[gpsd-dev] TSIP Accutime Gold patches

[Eric S. Raymond]

Igor Socec <address@hidden>:
> I attached six patches.

(1) tsip_receiver_detect_and_new_packets.patch
(2) bits_putbed64_implementation.patch
(3) bits_putbef32_fix.patch
(4) satellite_is_used.patch
(5) tsip_packet_fix.patch
(6) json_add_data_type.patch

I've now done a second pass over these patches.  Here is the current state 
of things:

I merged (2) and (3). 

I made fixes equivalent to (4) without the variable name change.

I've merged an equivalent of (5).  The packet analyzer in the git repo
should now pass 0x1c packets to the TSIP driver and accept either the TSIP
generic or Accutime Gold length of the 0xbb packet.  Please note that all
TSIP packets need to be length-checked as strictly as possible in order to
avoid false matches with Garmin Binary packets.

I have partially merged (1).  I have incorporated your initialization code
for generic TSIP and Accutime Gold and part of your handling of response 0x1c.
Note one difference, though:  instead of creating a public member slot
session->specific_receiver, I have put that information where I think it
belongs, in the TSIP private driver structure. If, in the future, we need
to distinguish among other TSIP subtypes, we'll create new enumerated
values for that.

I have not merged (6) at all.  I see the appeal, but I'm reluctant to
pretend that the type universe of JSON includes short integral types
and bool when it doesn't really.

I don't want to merge the parts of (1) that create new gps_data_t
members, as this would break object library compatibility for
everybody. I think there are other parts of (1) that should be merged
into the mainline code, though, and we should explore that.

Please try to come up with a new minimal patch set from git head to what you
need and mail that to the dev list.  
-- 
                <a href="http://www.catb.org/~esr/";>Eric S. Raymond</a>