On my setup with glibc-2.8, benchmark_residual linkage fails with
undefines references to clock_gettime(). Adding -lrt fixes that.
The following is a small patch for it.
Regards.
--
O.S.
diff --git a/configure.ac b/configure.ac
index 993ac33..392485e 100644
--- a/configure.ac
+++ b/configure.ac
The current text of the comment in the libFLAC/lpc.c states that
the "negative shift is a NOP in the decoder", but I don't think
this is accurate. The patch for a slightly reworded comment is attached.
Also this patch is connected to:
Erik,
The file indeed reached +/- 1 (one channel is the output of a magnetic
switched device, the audio signal is not strictly on-off but it has a
characteristic pattern that saturates).
I performed the test you've suggested. I used Audacity to convert
32-float to 24-signed. The original
On 22 January 2016 at 07:09, Erik de Castro Lopo wrote:
>> However, as a last attempt, I used Winrar on the original file and to my
>> surprise it was compacted to 79 Mb, only about 33% more than the FLAC
>> version representing a file with half the data.
>
> If the orignal
Ozkan Sezer wrote:
> On my setup with glibc-2.8, benchmark_residual linkage fails with
> undefines references to clock_gettime(). Adding -lrt fixes that.
> The following is a small patch for it.
Applied. Thanks.
Erik
--
--
lvqcl wrote:
> /share/compat.h includes win_utf8_io.h which in turn includes
> windows.h. But only flac/decode.c and share/grabbag/file.c
> need it. It was suggested earlier to guard this include by
> some #ifdef. I tried it, and it brokes compilation of some files
> because of missing includes.
lvqcl wrote:
> The current text of the comment in the libFLAC/lpc.c states that
> the "negative shift is a NOP in the decoder", but I don't think
> this is accurate. The patch for a slightly reworded comment is attached.
>
> Also this patch is connected to:
>