Home Forums Feature requests Generation Data Groups (I am really OLD)

Viewing 1 reply thread
  • Author
    Posts
    • inf2know
      Participant
      Post count: 17

      Way back when, in a galaxy far far away was a technology called a mainframe.
      In 1969 the "evil god" (IBM) created the "ultimate" system called the 360.
      One of the early challenges for the business world (who else could afford the $millions) was managing files.
      When disk finally arrived on the scene (will not embarrass anyone about the size and cost) managing files quickly became a MAJOR issue. Tape did not pose a problem because the "contents" of the tape was written on a sticky paper called an external label. This "label" was the early beginning of meta data (but we did not know we were creating meta data)

      When our files found their way to disk, if was not possible to put a sticky label on each file. (Side note: with removable disks, the label was attached to the disk shell)

      IBM developed a concept know as Generation data groups. A human would define a base name (c:myfilesmyfile.txt) and the number of generations desired (1-999). then the OS would "build names" based on the following naming technique.
      c:myfilesmyfile.txt.001
      c:myfilesmyfile.txt.002
      c:myfilesmyfile.txt.003 etc
      They were "created" by specifying a "new" file like c:myfilesmyfile.txt(+1) and
      "retrieved" by specifying c:myfilesmyfile.txt(0) for the current or c:myfilesmyfile.txt(-1) for the next oldest.

      This was a very usable technique (if generations to "keep"were N then ONLY N versions were kept. when N+1 was created then the oldest was deleted.)
      One might see :
      c:myfilesmyfile.txt.126
      c:myfilesmyfile.txt.127
      c:myfilesmyfile.txt.128

      This was a great technique for "daily and "weekly" executions.

      One would retrieve ALL the files for reading by specifying myfilesmyfile.txt (no suffix) which would cause ALL the files to be read.
      So we would typically "read" all the files created daily (or more frequently) without ever knowing the exact filename OR how many files were actually created.
      After the weekly file creation was SUCCESSFULLY created and checked then all of the old versions were deleted.

      This technique was very usable until it became obvious to the company I worked for that we loved the technique but needed more information in the file name.
      I personally developed an extension of the process that COULD be included in the windows space if I was a current developer working for Microsoft.

      Personally I would love to be able to create files names like:
      c:programfilesmyprog.YYMMDD_HHMMSS.exe and have the OS "manage" them
      The naming convention I currently use is to "add" either a datestamp or version number to most files.
      It is my personal "solution" to version control and works well, but is way to manual
      At least the time stamps contain the create modify and last use date, but I would LOVE to add a use count!

      My most interesting observation from the OLD generation is that with computers today, there is NO NEW PROBLEMS.
      Most (if not all) problems were "discovered" in the past mainframe days and solutions were created. In fact it was the problems and solutions that created an entire industry called "ISV" Independent Software Vendor. But many of us felt it meant IBM software vendor. Today we have ASV, MSV and USV (Apple, Microsoft and Unix(all flavors).

      Tabbles is a partial solution to the HUGE file management problem that has exploded with the challenge of unlimited and free data storage. (not literally but almost free). It is unlimited (relatively) as I just add 1TB (soon 1PB) external drives to USB ports. I have an extensive naming convention but have been relying mor and more on search because the NEED for Tabbles has not been address by Microsoft.

      Please excuse the length of this post but storage management has been my passion since 1969 and that makes me OLD.
      What concerns me is that the problems and solutions that we encountered are being re discovered and solved without knowledge of the "history" of the solutions.

      Many of the "system" type apps created for windows are solutions to "new" problems that are not new. I LOVE what I see with Tabbles and hope you go a long way!

      Keep up the GREAT work!
      Bob Levy
      boblevy@inf2know.com or
      boblevy@storageman.com

    • Andrea
      Keymaster
      Post count: 961

      Dear Bob,

      thanks a LOT for your post: sharing your experiences from the dark ages of IT (… 😀 …) means a lot to us as we realize that there is still so much we can do with Tabbles and probably most of the useful stuss is just 5 lines of code away!

      Particularly interesting about our post:

      1)
      Personally I would love to be able to create files names like:
      c:programfilesmyprog.YYMMDD_HHMMSS.exe and have the OS "manage" them
      The naming convention I currently use is to "add" either a datestamp or version number to most files.

      in the very beginning of Tabbles we considered the options of tagging files by renaming them and then search the files based on the name. This would have resulted in having file names like: invoices_gas_2009.pdf or HolidayInGreece_Mark_Deborah_Beach_Island_2008.jpg. We discarded the ida quickly as it would have created more problems than solutions. On top of it we believe that naming files is deviation from real life: you don’t name each of the dishes and chairs you own, but you can still distinguish them one another and find them when you need them… this is the goal the we set when we began working on Tabbles: surpass the file and folder names and replace them with visual tags.

      2)

      It is my personal "solution" to version control and works well, but is way to manual
      At least the time stamps contain the create modify and last use date, but I would LOVE to add a use count!

      Improving the way we sort files (by last open/modified/created) would not be too complicated for us. Given that we are in reality building a sort virtual file system, even adding some other options would be possible.
      And versioning is something we are already thinking about: as long as versioning would be "manual" (the user tells the software to go up one version for that file), it would be technically easy to implement. The reason why we didn’t do it yet is that we feel that we lack a good inspiration on GUI for the versioning…any suggestions here would be very welcome! 🙂

      Bob, please keep the comments coming! You might also consider replacing the word "old" with "experienced" 🙂

      Best regards,

      Andrea

Viewing 1 reply thread
  • You must be logged in to reply to this topic.