Histograms for H264EncoderImpl/H264DecoderImpl initialization and errors

Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
Committed: https://crrev.com/12f4cda0867b7264f0ad65b2b00e592de9a2a753
Cr-Commit-Position: refs/heads/master@{#11736}

[hbos, 2016-02-22 13:45:59]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl initialization and errors

All stats are counts, an instance of H264EncoderImpl/H264DecoderImpl
will report at most 1 event for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it retries in a
loop. The Init stats will give us an idea of usage counts for the
encoder/decoder. The Error stats will give us an idea of how many of
these usages encounters some type of problem, such as encode or decode
errors.

- WebRTC.Video.H264EncoderImpl.Init: Occurs at InitEncode.
- WebRTC.Video.H264EncoderImpl.Error: Occurs if any type of error occurs
  during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Init: Occurs at InitDecode.
- WebRTC.Video.H264DecoderImpl.Error: Occurs if any type of error occurs
  during initialization, AVGetBuffer2 or decoding.

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

All stats are counts, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 init
and 1 error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Init: Occurs at InitEncode.
- WebRTC.Video.H264EncoderImpl.Error: Occurs if any type of
  error occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Init: Occurs at InitDecode.
- WebRTC.Video.H264DecoderImpl.Error: Occurs if any type of
  error occurs during initialization, AVGetBuffer2 or
  decoding.

BUG=chromium:500605, chromium:468365
==========

[hbos, 2016-02-22 13:52:12]

hbos@webrtc.org changed reviewers:
+ henrik.lundin@webrtc.org, stefan@webrtc.org

[hbos, 2016-02-22 13:52:13]

Please take a look, stefan & hlundin.

I looked at other examples of histograms in WebRTC. I will have to make a
chromium-CL modifying histograms.xml as well.

(Do I understand min/max/bucket count correctly to think 1/1/1 are appropriate
values for this?)

[hbos, 2016-02-22 20:30:18]

hbos@webrtc.org changed reviewers:
+ rkaplow@chromium.org

[hbos, 2016-02-22 20:30:19]

+rkaplow for finch ambassador-ness (note: this is the webrtc repo with webrtc
macros for the histograms)

[stefan-webrtc, 2016-02-22 20:54:54]

My understanding of RTC_HISTOGRAM_COUNTS suggests that what you are doing is
correct. Is your plan to compare number of inits with number of errors? In that
case maybe it's better to use RTC_HISTOGRAM_PERCENTAGE?

[hbos, 2016-02-22 21:04:52]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

All stats are counts, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 init
and 1 error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Init: Occurs at InitEncode.
- WebRTC.Video.H264EncoderImpl.Error: Occurs if any type of
  error occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Init: Occurs at InitDecode.
- WebRTC.Video.H264DecoderImpl.Error: Occurs if any type of
  error occurs during initialization, AVGetBuffer2 or
  decoding.

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

All stats are counts, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 init
and 1 error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Init: Occurs at InitEncode.
- WebRTC.Video.H264EncoderImpl.Error: Occurs if any type of
  error occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Init: Occurs at InitDecode.
- WebRTC.Video.H264DecoderImpl.Error: Occurs if any type
  of error occurs during initialization, AVGetBuffer2 or
  decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

[hbos, 2016-02-22 21:10:31]

On 2016/02/22 20:54:54, stefan-webrtc (holmer) wrote:
> My understanding of RTC_HISTOGRAM_COUNTS suggests that what you are doing is
> correct. Is your plan to compare number of inits with number of errors? In
that
> case maybe it's better to use RTC_HISTOGRAM_PERCENTAGE?

Yes I do want to compare the number of inits with errors. However I want to
compare this across clients. A single client/session may either succeed or fail
to do encoding/decoding, i.e. likely have 100% success or 0% success. So if I
understand this correctly RTC_HISTOGRAM_COUNTS makes more sense and figure out
the % based on total number of inits and total number of errors across all
usages. Plus, total numbers might be nice.

[hbos, 2016-02-23 08:26:13]

PTAL stefan, hlundin. Want this in before branch cut (i.e. before roll on
thursday).

[stefan-webrtc, 2016-02-23 08:38:51]

On 2016/02/22 21:10:31, hbos wrote:
> On 2016/02/22 20:54:54, stefan-webrtc (holmer) wrote:
> > My understanding of RTC_HISTOGRAM_COUNTS suggests that what you are doing is
> > correct. Is your plan to compare number of inits with number of errors? In
> that
> > case maybe it's better to use RTC_HISTOGRAM_PERCENTAGE?
> 
> Yes I do want to compare the number of inits with errors. However I want to
> compare this across clients. A single client/session may either succeed or
fail
> to do encoding/decoding, i.e. likely have 100% success or 0% success. So if I
> understand this correctly RTC_HISTOGRAM_COUNTS makes more sense and figure out
> the % based on total number of inits and total number of errors across all
> usages. Plus, total numbers might be nice.

Yes, you are right. lgtm

[hbos, 2016-02-23 08:46:31]

Sweet. Looksies, Other Henrik? :}

[hbos, 2016-02-23 09:48:20]

Oh btw, do you know if the sibling CL that modifies chromium's histograms.xml
(https://codereview.chromium.org/1719273002/) has to land before or after this
CL or if it doesn't matter?

[hbos, 2016-02-23 09:48:51]

The CQ bit was checked by hbos@webrtc.org to run a CQ dry run

[commit-bot: I haz the power, 2016-02-23 09:48:57]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1716173002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1716173002/1

[stefan-webrtc, 2016-02-23 10:05:08]

On 2016/02/23 09:48:20, hbos wrote:
> Oh btw, do you know if the sibling CL that modifies chromium's histograms.xml
> (https://codereview.chromium.org/1719273002/) has to land before or after this
> CL or if it doesn't matter?

Doesn't matter.

[commit-bot: I haz the power, 2016-02-23 11:49:12]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-02-23 11:49:12]

Dry run: Try jobs failed on following builders:
  mac_baremetal on tryserver.webrtc (JOB_TIMED_OUT, no build URL)

[hlundin-webrtc, 2016-02-23 13:55:54]

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:374:
RTC_HISTOGRAM_COUNTS("WebRTC.Video.H264DecoderImpl.Init",
Did you consider using RTC_HISTOGRAM_ENUMERATION instead? It seems more fitting
than using a histogram with one bucket.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.h (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.h:67: void
ReportInit();
Methods should be declared before data members.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.cc:416:
RTC_HISTOGRAM_COUNTS("WebRTC.Video.H264EncoderImpl.Init",
Same again.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.h (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.h:62: void
ReportInit();
Same again.

[hbos, 2016-02-23 14:42:12]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

All stats are counts, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 init
and 1 error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Init: Occurs at InitEncode.
- WebRTC.Video.H264EncoderImpl.Error: Occurs if any type of
  error occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Init: Occurs at InitDecode.
- WebRTC.Video.H264DecoderImpl.Error: Occurs if any type
  of error occurs during initialization, AVGetBuffer2 or
  decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.State:
  * kH264EncoderStateInit: Occurs at InitEncode.
  * kH264EncoderStateError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.State:
  * kH264DecoderStateInit: Occurs at InitDecode.
  * kH264DecoderStateError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

[hbos, 2016-02-23 14:43:26]

Thanks, changed to RTC_HISTOGRAM_ENUMRATION instead. Please take another look.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:374:
RTC_HISTOGRAM_COUNTS("WebRTC.Video.H264DecoderImpl.Init",
On 2016/02/23 13:55:53, hlundin-webrtc wrote:
> Did you consider using RTC_HISTOGRAM_ENUMERATION instead? It seems more
fitting
> than using a histogram with one bucket.

Oh, that makes a lot of sense. Done.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.h (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.h:67: void
ReportInit();
On 2016/02/23 13:55:53, hlundin-webrtc wrote:
> Methods should be declared before data members.

Done.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.cc:416:
RTC_HISTOGRAM_COUNTS("WebRTC.Video.H264EncoderImpl.Init",
On 2016/02/23 13:55:54, hlundin-webrtc wrote:
> Same again.

Done.

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
File webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.h (right):

https://codereview.webrtc.org/1716173002/diff/1/webrtc/modules/video_coding/c...
webrtc/modules/video_coding/codecs/h264/h264_encoder_impl.h:62: void
ReportInit();
On 2016/02/23 13:55:54, hlundin-webrtc wrote:
> Same again.

Done.

[hbos, 2016-02-23 14:45:53]

On 2016/02/23 14:43:26, hbos wrote:
> Thanks, changed to RTC_HISTOGRAM_ENUMRATION instead. Please take another look.

Hang on I'll rename from "State" to "Event", it seems more fitting.

[hbos, 2016-02-23 14:51:21]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.State:
  * kH264EncoderStateInit: Occurs at InitEncode.
  * kH264EncoderStateError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.State:
  * kH264DecoderStateInit: Occurs at InitDecode.
  * kH264DecoderStateError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

[hbos, 2016-02-23 14:52:19]

On 2016/02/23 14:45:53, hbos wrote:
> On 2016/02/23 14:43:26, hbos wrote:
> > Thanks, changed to RTC_HISTOGRAM_ENUMRATION instead. Please take another
look.
> 
> Hang on I'll rename from "State" to "Event", it seems more fitting.

There. PTAL hlundin.

[hbos, 2016-02-23 15:01:22]

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:42:
kH264DecoderEventMax = 16,
The example I looked at was WebRTC.Video.Encoder.CodecType which had a much
higher "Max" than the number of enum entries (64). I thought the margin was in
case more entries are added in the future and also used a higher value. But
perhaps this is not necessary, and max should be either 1 or 2 depending on if
its inclusive or exclusive. Do any of you know why a higher Max value was used
in that case and if it makes sense in this case?

[hlundin-webrtc, 2016-02-24 09:17:57]

lgtm

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:42:
kH264DecoderEventMax = 16,
On 2016/02/23 15:01:22, hbos wrote:
> The example I looked at was WebRTC.Video.Encoder.CodecType which had a much
> higher "Max" than the number of enum entries (64). I thought the margin was in
> case more entries are added in the future and also used a higher value. But
> perhaps this is not necessary, and max should be either 1 or 2 depending on if
> its inclusive or exclusive. Do any of you know why a higher Max value was used
> in that case and if it makes sense in this case?

I don't know the history behind CodecType. The description in metrics.h says
"|boundary| should be above the max enumerator sample" which is interpret as
being exclusive. I don't know what happens if you later on want to add more enum
values and have to raise the boundary.

[hbos, 2016-02-24 09:54:45]

The CQ bit was checked by hbos@webrtc.org

[hbos, 2016-02-24 09:54:45]

The patchset sent to the CQ was uploaded after l-g-t-m from stefan@webrtc.org
Link to the patchset: https://codereview.webrtc.org/1716173002/#ps40001 (title:
"Renamed "...State" enums to "...Event"")

[commit-bot: I haz the power, 2016-02-24 09:55:04]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1716173002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1716173002/40001

[commit-bot: I haz the power, 2016-02-24 11:03:08]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

[commit-bot: I haz the power, 2016-02-24 11:03:09]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2016-02-24 11:03:15]

Description was changed from

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365
==========

to

==========
Histograms for H264EncoderImpl/H264DecoderImpl
initialization and errors.

The stats are counts using enumeration, an instance of
H264EncoderImpl/H264DecoderImpl will report at most 1 Init
and 1 Error for its entire lifetime. This is to avoid
spamming reports if initialization or coding fails and it
retries in a loop. The Init stats will give us an idea of
usage counts for the encoder/decoder. The Error stats will
give us an idea of how many of these usages encounters some
type of problem, such as encode or decode errors.

- WebRTC.Video.H264EncoderImpl.Event:
  * kH264EncoderEventInit: Occurs at InitEncode.
  * kH264EncoderEventError: Occurs if any type of error
    occurs during initialization or encoding.
- WebRTC.Video.H264DecoderImpl.Event:
  * kH264DecoderEventInit: Occurs at InitDecode.
  * kH264DecoderEventError: Occurs if any type of error
    occurs during initialization, AVGetBuffer2 or decoding.

Chromium sibling CL:
https://codereview.chromium.org/1719273002/

BUG=chromium:500605, chromium:468365

Committed: https://crrev.com/12f4cda0867b7264f0ad65b2b00e592de9a2a753
Cr-Commit-Position: refs/heads/master@{#11736}
==========

[commit-bot: I haz the power, 2016-02-24 11:03:16]

Patchset 3 (id:??) landed as
https://crrev.com/12f4cda0867b7264f0ad65b2b00e592de9a2a753
Cr-Commit-Position: refs/heads/master@{#11736}

[hbos, 2016-02-24 11:11:25]

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:42:
kH264DecoderEventMax = 16,
On 2016/02/24 09:17:57, hlundin-webrtc wrote:
> On 2016/02/23 15:01:22, hbos wrote:
> > The example I looked at was WebRTC.Video.Encoder.CodecType which had a much
> > higher "Max" than the number of enum entries (64). I thought the margin was
in
> > case more entries are added in the future and also used a higher value. But
> > perhaps this is not necessary, and max should be either 1 or 2 depending on
if
> > its inclusive or exclusive. Do any of you know why a higher Max value was
used
> > in that case and if it makes sense in this case?
> 
> I don't know the history behind CodecType. The description in metrics.h says
> "|boundary| should be above the max enumerator sample" which is interpret as
> being exclusive. I don't know what happens if you later on want to add more
enum
> values and have to raise the boundary.

Ok. Landed as-is. rkaplow if you know I could adjust "Max" in a follow-up if
necessary?

[rkaplow, 2016-02-24 16:10:08]

lgtm

sorry for delay

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
File webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc (right):

https://codereview.webrtc.org/1716173002/diff/40001/webrtc/modules/video_codi...
webrtc/modules/video_coding/codecs/h264/h264_decoder_impl.cc:42:
kH264DecoderEventMax = 16,
On 2016/02/24 11:11:25, hbos wrote:
> On 2016/02/24 09:17:57, hlundin-webrtc wrote:
> > On 2016/02/23 15:01:22, hbos wrote:
> > > The example I looked at was WebRTC.Video.Encoder.CodecType which had a
much
> > > higher "Max" than the number of enum entries (64). I thought the margin
was
> in
> > > case more entries are added in the future and also used a higher value.
But
> > > perhaps this is not necessary, and max should be either 1 or 2 depending
on
> if
> > > its inclusive or exclusive. Do any of you know why a higher Max value was
> used
> > > in that case and if it makes sense in this case?
> > 
> > I don't know the history behind CodecType. The description in metrics.h says
> > "|boundary| should be above the max enumerator sample" which is interpret as
> > being exclusive. I don't know what happens if you later on want to add more
> enum
> > values and have to raise the boundary.
> 
> Ok. Landed as-is. rkaplow if you know I could adjust "Max" in a follow-up if
> necessary?

usually people don't even set the numbers here manually.
It should be fine to adjust the max value, yes.