Nobody uses folders!



  • I had an interesting conversation with a co-worker the other day. I wrote a document repository for our company which we have started using to organize all our courses online. It is setup to resemble your typical file system. I suggested setting up a folder for each language then within those a folder for each course, subfolders within for assignments, quizzes, reading material, etc. Seemed like pretty typical/logical setup. After about 2 months into the project the person assigned to this project (in which he was already behind schedule) came to me with what I thought was a pretty novel idea – ‘Get rid of all folders and put all the files into one folder, just include in the name what they are’. Mind you, this would be almost a thousand files. You can imagine what the listing view would look like and how it wouldn’t be the easiest thing to find particular files (not to mention trying to deal with naming standards, moving courses, downloading an entire course, etc). I didn’t quite know how to react when I said I didn’t believe it was the best idea and he stated ‘nobody uses folders anymore’. I replied ‘I believe all operating systems use them to organize files’. It started to get heated and the last thing I heard as he stormed away was ‘I don’t mean now, I mean in 10 years!’

     

    I always try to think about how my application may be used in the future and prevent it from becoming obsolete, but can’t really look that far ahead as we really need this setup soon.



  • I would be funny if an operating system didn't use folders or any kind of hierarchy.  ever time you try to browse for a file you have to wait for 10 minutes while the hard drive looks for every file...



  • Could be he was referring to metadata.  If files could have numerous attributes which tell not only their type but their project and author and so on, the need for directories would be greatly diminished.  Every OS is planning to add support for generic metadata to its filesystem, so it is likely we will have it in ten years.

    So maybe in ten years, his idea will be valid. 



  • [quote user="VGR"]

    Could be he was referring to metadata.  If files could have
    numerous attributes which tell not only their type but their project
    and author and so on, the need for directories would be greatly
    diminished.  Every OS is planning to add support for generic
    metadata to its filesystem, so it is likely we will have it in ten
    years.

    So maybe in ten years, his idea will be valid. 

    [/quote]

     
    Actually, I do have it setup already where you can tag metadata (like keywords, description, author, format, etc) to each file and folder in order to create SCORM packages for each learining object. The folder structure is mainly for browsing while the metadata stored in the db is for searching and packaging the course.



  • otherwise you just use a database to store the metadata, eh?  then it's even possible to optimize it for searching if there's the potential for more than 2 or so documents.  In that case, documents could be stored utilizing a lack of folders and a nifty NAS system for scalability.  Of course, you can still have a pretend directory structure..  using one common function and a couple attributes you could just have a bunch of index files in each of those directories that quickly and happily retrieves documents.  then every document is given a unique id or sequenced number and - bada-bing - no folders are needed for storing the files themselves.  a quick CMS would allow you to update the files and ensure the naming went well.  etc. etc. blah blah blah.

    aside to this aside, the term "generic metadata" is quite a wtf in and of itself...



  • gosh yer quick!



  • [quote user="pauluskc"]

    otherwise you just use a database to store the metadata, eh?  then it's even possible to optimize it for searching if there's the potential for more than 2 or so documents.  In that case, documents could be stored utilizing a lack of folders and a nifty NAS system for scalability.  Of course, you can still have a pretend directory structure..  using one common function and a couple attributes you could just have a bunch of index files in each of those directories that quickly and happily retrieves documents.  then every document is given a unique id or sequenced number and - bada-bing - no folders are needed for storing the files themselves.  a quick CMS would allow you to update the files and ensure the naming went well.  etc. etc. blah blah blah.

    aside to this aside, the term "generic metadata" is quite a wtf in and of itself...

    [/quote]

     I have flip-flopped a few times on the sequncial numbering approach when storing files on the server for display on a website. On one hand it makes things like moving files/folders much easier as you are just updating a 'parent_id' sort of thing and not actually doing a file/directory operation but it can be more work when you want to allow people to download the files or create a zip file of an entire folder (since you need to rename the files on the server first). I guess it just depends on the project.
     



  • [quote user="VGR"]

    Could be he was referring to metadata.  If files could have
    numerous attributes which tell not only their type but their project
    and author and so on, the need for directories would be greatly
    diminished.  Every OS is planning to add support for generic
    metadata to its filesystem, so it is likely we will have it in ten
    years.

    So maybe in ten years, his idea will be valid. 

    [/quote]

     
    If the metadata is used to present a hierarchical listing of the files, much in the vein of a filesystem of today perhaps, does it really matter if they're really directories or just metadata? So no, his idea isn't likely to ever be "valid," though it someday may not be "completely retarded." 



  • only more work when you setup the function the first time.  then it's no work.  you can use memory for a temporary storage solution (if the files aren't prohibitively huge) and then just copy the files by number into appropriately named files (retrieved from DB which organizes things) in memory and then zip that and start the download..  then remember to free the memory (no leakies here).  php even has a function for doing this task:  http://us3.php.net/manual/en/function.ziparchive-addfile.php

    plus, using numbers for filenames can help make logging and statistics much faster and scalable as you only have to duplicate a number (which stores and compares reliably, efficiently, and quickly) vs trying to work with strings which could contain funky chars that aren't translated right at the wrong time, etc.



  • OooOooOooooo...SCORM :)

    This isn't Microsoft Class Server (or whatever they call it now), is it?

    -cw



  • No, no, he didn't mean "In 10 years," he meant "20 years ago, and only on limited floppy-disc-only systems".

     (MS-DOS 1.0 did not have folders, Mac System 1.0 had them, but they were implemented as a file attribute, thus allowing only 1 level of folders)
     



  • [quote user="CodeWhisperer"]

    OooOooOooooo...SCORM :)

    This isn't Microsoft Class Server (or whatever they call it now), is it?

    -cw

    [/quote]

    Nope - we use Moodle for content delivery but users can package our content into a SCORM complient format to search on and use in other course delivery software like Blackboard or WebCT (I guess they are one in the same now though aren't they?). Never heard of Microsoft having a course delivery software.



  • [quote user="HitScan"][quote user="VGR"]

    Could be he was referring to metadata.  If files could have
    numerous attributes which tell not only their type but their project
    and author and so on, the need for directories would be greatly
    diminished.  Every OS is planning to add support for generic
    metadata to its filesystem, so it is likely we will have it in ten
    years.

    So maybe in ten years, his idea will be valid. 

    [/quote]

     
    If the metadata is used to present a hierarchical listing of the files, much in the vein of a filesystem of today perhaps, does it really matter if they're really directories or just metadata? So no, his idea isn't likely to ever be "valid," though it someday may not be "completely retarded." 

    [/quote]

    What's more, you can't abandon folders until you've got that metadata-based heirarchy set up.  By way of comparison, you wouldn't sell your car today, because you think hybrids will be affordable in 10 years.   You still need a way to get to work in the morning.

    Folders offer heirarchical organization.  If that's what you need, but you refuse to use folders, you're just going to reinvent the wheel.  Poorly.
     



  • [quote user="VGR"]

    Could be he was referring to metadata.  If files could have numerous attributes which tell not only their type but their project and author and so on, the need for directories would be greatly diminished.  Every OS is planning to add support for generic metadata to its filesystem, so it is likely we will have it in ten years.

    So maybe in ten years, his idea will be valid. 

    [/quote]

    That's the beauty of his idea. He's like sooo far ahead of us all that he realizes we have metadata support now! Just put the metadata in the file name! And let users append text to any file name too. Bingo! You got tagging!

    Then its just a simple matter of getting users to write regular expressions to search for files.

    Truly visionary.

    --Rank



  • [quote user="VGR"]

    Could be he was referring to metadata.  If files could have
    numerous attributes which tell not only their type but their project
    and author and so on, the need for directories would be greatly
    diminished.  Every OS is planning to add support for generic
    metadata to its filesystem, so it is likely we will have it in ten
    years.

    So maybe in ten years, his idea will be valid. 

    [/quote]

     I strongly doubt it. The price of entry for metadata is too high for me to see it every being a replacement for directories in a multi-user system.  For one educated, organized user, it's a valid option now. I've seen attempts to implement several metadata systems across a group of users, and it always comes down to one compound question, with no right answers.

    1. Do you require metadata?

    No: Given the option to not spend a few minutes filling out metadata when creating a file, 99.9% of people will skip the metadata, thus making the system useless.

    Yes: Everyone bitches that it takes too long to enter files and no on wants to use the system. If your lucky enough to have an essential system that people are forced to use, go to question 2.

    2. Do you manage the metadata options? 

    No: Even with case insensitivity and matching tricks, the data becomes horribly fragmented and unusable.

    Yes: Everyone bitches even more about how they need new options. Ignorance of synonyms, and laziness still leads to fragmentation of data and bloated drop down menus.

    People want a metadata system that creates order out of chaos with 100% accuracy, and I just haven't seen that, and don't expect to.

     


Log in to reply