Thanks. Miklos
On Tue, Dec 17, 2013 at 3:45 PM, Martin Braun wrote:
> On Mon, Dec 16, 2013 at 5:18 PM, Miklos Maroti
> wrote:
>>
>> Can you guys add me to the list of developers (user name: mmaroti)?
>> Then I can file the bug report.
>>
>
> In case this wasn't the case, you now have access to t
On Mon, Dec 16, 2013 at 5:18 PM, Miklos Maroti wrote:
> Can you guys add me to the list of developers (user name: mmaroti)?
> Then I can file the bug report.
>
>
In case this wasn't the case, you now have access to the wiki.
MB
___
Discuss-gnuradio mail
Hi Tom,
On Mon, Dec 16, 2013 at 7:27 PM, Tom Rondeau wrote:
> On Mon, Dec 16, 2013 at 1:08 PM, Miklos Maroti
> wrote:
>> Hi Guys,
>>
>> Ok, I have sent a pull request.
>>
>> On a related note: I do not understand the scaling performed in the
>> constellation constructor. Why are the constellati
On Mon, Dec 16, 2013 at 1:08 PM, Miklos Maroti wrote:
> Hi Guys,
>
> Ok, I have sent a pull request.
>
> On a related note: I do not understand the scaling performed in the
> constellation constructor. Why are the constellation points scaled to
> have average 1.0 amplitude? It prevents me to do in
Hi Guys,
Ok, I have sent a pull request.
On a related note: I do not understand the scaling performed in the
constellation constructor. Why are the constellation points scaled to
have average 1.0 amplitude? It prevents me to do interesting stuff
with constellations. Why cannot I set the d_scalefa
Hi Marcus,
> Anyway, I think Miklos is right: There is an error in calc_soft_dec in
> using the pre_diff_code even in the case that it is not set or its
> application disabled.
>
> Miklos, would you be as nice as to file a bug for that? You're the one
> who successfully diagnosed that.
Can you gu
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Hi Phil, hi Miklos,
as far as I can tell, the default constructor of constellation doesn't
initialize d_pre_diff_code, which is correct (well, kind of correct)
considering that it does not set any constellation points, either.
So I guess both members
On Mon, Dec 16, 2013 at 8:16 AM, Miklos Maroti wrote:
> Hi Marcus,
>
> Ok, that makes sense. It does not crash with QPSK, but it does with
> BPSK and 8PSK, and BPSK and 8PSK do not set the pre_diff_code. This
> explains why there is crash in one case and not the other. Although
> neither sets the
On 12/16/2013 08:16 AM, Miklos Maroti wrote:
> Hi Marcus,
>
> Ok, that makes sense. It does not crash with QPSK, but it does with
> BPSK and 8PSK, and BPSK and 8PSK do not set the pre_diff_code. This
> explains why there is crash in one case and not the other. Although
> neither sets the apply_pre
Hi Marcus,
Ok, that makes sense. It does not crash with QPSK, but it does with
BPSK and 8PSK, and BPSK and 8PSK do not set the pre_diff_code. This
explains why there is crash in one case and not the other. Although
neither sets the apply_pre_diff_code, so even QPSK should not set
pre_diff_code.
I
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Well, after having a debug build and then adding a lot of "volatile"s
into the code at the crashing function, it looks as if the
d_pre_diff_code of the constellation has size 0; and thus, you can't
access the 0. (read: first) element. How this could ha
Hi Marcus,
I do not think it is a compiler bug, but something uninitialized.
Constellation_qpsk and constellation_calcdist does not crash (I did
not verify if they work). However, if I generate the LUT for
constellation_qpsk, then it crashes.
# uname -a
Linux maroti-t410 3.11.0-14-generic #21-Ubu
Could you post the whole backtrace (as a github gist, pastebin etc)?
That would be very helpful, along with information such as:
Compiler (Gcc (version), VC++, clang ...), Platform (Linux, win, mac,...),
architecture (x86(64), ARMv9, IBM S7390,...) and GNU Radio source version.
Thanks and happy h
Seems like a strange error for C++. I usually get for loop errors with C
where you can't declare variables in the for loop. Perhaps try to declare j
outside the for loop.
AM KC3BBL
On Sun, Dec 15, 2013 at 7:05 PM, Miklos Maroti wrote:
> Hi Guys,
>
> Constellation soft decoder crashes for conste
14 matches
Mail list logo
