Hi folks
I've tried asking this in private mail for quite some time, but sadly
got no reply. So I end up needing to raise the question here.
Sangoma's s setup process includes a small patch to Zaptel. I have some
technical reservations with that patch. It seems that under certain
circumstances it may cause unexpected behavior when used with other
Zaptel channel drivers. I also don't understand why a safer method is
not used.
[Resending message, with a proper subject. After the morning coffee]
To be more specific:
The patch defines a number of ioctl numbers in the range of 60-65 in
zaptel.h . This is perfectly OK: that range is dedicated to
driver-specific ioctl-s. If users-pace ever calls those ioctl-s on
channels belonging to spans handled by that specific driver, Zaptel
will deliver them to the driver.
But then comes the patch to zaptel.c . Under certain circumstances
(which don't seem to be specific to the Sangoma cards. Might be specific
to digital spans) Zaptel is to call the ioctl handler.
Has it been tested with all other Zaptel drivers? Is it safe to apply
for a general-purpose package of Zaptel? A better strategy would
probably be to add an extra function pointer to struct zt_span and call
it in case it is not NULL in such a case.
For that reason I have rejected the Sangoma patch from Debian packages
and tried contacting Sangoma's developers for clarifications. Sadly, I
have not received replies. Others have different opinions than me and
thus the Sangoma patch is included in some binary distributions of
Zaptel Maybe this is just because the potential problem was not clear.
So right now I'm waiting for clarifications. Until then I would
recommend for anybody using Zaptel with non-Sangoma digital cards not to
apply that patch.
I see no reason why a patch that slightly adapts Zaptel to allow better
usage with Sangoma hardware could be applied. But such a patch should
first and foremost be technically sound.
--
Tzafrir Cohen
icq#16849755 jabber:tzafrir.cohen at xorcom.com
+972-50-7952406 mailto:tzafrir.cohen at xorcom.com
http://www.xorcom.com iax:guest at local.xorcom.com/tzafrir
Tzafrir Cohen wrote:> Hi folks > > I've tried asking this in private mail for quite some time, but sadly > got no reply. So I end up needing to raise the question here. > > Sangoma's s setup process includes a small patch to Zaptel. I have some > technical reservations with that patch. It seems that under certain > circumstances it may cause unexpected behavior when used with other > Zaptel channel drivers. I also don't understand why a safer method is > not used. > > [Resending message, with a proper subject. After the morning coffee] > > > To be more specific: > The patch defines a number of ioctl numbers in the range of 60-65 in > zaptel.h . This is perfectly OK: that range is dedicated to > driver-specific ioctl-s. If users-pace ever calls those ioctl-s on > channels belonging to spans handled by that specific driver, Zaptel > will deliver them to the driver. > > But then comes the patch to zaptel.c . Under certain circumstances > (which don't seem to be specific to the Sangoma cards. Might be specific > to digital spans) Zaptel is to call the ioctl handler. > > Has it been tested with all other Zaptel drivers? Is it safe to apply > for a general-purpose package of Zaptel? A better strategy would > probably be to add an extra function pointer to struct zt_span and call > it in case it is not NULL in such a case. > > > For that reason I have rejected the Sangoma patch from Debian packages > and tried contacting Sangoma's developers for clarifications. Sadly, I > have not received replies. Others have different opinions than me and > thus the Sangoma patch is included in some binary distributions of > Zaptel Maybe this is just because the potential problem was not clear. > > So right now I'm waiting for clarifications. Until then I would > recommend for anybody using Zaptel with non-Sangoma digital cards not to > apply that patch. > > I see no reason why a patch that slightly adapts Zaptel to allow better > usage with Sangoma hardware could be applied. But such a patch should > first and foremost be technically sound. > >Just out of curiosity, I have yet to see any issues with Sangoma cards and the way they ride on top (and patch) the Zaptel drivers. This personal dataset is around one hundred productions boxes. Two questions: 1. Have you found something that is broken? 2. If not, why are you so set on fixing things that are not broken? Your company (Xorcom) is a direct competitor of Sangoma, is that correct? I rarely answer questions or give competitors ideas that may come back and hurt my business. Thanks, Steve
On Nov 11, 2007 1:07 PM, Steve Totaro wrote:> [...] > > Your company (Xorcom) is a direct competitor of Sangoma, > is that correct? I rarely answer questions or give competitors > ideas that may come back and hurt my business.with all due respect, you are being presumptive projecting your philisophy on to others and expecting that they would do the same. I am not saying there is anyting wrong with your philisophy, just that everyone can have their own motives & objectives. -baji. --
(clarification) On Nov 11, 2007 1:07 PM, Steve Totaro wrote:> [...] > > Your company (Xorcom) is a direct competitor of Sangoma, > is that correct? I rarely answer questions or give competitors > ideas that may come back and hurt my business.with all due respect, you are being presumptive projecting your philisophy on to others and expecting that they would act the same way, & for the same reasons, as you would. I am not saying there is anyting wrong with your philisophy, just that everyone can have their own motives & objectives. -baji. --