think? May I put your tarball to our FTP archive?
David> Some minor patches against libsmi-0.2 are required for two reasons:
David> . "typename" is C++ reserved keyword.
David> ==> Changed to "type_name".
David> . C++ complains when string.c_str() is not passed as "const char *"
David> ==> Added a few "const" keywords, mostly in function prototypes.
David> Perhaps these patches can be applied to future releases? As LIBSMI
David> expands, more light changes may be necessary, but there should be
David> nothing more intrusive than this.
I'm fine with these changes. I've applied them to the libsmi CVS module.
--
!! This message is brought to you via the `libsmi' mailing list.
!! Please do not reply to this message to unsubscribe. To subscribe or
!! unsubscribe, send a mail message to <libsmi-request(a)ibr.cs.tu-bs.de>.
!! See http://www.ibr.cs.tu-bs.de/projects/libsmi/ for more information.