gst-inspect-1.0 segfaults on tracing logs where it fails to find
element stats. So on the pipelines where we get the following WARNING
during execution will afterwards crash with a segfault as the
g_ptr_array has a index for it but it is just a NULL pointer.
WARN default gst-stats.c:444:do_message_stats: no element stats found for ix=X
An example of an pipeline which can reproducibly create a trace log
where this occurs would be this
GST_DEBUG="GST_TRACER:7" GST_TRACERS="stats;rusage;latency" gst-launch-1.0 videotestsrc num-buffers=120 ! autovideosink &> trace.log
gst-stats-1.0 trace.log
We do not have a way to know the format modifiers to use with string
functions provided by the system. `G_GUINT64_FORMAT` and other string
modifiers only work for glib string formatting functions. We cannot
use them for string functions provided by the stdlib. See:
https://developer.gnome.org/glib/stable/glib-Basic-Types.html#glib-Basic-Types.description
F.ex.:
```
../tools/gst-stats.c:921:11: error: too many arguments for format [-Werror=format-extra-args]
printf ("Number of Buffers passed: %" G_GUINT64_FORMAT "\n", num_buffers);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../tools/gst-stats.c:922:11: error: unknown conversion type character 'l' in format [-Werror=format=]
printf ("Number of Events sent: %" G_GUINT64_FORMAT "\n", num_events);
^~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86_64/include/glib-2.0/glib/gtypes.h:32,
from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86_64/include/glib-2.0/glib/galloca.h:32,
from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86_64/include/glib-2.0/glib.h:30,
from ../gst/gst.h:27,
from ../tools/tools.h:28,
from ../tools/gst-stats.c:30:
/builds/nirbheek/cerbero/cerbero-build/dist/windows_x86_64/lib/glib-2.0/include/glibconfig.h:69:28: note: format string is defined here
#define G_GUINT64_FORMAT "llu"
^
```
and
```
../tests/misc/netclock-replay.c: In function 'main':
../tests/misc/netclock-replay.c:98:23: error: unknown conversion type character 'l' in format [-Werror=format=]
if (sscanf (line, "%" G_GUINT64_FORMAT " %" G_GUINT64_FORMAT " %"
^~~
In file included from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86/include/glib-2.0/glib/gtypes.h:32,
from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86/include/glib-2.0/glib/galloca.h:32,
from /builds/nirbheek/cerbero/cerbero-build/dist/windows_x86/include/glib-2.0/glib.h:30,
from ../tests/misc/../../libs/gst/net/gstntppacket.c:38,
from ../tests/misc/netclock-replay.c:31:
/builds/nirbheek/cerbero/cerbero-build/dist/windows_x86/lib/glib-2.0/include/glibconfig.h:69:28: note: format string is defined here
#define G_GUINT64_FORMAT "llu"
^
```
This is needed for upgrading glib inside Cerbero which builds with
`-Werror` on Windows:
https://gitlab.freedesktop.org/gstreamer/cerbero/merge_requests/419
Errors causing the pipeline to fail going from NULL to PAUSED
were not displayed, and the pipeline was not dumped either in
those cases.
In addition, dumping the pipeline from the sync handler means
the dump matches exactly the state of the pipeline at the
moment the error was posted.
Concurrent Windows' colored debug message and g_print will print
string hard to read. Instead, use gst_print* which serialize
debug output and the APIs call.
We use to display the latency of each element in random order which is
not very convenient when comparing latency between different runs.
Sort them by "first activity" (the first latency reported for each
element) so it's consistent betwen runs.
This is the same logic when sorting and displaying element stats.
Latency is easier to read when formatted as time rather than displayed
as a flat number in ns.
Especially when displaying GST_CLOCK_TIME_NONE which is now formated as
99:99:99.999999999 instead of 18446744073709551615.
This will output latency information when parsing a log file with gst-stats that
has latency trace information. It will show the min, max and mean latency for
the pipeline and all its elements. It will also show the reported latency for
each element of the pipeline. Output example:
Latency Statistics:
pulsesrc0_src|fakesink0_sink: mean=190000043 min=190000043 max=190000043
Element Latency Statistics:
flacparse0_src: mean=45561281 min=654988 max=90467575
flacenc0_src: mean=89938883 min=81913512 max=97964254
flacdec0_src: mean=45804881 min=228962 max=91380801
Element Reported Latency:
pulsesrc0: min=10000000 max=200000000 ts=0:00:00.262846528
flacenc0: min=104489795 max=104489795 ts=0:00:00.262898616
flacparse0: min=0 max=0 ts=0:00:00.262927962
When printing a GstStructure property (e.g. the "stats" property in
rtpsession) the first field is printed on the same line of the type
description, and this is both inconsistent compared to how Enum values
are printed and confusing as the reader might miss the first field.
To fix this, add a newline before printing GstStructure fields in
properties.
NOTE: this does not change the existing inconsistent behavior of an
extra newline *after* a GstStructure property, but the latter is not as
annoying and it would take more effort to fix because GstStructure
fields are printed in CAPS descriptions too.
gstharness.c: Use G_GSIZE_FORMAT instead of hard-coding %zu
error: unknown conversion type character 'z' in format [-Werror=format]
gst-inspect.c: GPid is void* on non-UNIX, and we only use it on UNIX
error: initialization makes pointer from integer without a cast [-Werror]
gstmeta.c: Use and then discard value
error: value computed is not used [-Werror=unused-value]
With this, gstreamer builds with -Werror on MinGW
Commit 56b4fbef5e refactored the pipe code
to use GLib utility, but the patch was hading some other changed. LESS
env was now hardcoded in the middle instead of from a define and was
changed from FXR to -RX. The "-" is not even valid for LESS env, and
with the lost of F, we would still use a pager when the content fits the
terminal.
Setup it only if we have something to print out about inspected results.
Otherwise, gst_tools_print_version() output will be redirected to pager and also
exit immediately without waiting child process.
Not only this will make colored output work on old terminals and console
as well, terminals can theme the actual colors this way to make it fit
with their different themes this way.
Let's make the output a bit pretty to read. The colored output can be
disabled with `--no-colors` option or by setting `GST_INSPECT_NO_COLORS'
env (to any value).
The chosen colors are based on the popular Solarized theme, which is
targeted for both dark and light backgrounds.
Note:
* We only support true colors. If the terminal doesn't signal support for
that via 'COLORTERM' env, we disable colored output.
* We don't add colors to --print-plugin-auto-install-info output, as
that's meant for machines, not humans. Not only machines don't care
about beauty, the existing ones will likely not expect colors and choke
on it and we'll get angry mob at our doors.
[1] https://ethanschoonover.com/solarized
When printing info about a specific plugin, there is no need to prefix
some of the details with plugin's name. It's not only redundant but also
inconsistent and makes the task of adding consistent coloring to the
output (which we'll do in a follow patch), harder.
This emulates the default behaviour of git help pages, and also fixes
a bug on macOS where `less -F` doesn't display anything at all when
the output is shorter than one terminal screen.
Also moved the DEFAULT_PAGER define to after the includes, because
it's an unprefixed define.
Fixes https://gitlab.freedesktop.org/gstreamer/gstreamer/issues/330
