| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was causing problems in sign extension.
On amd64/Gentoo sign was extended:
(unsigned int)(char 0x80) became: 0xfffffff80
But on Pi2/Wheezy the same thing became: 0x80.
The obvious fix is to make it unsigned, as god intended.
All regression tests pass on pi2/wheezzy and amd64/gentoo.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When an entire 4K buffer of packet data contains no newlines, it's
simply retained for future use. But the code was failing to clear the
previous response in that case, causing the caller to see duplicate
data.
Since this bug only mattered with lines longer than 4K bytes, and
since even then it varied as a function of the alignment of lines
vs. 4K boundaries, it was never seen on previous regression tests.
TESTED:
Ran "scons build-all check" with the updated code and the corrected
test data for RTCM3.2.
Signed-off-by: Gary E. Miller <gem@rellim.com>
|
|
|
|
|
| |
Also update the affected regression files. gpsd had been
throwing away the DGPS status.
|
| |
|
|
I have seen up to about 490 in RTCM3.
|