Re: [hylafax-users] consecutive bad lines > total bad lines?

[Lee Howard <faxguy@xxxxxxxxxxxxxxxx>]

On 2003.12.11 08:14 Giulio Orsero wrote:
> cvs20031206, rh6x, rockwell class1.
>
> RECV: 2291 total lines, 0 bad lines, 1 consecutive bad lines

The problem existed prior to the ECM development and was perpetuated by 
that development.  The problem is that the decoder doesn't take into 
account that RTCrow is supposed to reference the row number after the 
last image row (the first row of the RTC/EOFB) rather than the last 
line of image data.  I think that this was probably a result of 
confusion about the first row being numbered as zero (row 0).

The attached patch fixes this as well as 1) preventing the bottom one 
line of image data from being trimmed, and 2) it should correct some 
line-counting disagreement that we've seen between the HylaFAX CQ 
decoder and some Class 2 CQ mechanisms.

Lee.

diff -Nru hylafax-4.2.0.orig/faxd/G3Decoder.c++ hylafax-4.2.0/faxd/G3Decoder.c++
--- hylafax-4.2.0.orig/faxd/G3Decoder.c++	Wed Dec 10 13:33:42 2003
+++ hylafax-4.2.0/faxd/G3Decoder.c++	Thu Dec 11 09:52:31 2003
@@ -217,7 +217,7 @@
 	b1 = *pb++;
 #define	badlength(a0,lastx) do {			\
     if (isG4)						\
-	RTCrow = rowref-1;				\
+	RTCrow = rowref;				\
     badPixelCount("2D", a0, lastx);			\
     rowgood = false;					\
 } while (0)
@@ -228,7 +228,7 @@
 #define	badlength(a0,lastx) do {			\
     nullrow = (a0 == 0);				\
     if (nullrow && ++RTCrun == 5 && RTCrow == -1)	\
-	RTCrow = rowref-5;				\
+	RTCrow = rowref-4;				\
     badPixelCount("1D", a0, lastx);			\
     rowgood = false;					\
 } while (0)