Klaus Schmidinger wrote:

tinconn@virgilio.it wrote:

...

Hi

a new plugin channelscan has been released. Channelscan uses a transponders .ini file from win project like prog-DVB software for create a channels.conf

The homepage of the plugin is at

http://www.kikko77.altervista.org goto

...

download page for the channelscan-0.0.4.tar.bz2 archive

---

---

...

changelog: 2005-04-11: Version 0.0.4

• Channelscan is now faster because sets the filters for

sdt, pat and

...

pmt itself, so it now knows when scanning of a transponder is

over.(idea from Luca

...

Olivetti) (write by Thomas Bergwinkl)

I see you've copied large parts of VDR's pat and sdt code in order to introduce what I assume shall improve performance.

In ealier version the plugin just switched to all transponders of a satellite, waited 10 seconds and let vdr add new channels. But of course it is not always necessary to wait so long, or sometimes after 10 seconds not all channels were added by vdr. So Luca Olivetti wanted a way to get a notification when a transponder scan is complete. So he suggested to wait until 'sdt.getSectionNumber() == sdt.getLastSectionNumber()'.

Of course this is not enough, you also have to wait for the pmt data. But there is the problem, that some pids are in pat, but don't get broadcasted. To speed up the scan, the plugin sets 10 filter for pmt-pids at a time. So when there is such a pmt pid, which isn't in the stream, it can't delay processing the other pids (as long not all 10 are missing, which is improbable)

A second attempt to avoid those timeouts is that only the pmt pid filter will be set for services which were really in sdt. Because it makes no sense to wait for the pmt data for services, which are not in channels.conf (which pids should then be set/updated?).

So this changes did i make: - sdtfilter tells patfilter which SIDs it received - patfilter sets only for these services the filters for pmtpids - patfilter sets up to 10 filters for pmtpids simultaneous - after all sections of sdt and the necessary pmtpids were proceeded, the plugin knows that the scan for the transponder is then over

Would you mind elaborating on all these changes and letting me know what and why they do, so maybe the official VDR code might also benefit?

The main idea for the modifications in the plugin was that more than one filter for pmtpids will be set at a time. But I'm not sure, if this is a good idea for vdr's patfilter. I think the number of filters which can be set simultaneous is limited. So when vdr sets too much filters for proceeding pmt data then perhaps there aren't enough left for recordings? Perhaps it would be possible to set always as much filters as possible and stop them when filters for more important things (like a recording) are needed. So perhaps priorities for filters would be a good idea?

Thomas

Thomas Bergwinkl wrote:

Klaus Schmidinger wrote:

[..]

...

I see you've copied large parts of VDR's pat and sdt code in order to introduce what I assume shall improve performance.

In ealier version the plugin just switched to all transponders of a satellite, waited 10 seconds and let vdr add new channels. But of course it is not always necessary to wait so long, or sometimes after 10 seconds not all channels were added by vdr. So Luca Olivetti wanted a way to get a notification when a transponder scan is complete. So he suggested to wait until 'sdt.getSectionNumber() == sdt.getLastSectionNumber()'.

For reference, here's the original thread: http://thread.gmane.org/gmane.linux.vdr/21006

Bye

tinconn@virgilio.it wrote:

Hi

a new plugin channelscan has been released. Channelscan uses a transponders .ini file from win project like prog-DVB software for create a channels.conf

A small bugfix... in channelscan.c

- if ((iPosition > 0) && (iPosition < oTransp.iArrTranspodersFile)) { + if ((iPosition >= 0) && (iPosition <= oTransp.iArrTranspodersFile)) {

otherwise cant select first and last inifile.