Improve reliability on slow internet connections

[nickb937]

The following patch will reduce glitching by increasing the tolerances that the application will allow for delayed key response. I've been using this for 12 months on a 3G/GPRS connection with positive results. I would think it would be of benefit to other network connection types because there will always be occasions where ECM replies take 700ms or 1000ms to come back, due to unforseen network congestion.

diff --git a/device.c b/device.c
index 2c1a5ec..5535d76 100644
--- a/device.c
+++ b/device.c
@@ -978,10 +978,10 @@ bool cScCiAdapter::Assign(cDevice *Device, bool Query)

 #define MAX_CSA_PIDS 8192
 #define MAX_CSA_IDX  16
-#define MAX_STALL_MS 70
+#define MAX_STALL_MS 140

 #define MAX_REL_WAIT 100 // time to wait if key in used on set
-#define MAX_KEY_WAIT 500 // time to wait if key not ready on change
+#define MAX_KEY_WAIT 2000 // time to wait if key not ready on change

 #define FL_EVEN_GOOD 1
 #define FL_ODD_GOOD  2

[bas-t]

I have to ask: is this meant to make it possible to use cardsharing over the internet?

[nickb937]

That's already possible. This just makes it work better.

[bas-t]

Where I live, the use of FFdecsawrapper is legal, but not wihen using internet cardsharing servers.
I don't want to give our government a good reason to make it illegal here too, so I probably should reduce these settings. Not increase them. I'll think about it.

[bas-t]

On the other hand, it might make it possible to plan like 6 or even more recordings in MythTV starting at the exact same time. ATM I'm stuck with max 4.
That would be a valid reason to use your diff.
Have to test this first.

[bas-t]

I could not detect any improvement regarding the 'start many recordings beginning at the exact same time' issue.
Prove me wrong.

[nickb937]

Have you tried running one process for each adapter rather than having one process handle all adapters?

[bas-t]

No, I did not.
I'll try this when I have time.

[bas-t]

Running one process for each adapter takes way to much cpu cycles (like almost 3 times to much) when recording more then 1 show on the same adapter.
I'm not going to dive into this anytime soon, so I'm closing this now.