22 Jul
2013
22 Jul
'13
12:36 p.m.
On Mon, Jul 22, 2013 at 05:00:12PM +0800, Harald Welte wrote:
Hi,
Also, rather than that a real async hlr interface is
needed. There are many areas where contributions are welcome. So far I yet have to see
anyone else making any reasonably sized contribution to the sgsn. This always makes me
sad. Not even the past and current llc sequence number / tlli problems which should have
been visible to any user seem to have encouraged much in terms of submitting fixes :/
we see an increased use of OpenBSC while the amount of devs mostly
stay the same. I wondered about ways to handle or improve this. The
below is a list of random thoughts.
* I started with specifying simple tasks and offering guidance when
someone wants to implement that. One could go to Universities and
hacker spaces to find people motivated to try it. So if there is
1/10 success rate on such projects it would already be positive.
* Obstacles. One needs to have access to a base station to do meaningful
work. Now thanks to you there are plenty of individuals with a BS11,
and then there are Nokia/Ericsson/nanoBTS and sysmoBTS out there. We
also have public events like the XXC3, Camp, OHM. So maybe we should
be more active in announcing that we want implementers at these
events? Or maybe even hold a two day event in Berlin to ask interested
people to implement things?
* Seek for monetary support. Sure some of our commercial users, use
the software because it is of zero cost and would never pay a dime
for anything. But maybe there are others that want to contribute some
money for features? We could do some more ads that companies like
sysmocom offer high class, cost-effective customisations to OpenBSC.
* Adopt a more shiny/structured website. On the one hand the wiki
is a manifestation that users don't even correct content after they
found a solution but maybe we can do things to make the entry for
developers/contributors more easy. In fact I had already set-up
the software powers a book like this[1]. Maybe we can start with
a couple of 'unreliable' guides to {GSM, OpenBSC, Testing}. I had
also planned to create such material for the employee's of sysmocom.
* Adopt a model like it is/was(?) used by PackageKit. Contributors
do get direct access.. the public needs to wait a penalty time to
see the code. It makes it clear that value contributors more than
users.
Linking a regular expression matching library seems
excessive and an additional dependency for no good reason. Feel free to implement a simple
prefix match, possibly by reusing code from the smpp route matching.
man regcomp, it is part of posix.
[1] http://book.seaside.st/book