[RFC] Get rid of frames-(m)sec conversion #13288

Voyager1 · 2018-01-02T10:19:55Z

Description

A relatively small change to isolate the repeated conversions between frames (aka 1/75th of a second) and seconds or milliseconds. CFileItem's startOffset and endOffset are expressed in "frames" so there's repeated conversion going on in several places. Plus having the conversion all over the place is error prone.
UPDATE
I've added code to remove the frames conversion and put all offsets in milliseconds. At first sight the code seems to work fine.

Note: the placement of the helper functions is currently CUtil but it's up for comments.

Motivation and Context

see above

How Has This Been Tested?

Code compiles. Some runtime testing done but needs more testing.

Screenshots (if appropriate):

Types of change

Bug fix (non-breaking change which fixes an issue)
Improvement (non-breaking change which improves existing functionality)
New feature (non-breaking change which adds functionality)
Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

My code follows the Code guidelines of this project
My change requires a change to the documentation, either Doxygen or wiki
I have updated the documentation accordingly
I have read the CONTRIBUTING document
I have added tests to cover my change
All new and existing tests passed

xbmc/Util.h

@@ -204,6 +204,13 @@ class CUtil
   */
  static int GetRandomNumber();

+  static int ConvertSecsToOffset(double secs) { return static_cast<int>(secs * 75); }
+  static double ConvertOffsetToSecs(int offset) { return offset / 75.0; }


MartijnKaijser · 2018-01-02T11:41:59Z

jenkins build this please

MartijnKaijser · 2018-01-02T14:06:17Z

Let's try again now.
jenkins build this please

xbmc/CueDocument.cpp

@@ -456,7 +457,7 @@ int CCueDocument::ExtractTimeFromIndex(const std::string &index)
  int secs = atoi(time[1].c_str());
  int frames = atoi(time[2].c_str());

-  return (mins*60 + secs)*75 + frames;
+  return CUtil::ConvertSecsToOffset(mins*60 + secs) + frames;


Voyager1 · 2018-01-03T17:12:27Z

@notspiff @FernetMenta the "weird" unit of 75 frames comes from the Cue-sheet standard: see https://en.wikipedia.org/wiki/Cue_sheet_(computing) .
I have added two commits to completely get rid of that unit (except when reading cue-sheets of course) so now start and end offsets are in milliseconds. That saves us a whole bunch of conversion. The only conversion still going on is between seconds (double) and milliseconds (int64_t).

Voyager1 · 2018-01-03T17:21:15Z

jenkins build this please

Voyager1 · 2018-01-03T18:17:09Z

jenkins build this please

Voyager1 · 2018-01-03T19:22:37Z

once again, jenkins build this please

Voyager1 · 2018-01-03T20:41:41Z

at last. Jenkins is good - just the usual OSX test failure...

FernetMenta · 2018-01-04T09:00:29Z

@Voyager1 nice!

ace20022 · 2018-01-04T11:04:50Z

Thanks, looks good at a glance.

DaveTBlake · 2018-01-04T15:37:04Z

I've added code to remove the frames conversion and put all offsets in milliseconds.

Does that include the offset values held in the music DB? If not would it not be best to convert cuesheet values when originally read and saved msec values in the library too? But that would ential converting existing song data and a db bump. I will take that as a separate PR if liked.

Voyager1 · 2018-01-04T17:37:08Z

Does that include the offset values held in the music DB

ha - I did not consider the existing ones. As of right now, they are in "frames" but new data will be stored in milliseconds. So existing data will be read as milliseconds thus will be wrong. Another option would be to store in db as "frames", ie. do a conversion upon read/write. But I like your suggestion (much cleaner IMO) and if you feel up to doing the db bump, that would be great!

Voyager1 · 2018-01-05T17:21:24Z

@DaveTBlake I just want to clarify that three (not two) columns in the Song table will now be interpreted as milliseconds: iDuration, iStartOffset, iEndOffset.
UPDATE: this statement is incorrect. iDuration is actually in seconds.

DaveTBlake · 2018-01-05T17:48:44Z

I get your thinking @Voyager1 but IMO Song table iDuration needs to remain as seconds. It is something all songs have, and for human use rather than player (which does not take any notice of what the library has anyway) - think track listings on the back of a CD case - and second accuracy is enough.

Music from cuesheets is not mainstream, and the only time that iStartOffset, iEndOffset is used. Despite the odd cuesheet standard using frames, I suspect that all offsets are to second accuracy - the example cuesheets I have seem have all been like that, and it seems a natural granularity for music. But in line with use elsewhere we can store as msec, after all they have never been the same unit as iDuration

Happy to do the db bump, I can roll it into some other small changes.

Voyager1 · 2018-01-05T18:55:06Z

@DaveTBlake my bad... iDuration is (and has been) in seconds so indeed there's no change (I don't know what I was thinking!). Please go ahead with the bump. Next to the version increase, I think you'll just need an UPDATE SQL statement to multiply all iStartOffset and iEndOffset * 1000 / 75 (or * 40/3)

…ths of sec) and seconds/milliseconds

…liseconds

…sion

Voyager1 · 2018-01-09T19:50:12Z

@DaveTBlake - since I didn't hear back I pushed an additional commit for the db bump. This way we have a complete PR that can be merged and tested.

Voyager1 · 2018-01-09T21:33:57Z

jenkins build this please

DaveTBlake · 2018-01-10T07:25:36Z

Thanks for that @Voyager1, just not sure what I didn't reply to?

Voyager1 · 2018-01-10T09:00:43Z

just not sure what I didn't reply to

you may have missed my message on slack. Never mind...

popcornmix · 2018-01-10T15:12:26Z

Kodi no longer builds for me after this PR.
PRIi64 does not get defined by inttypes.h with gcc 4.9.3.
If I replace PRIi64 with PRId64 it builds again.

Voyager1 · 2018-01-10T15:43:09Z

If I replace PRIi64 with PRId64 it builds again.

weird because jenkins had built it without issues. That said I think it would be ok to use PRId64 instead because it's the same for an output format specifier. Feel free to PR the fix. thanks!

notspiff · 2018-01-10T15:49:11Z

it's likely due to interaction with #13336 (system.h removal).

Rechi · 2018-01-10T21:19:25Z

jenkins told it works on top of #13336

…(one more 75 missed)

[video] Fix CGUIWindowVideoBase::GetResumeItemOffset after #13288 (on…

Voyager1 added Type: Cleanup non-breaking change which removes non-working or unmaintained functionality RFC PR submitted for gathering feedback labels Jan 2, 2018

Voyager1 requested review from FernetMenta and ace20022 January 2, 2018 10:19

FernetMenta reviewed Jan 2, 2018

View reviewed changes

FernetMenta approved these changes Jan 2, 2018

View reviewed changes

Voyager1 changed the title ~~Get rid of repeated constants in frames-(m)sec conversion~~ Get rid of frames-(m)sec conversion Jan 3, 2018

Voyager1 commented Jan 3, 2018

View reviewed changes

Voyager1 changed the title ~~Get rid of frames-(m)sec conversion~~ [RFC] Get rid of frames-(m)sec conversion Jan 3, 2018

Voyager1 force-pushed the offset-conversion branch from 17b77e6 to 3cf2e5f Compare January 3, 2018 17:08

ace20022 approved these changes Jan 4, 2018

View reviewed changes

Voyager1 added 5 commits January 9, 2018 20:37

Get rid of repeated constants in conversion between frames/offset (75…

5bd5ef7

…ths of sec) and seconds/milliseconds

Get rid of the old frames unit and store startOffset/endOffset in mil…

f0067b7

…liseconds

Cue sheet standard format still uses MM:SS:FF format so ensure conver…

8dd28a0

…sion

Bulk Rename Offset to MilliSecs

6efbe1d

fix missing PRIi64 macro

99435bb

update music db start and end offset from frames to msec

d1d6fc5

Voyager1 force-pushed the offset-conversion branch from a556700 to d1d6fc5 Compare January 9, 2018 19:48

Voyager1 merged commit db046cf into xbmc:master Jan 10, 2018

DaveTBlake added Component: Music v18 Leia labels Jan 10, 2018

DaveTBlake added this to the L 18.0-alpha1 milestone Jan 10, 2018

Voyager1 deleted the offset-conversion branch January 14, 2018 10:28

ksooo added a commit to ksooo/xbmc that referenced this pull request Jan 18, 2018

[video] Fix CGUIWindowVideoBase::GetResumeItemOffset after xbmc#13288 …

dbab3db

…(one more 75 missed)

ksooo mentioned this pull request Jan 18, 2018

[video] Fix CGUIWindowVideoBase::GetResumeItemOffset after #13288 (on… #13389

Merged

wiromare added a commit to wiromare/xbmc that referenced this pull request Jan 18, 2018

fix GetResumeItemOffset after PR xbmc#13288

4abfdbe

ksooo added a commit that referenced this pull request Jan 18, 2018

Merge pull request #13389 from ksooo/video-fix-getresumestring

33d85ac

[video] Fix CGUIWindowVideoBase::GetResumeItemOffset after #13288 (on…

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[RFC] Get rid of frames-(m)sec conversion #13288

[RFC] Get rid of frames-(m)sec conversion #13288

Voyager1 commented Jan 2, 2018 •

edited

Loading

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

MartijnKaijser commented Jan 2, 2018

MartijnKaijser commented Jan 2, 2018

This comment was marked as spam.

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

FernetMenta commented Jan 4, 2018

ace20022 commented Jan 4, 2018

DaveTBlake commented Jan 4, 2018

Voyager1 commented Jan 4, 2018

Voyager1 commented Jan 5, 2018 •

edited

Loading

DaveTBlake commented Jan 5, 2018

Voyager1 commented Jan 5, 2018

Voyager1 commented Jan 9, 2018

Voyager1 commented Jan 9, 2018

DaveTBlake commented Jan 10, 2018

Voyager1 commented Jan 10, 2018

popcornmix commented Jan 10, 2018

Voyager1 commented Jan 10, 2018

notspiff commented Jan 10, 2018 •

edited

Loading

Rechi commented Jan 10, 2018

[RFC] Get rid of frames-(m)sec conversion #13288

[RFC] Get rid of frames-(m)sec conversion #13288

Conversation

Voyager1 commented Jan 2, 2018 • edited Loading

Description

Motivation and Context

How Has This Been Tested?

Screenshots (if appropriate):

Types of change

Checklist:

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

This comment was marked as spam.

MartijnKaijser commented Jan 2, 2018

MartijnKaijser commented Jan 2, 2018

This comment was marked as spam.

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

Voyager1 commented Jan 3, 2018

FernetMenta commented Jan 4, 2018

ace20022 commented Jan 4, 2018

DaveTBlake commented Jan 4, 2018

Voyager1 commented Jan 4, 2018

Voyager1 commented Jan 5, 2018 • edited Loading

DaveTBlake commented Jan 5, 2018

Voyager1 commented Jan 5, 2018

Voyager1 commented Jan 9, 2018

Voyager1 commented Jan 9, 2018

DaveTBlake commented Jan 10, 2018

Voyager1 commented Jan 10, 2018

popcornmix commented Jan 10, 2018

Voyager1 commented Jan 10, 2018

notspiff commented Jan 10, 2018 • edited Loading

Rechi commented Jan 10, 2018

Voyager1 commented Jan 2, 2018 •

edited

Loading

Voyager1 commented Jan 5, 2018 •

edited

Loading

notspiff commented Jan 10, 2018 •

edited

Loading