file/directory only library option

[nosferatu_cat]

I have one of those dread plus size collections of files.  I've never used tags much except to get along with software that relies on tags for names.  Back in the time of giant ground sloths and Napster I started organizing my files by directory.  It works great for me.  Would it be easy to add an option for SMB sources that considers only directory info, i. e. no tags, no albums, etc.?  I'm thinking in terms of a standard file browser.  Seems like it would provide snappier response time for people who don't care about tags. 

Thanks

[the_bertrum]

There are many tag editor solutions that will take your folder hierarchy and tag the files based on that. Doing so would replicate your meta data from the folders to the tags so it was still in the folders for your use, but also in the tags for moOde to use.

[nosferatu_cat]

That's not the problem.  The tag database becomes a serious performance problem when the number of files reaches a certain level. It's a known problem.  In order to support tags you have to have the overhead of indices, etc.  The Raspberry Pi is not good with databases beyond a certain size.   I have enough files to have passed the critical point.  I've used tag editors.  They work.  I only use them when I'm dealing with player software that gets title/artist information from tags and ignores the file names.  I like the file/directory approach.  It has all the features I care about.  I can find anything I want quickly and get all the information I care about from the file names.  If I want discographical information I can look it up elsewhere.  Bonus, the file/directory approach requires minimal processing power.  There's some caching but storage of information is in my case in a separate location.  My files are on an NAS.  

There are many tag editor solutions that will take your folder hierarchy and tag the files based on that.  Doing so would replicate your meta data from the folders to the tags so it was still in the folders for your use, but also in the tags for moOde to use.

[Tim Curtis]

moOde and MPD are designed and optimized around tagged collections and rich metadata and not raw folders and directories. This is what the majority of moOde users want and find easiest to use.

If you want to navigate your collection via its folder tree and file names you'll need to seek out another player.

[nosferatu_cat]

Darn.  I was afraid it was going to be like that. I didn't realize MPD was designed like that.   I was hoping that the file/directory tree was independent from the database and it would be easy to separate the two.  I know some people like tags.  Years ago I had experience with trying to deal with a user maintained tag system in a retail environment.  The big problem with the tag idea is that when people are given a text field where they can enter anything, they will.  You can't have meaningful indices and queries without key integrity.  Music tags are the same way unless you get all your files from the same commercial source.  You don't want to see what my library looks like. 

moOde and MPD are designed and optimized around tagged collections and rich metadata and not raw folders and directories. This is what the majority of moOde users want and find easiest to use.

If you want to navigate your collection via its folder tree and file names you'll need to seek out another player.

[swizzle]

That's not the problem.  The tag database becomes a serious performance problem when the number of files reaches a certain level. It's a known problem.  In order to support tags you have to have the overhead of indices, etc.  The Raspberry Pi is not good with databases beyond a certain size.   I have enough files to have passed the critical point. 

I suspect it’s more a limitation of the webui than mpd as really large lists aren’t a common enough use case that anybody really optimizes for them when developing a browser. 

There are ways to limit how much of a list you have loaded into memory at any one time which would solve the memory issue but also limits some of the things we can do as the data isn’t in memory (like the quick jump alphabetical indexes or the find as you type search).

For example: https://clusterize.js.org/

[the_bertrum]

The big problem with the tag idea is that when people are given a text field where they can enter anything, they will.  You can't have meaningful indices and queries without key integrity.  Music tags are the same way unless you get all your files from the same commercial source.  You don't want to see what my library looks like. 

What you say for tags is also true of your folder hierarchy, the folder names are after all just free text and it is only because you take care to put things in the right place that the system remains in order. Take the same care over your tags, and all is well.

[nosferatu_cat]

The difference between a directory based system and a tag system is the number of sources for the information.  A directory system is for one user who's also the creator. It's consistent by definition assuming you're capable of following your own rules. If you have more than one source for files a tag system is a TEGWAR, The Exciting Game without Any Rules.  Any given tag can contain different information depending on who put the information into it.  One of the first non Codd/Date rules although they have something similar which I've forgotten, for database keys is that they be unique.  You can't have more than one key for the same concept or the system breaks. Suppose one source uses a genre key "Baroque" and another uses "17th-18th century" for the same piece? How about a composer tag as "J. S Bach" or "Bach?"   I have files from many sources, my personal music collection, friends' music collections, commercial sources...  Tagging ranges from none to complete with different sets of rules for different sources.  Without a standards body and rigorous standards enforcement, tag utility isn't nearly as good as it could be. 

The other thing is that I'm not really interested in spending the week or so it would take me to implement consistent tagging for my files. For me it has zero benefit. I'm not insisting that everyone use my approach.  It's a valid approach for me and my set of files is well beyond the critical mass that would make tagging/re-tagging practical.  I understand that the design of MPD makes skipping the tag system and the database unworkable.  I was hoping that file browsing and the tag database were independent entities and you could install MOODE without tags.  Nope, doesn't work that way.  I get it.  

The big problem with the tag idea is that when people are given a text field where they can enter anything, they will.  You can't have meaningful indices and queries without key integrity.  Music tags are the same way unless you get all your files from the same commercial source.  You don't want to see what my library looks like. 

What you say for tags is also true of your folder hierarchy, the folder names are after all just free text and it is only because you take care to put things in the right place that the system remains in order.  Take the same care over your tags, and all is well.