Rel 1.2.4

[PoppyPop]

BugFix Add the same functionality for TV shows and movies (Auto-library add) Add perf optimisation options to handle huge initial library scan

[LukePulverenti]

Why switch to the model of manually setting parent ids and season ids? This is really not good. It is best to just copy the files in and let the normal library scanning process take it from there.

[PoppyPop]

Multiple reasons :

• I noticed that the refreshMetaData on series add it to library but with no parent. when you scan the library in the "normal" way, it result in a duplicate (issue #8)
• I have some trouble with filesystem modification detection i added this to optimize the process

For the solution

• I have another branch, wich provide option to activate/deactivate direct library add
• I can remove the direct add for movie and episode as the process cause timeouts (i think i will do it anyway, with time i think i take the wrong way)

Just give me the good path, but the parenting of Series is mandatory if we want to refresh metadata.

[LukePulverenti]

We shouldn't have the parenting. The primary purpose of this plugin is to be an automated alternative to manually copying a new file into a library folder. If there are any issues in that process then it is likely to be a core issue and should be reported accordingly. It's not the plugin's responsibility to try and work around it.

[PoppyPop]

I reworked the change, Removing all auto-add feature, as it was before. I keeped an parenting to not affect user and i will open an proper issue for the emby core.

For the responsibilty of the plugin, i think it's very hard to know if it's the plugin responsablity or the core for the parenting issue, the public api, provide an AddChild, wich make the job in an esay way.

[PoppyPop]

Ok, i restart from the beginning, in the AutoOrganize plugin there was a code that search for the Series, it was existing, when i add the Movie Organize function, i noticed my library was inconsistent because of duplicate. I made a temporary fix with the AddChild. I repeat Temporary.

I can remove the AddChild, but is like saying to the users, i will create duplicates. I can remove the MetadataRefresh on series, but this will lead to a remote call for each episodes, and lead to timeout and performances issues.

I repeat it a second time, i speak about existing code, that make duplicates, not a new functionality. (MediaBrowser/Emby.AutoOrganize#8) As you can see, i removed all other addChild.

[LukePulverenti]

I asked you to open a bug report in the core because I didn't catch that you are calling AddChild. After further review, I see you are still using that, and it should be removed.

[LukePulverenti]

As far as performance - if the same situation would happen when manually copying in a new video file, then yes the RefreshMetadata should be removed from the plugin.

[PoppyPop]

The same situation would not happen, it's a Series refresh and not an episode Refresh. When you copy, one or multiple episode, the series would be added one time, and each episode added and refreshed.

If i remove the existing code it will result in :

• For the auto-identification: for each episode, it will make a remotesearch of the series (the library does not contains a definition for the series, so the code do a search). This will increase the number of external request.
• For the manual choice, Users will make an proper identification of the series of one episode Now they have 2 choices :
  • wait or trigger a library scan and re-run an organization
  • Make an individual identification for each Series (and a RemoteSearch for identification)

This field will always lack the identified series until a library scan

We must find the good compromise between performances and user functionality.

• Add correct parenting when refreshing metadata (you can add an entire series in one time, with no performances issues). This lead to 1 Series RemoteCall per Series, 1 Episode RemoteCall per Episode.
• Remove adding Series to library (decrease performances, you cannot add an entire series in one time, you must scan multiple time and wait for library add between each scan) . This lead to 1 Series RemoteCall per Episode, 1 Episode RemoteCall per Episode.

For references, i checked the library scan, it make 1 Series RemoteCall per Series, 1 Episode RemoteCall per Episode.

[PoppyPop]

Response Part 2 : I tried to make a local series cache, so i can keep good performance and i do not relies on media library and parenting. The result is worse, now i understand why the plugin NEED to refresh Series metadata. The Code use providerManager.GetRemoteSearchResults<Episode, EpisodeInfo> method to obtain episode information. I lookup at the base code of TvdbEpisodeProvider and OmdbEpisodeProvider, they do not make any remote call, they relies to the metadata generated by the Series refresh.

When removing the metadata refresh, no episode can be identified, each research give no result. Obviously, the plugin become useless in this case.

I we see the Series MetaData Refresh is Mandatory, so we come back to the parenting issue.

In my vision there is 2 possible solutions :

• Add the functionality to the core (or)
  • Add a hook in the metadata refresh action to correct parenting if missing
  • Add a hook in the media scanner, that try to match an unparented item a each scan (and delete if duplicate)
• Keep the parenting in the plugin (as you say, it's a bad solution, the plugin become aware of the internal library structure)

I removed the parenting from the PR

[PoppyPop]

Hello,

Any news ? (I removed the parenting from the PR)

[LukePulverenti]

Can you check spelling on the html - comlpetion for example. Thanks.