• Hannes (unregistered)

    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Scan it to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.

  • Toga (unregistered)

    What the frist? The date? What's the date?

  • that sinking feeling (unregistered)
    the OCR software was a significant expense
    Ha! Sunk costs are such fun. Gotta use stuff otherwise WTF did I pay for?
  • (cs) in reply to Hannes
    Hannes:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Scan it to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.
    You forgot the wooden table.
  • (cs)

    On the bright side, since the diffs don't need any thought to resolve the sysadmin should be able to write a script to do the whole process. And if the OCR has enough options that it can be made to run for 5 minutes (in the name of getting a high-quality output, of course) then that's a good excuse to get a coffee.

    PS A proper WTF without nonsense padding! Good to see that snoofle is still doing some of the write-ups.

  • giammin (unregistered)
    The System:
    The manager wanted to make sure that nobody changed the instructions after they were posted, so he had the developers write up the script text, verify that it worked, then take a screen shot of it and put the image of the script on the wiki page.

    Don't get why an image cannot be changed in a wiki and text can...

  • (cs) in reply to giammin
    giammin:
    The System:
    The manager wanted to make sure that nobody changed the instructions after they were posted, so he had the developers write up the script text, verify that it worked, then take a screen shot of it and put the image of the script on the wiki page.

    Don't get why an image cannot be changed in a wiki and text can...

    Freight sect.

  • Ziplodocus (unregistered)

    So, hold on.

    They are creating code and storing it in source control. They are then storing the code in a wiki in image format?

    To make code changes, they are then running an OCR on the image of the code and then loading that into the solution?

    That can't be correct, please somebody tell me I have got that wrong...

  • ANON (unregistered) in reply to giammin
    giammin:
    The System:
    The manager wanted to make sure that nobody changed the instructions after they were posted, so he had the developers write up the script text, verify that it worked, then take a screen shot of it and put the image of the script on the wiki page.

    Don't get why an image cannot be changed in a wiki and text can...

    I don't even get why you are using a wiki in the first place when you don't want the content to be changed.

  • Miriam (unregistered) in reply to Steve The Cynic
    Steve The Cynic:
    Hanzo:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Photograph the printout on a wooden table with a polaroid camera, scan the photograph to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.
    You forgot the wooden table.
    Fixed! Everything would be a lot easier if the developers just faxed the necessary build commands to the SA and the manager, though!
  • Snowman25 (unregistered) in reply to Ziplodocus
    Ziplodocus:
    So, hold on.

    They are creating code and storing it in source control. They are then storing the code in a wiki in image format?

    To make code changes, they are then running an OCR on the image of the code and then loading that into the solution?

    That can't be correct, please somebody tell me I have got that wrong...

    Almost right. It's not the code that is written in the wiki. It's the changes and commands to build the application.

  • (cs)

    Man, what a POS of deployment process. Anyway, in the article it's explained that the OCR runs many times doing a diff against the code that can be changed during this process and since the OCR doesn't have any sort of AI, it would continue without knowing that the code has changed.

    Anyway, last time I used an OCR software was in 1996 and it was crap but got most of the job done. Not that I would put my career on its hands, but it worked.

    Now, what was the manager thinking? I mean, OCR? Really? Didn't some think: there's got to be a better way. Maybe a plugin for the Wiki which wouldn't allow modification? Ugh!

  • D. Vader (unregistered) in reply to Miriam

    Don't be too proud of this technological terror you've constructed. The ability to destroy a server is insignificant next to the power of the Force

    Miriam:
    Steve The Cynic:
    Hanzo:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Photograph the printout on a wooden table with a polaroid camera, scan the photograph to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.
    You forgot the wooden table.
    Fixed! Everything would be a lot easier if the developers just faxed the necessary build commands to the SA and the manager, though!
  • (cs) in reply to Snowman25
    Snowman25:
    commands to build the application.

    There's your first WTF. If I have to write nothing more than "$./deploy" something is broken in my deployment process.

  • (cs)
    Manager:
    Because the OCR software was a significant expense and is part of the procedure!

    This is the problem: someone in a management position who is too proud to admit that they made the wrong call. Someone like that, high enough, can wreak havoc on any company.

  • Gyro (unregistered) in reply to ubersoldat
    ubersoldat:
    There's your first WTF. If I have to write nothing more than "$./deploy" something is broken in my deployment process.
    Pfft. You don't have a crew to carry out your spoken orders? Deploy the weapon. Kill everything on that ship!
  • That admin guy (unregistered) in reply to ubersoldat
    ubersoldat:
    Maybe a plugin for the Wiki which wouldn't allow modification? Ugh!

    You mean like locking the page?

  • That admin guy (unregistered)

    TRWTF is that the Sys Admin didn't just use the scripts from source control and lie to the boss about running the OCR software. Either he is some kind of brown-noser or he believes in being passive aggressive by way of malicious compliance. In both cases, he should be fired.

  • Hannes (unregistered) in reply to Miriam
    Miriam:
    Steve The Cynic:
    Hanzo:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Photograph the printout on a wooden table with a polaroid camera, scan the photograph to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.
    You forgot the wooden table.
    Fixed! Everything would be a lot easier if the developers just faxed the necessary build commands to the SA and the manager, though!

    Wow. Be careful with who you call "Hanzo". >:(

  • RealUlli (unregistered) in reply to That admin guy
    That admin guy:
    TRWTF is that the Sys Admin didn't just use the scripts from source control and lie to the boss about running the OCR software. Either he is some kind of brown-noser or he believes in being passive aggressive by way of malicious compliance. In both cases, he should be fired.
    Changes should always be possible in such a situation. It's not that the page needs to be locked, it's that the changes need to be tracked to a specific user. That way, someone making changes in places where he's not supposed to change something will have to answer pointed questions...

    Anyway, this is TDWTF, not some help forum, what am I doing here... ;-)

  • (cs) in reply to ANON
    ANON:
    giammin:
    The System:
    The manager wanted to make sure that nobody changed the instructions after they were posted, so he had the developers write up the script text, verify that it worked, then take a screen shot of it and put the image of the script on the wiki page.

    Don't get why an image cannot be changed in a wiki and text can...

    I don't even get why you are using a wiki in the first place when you don't want the content to be changed.

    Because the wiki software was a significant expense and is part of the procedure! Duh.

  • Smug Unix User (unregistered)

    Why not use the "script" command to log all the actual commands that went into the box and then use curl to PUT the log to the wiki?

  • Ziplodocus (unregistered) in reply to That admin guy
    That admin guy:
    TRWTF is that the Sys Admin didn't just use the scripts from source control and lie to the boss about running the OCR software. Either he is some kind of brown-noser or he believes in being passive aggressive by way of malicious compliance. In both cases, he should be fired.

    Pfft, what he should have done is suggest using the OCR in another process, thus DOUBLING its usefulness.

    You'll never get anywhere in business with your attitude.

  • Admiral Motti (unregistered) in reply to D. Vader
    D. Vader:
    Don't be too proud of this technological terror you've constructed. The ability to destroy a server is insignificant next to the power of the Force

    Don't try to frighten us with your sorcerer's ways, Lord Vader. Your sad devotion to that ancient Jedi religion has not helped you fix up the crappy deployment process, or given you enough clairvoyance to find the flaw in it.

  • (cs)

    TRWFT is OCR software that can't handle fresh computer-printed text output-- probably even a monospaced font.

  • Miriam (unregistered) in reply to operagost
    operagost:
    TRWFT is OCR software that can't handle fresh computer-printed text output-- probably even a monospaced font.
    Knowing the manager, he probably spent lots of money on an expensive super-stylish design font that is non-proportional and based equally on Papyrus, Comic Sans and Wingdings, which he of course uses in the wiki. It was a significant expense, you know.
  • Erroneously suicidal. (unregistered)

    "What do you mean it was all a joke?? I DIDN'T lose me wife, kids, job, house and car?

    Well I've bought all the cyanide now, so it'd be a shame not to use it."

  • Rodnas (unregistered) in reply to Admiral Motti
    Admiral Motti:
    D. Vader:
    Don't be too proud of this technological terror you've constructed. The ability to destroy a server is insignificant next to the power of the Force

    Don't try to frighten us with your sorcerer's ways, Lord Vader. Your sad devotion to that ancient Jedi religion has not helped you fix up the crappy deployment process, or given you enough clairvoyance to find the flaw in it.

    I find your lack of faith disturbing

  • eman_ruoy (unregistered)

    And a simple copy/paste of the command line to the wiki and a simple copy/paste from the wiki to the command line of the next environment is not good enough?

    I realize that unauthorized changes are scary, but wiki's have a "blame" feature, too and/or can be rolled back/changed to the correct version.

    I agree with publicly mocking this Rube Goldberg solution.

  • Not Hans (unregistered) in reply to Hannes
    Hannes:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Scan it to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.

    You forgot to take a picture of it on a wooden table.

  • Nagesh (unregistered)

    Unfortunately, Managers are always proud of the mess they create.

  • (cs) in reply to giammin
    giammin:
    Don't get why an image cannot be changed in a wiki and text can...

    Because $bossman knows how to use word he knows that text can be edited. $bossman knows that images can't be changed (or, alternately, knows how to use MS Paint, and knows that changes would be obvious).

    I see that at my job... all job quotations are sent to the customer in .pdf format "because then they can't change anything". Apparently the format stands for Permanent Document Format.

  • evilspoons (unregistered)

    This has got to be the dumbest thing I've read on this site in a long time. Holy crap.

  • Blackmage (unregistered) in reply to Ziplodocus

    Should have used the OCR on the source code for the diff.

  • Neil (unregistered) in reply to Admiral Motti
    Admiral Motti:
    D. Vader:
    Don't be too proud of this technological terror you've constructed. The ability to destroy a server is insignificant next to the power of the Force.
    Don't try to frighten us with your top-posting ways, Lord Vader.
    FTFY
  • C-Derb (unregistered)

    TRWTF is knowing you are using OCR in your "release process", but still calling a consultant to figure out why said process does not work.

  • Hasse de great (unregistered) in reply to Hannes
    Hannes:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Scan it to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.
    where is the wooden table?
  • Does That Work? (unregistered)
    The manager wanted to make sure that nobody changed the instructions after they were posted, so he had the developers write up the script text, verify that it worked, then take a screen shot of it and put the image of the script on the wiki page.
    When the scripts are more than fit on a single screen, do they enter multiple screen shots into the wiki? Then OCR multiple images and combine then, without repeated scripts and no missed lines?
  • anonymous (unregistered) in reply to C-Derb
    C-Derb:
    TRWTF is knowing you are using OCR in your "release process", but still calling a consultant to figure out why said process does not work.
    TRWTF is not even supposing that OCR software would misread some text.

    Even further, what kind of incompetent moron is the SA, that never noticed wrong keywords in those scripts? I bet that more than once at least some SQL script got a DRUP command or something. Where does he shove the output logs?

  • (cs) in reply to anonymous
    anonymous:
    TRWTF is not even supposing that OCR software would misread some text.

    But there's an easy fix for that: blow the image up 5X before posting to the wiki. Always easier to read superlargefont text!

  • (cs)

    All of this reminds me of an (over 15 years ago) episode of "This Old House" (American public TV). They went to a design studio and entered the design into some cad program, and plotted it out (nice sized sheets). Then because the data was needed as input to another program, they used an absolute positioning mouse (graticule, etc.) to re-enter the data into the second program. A wonderful manual step (probably prone to error) just because someone didn't know the export/import commands to the programs. Of course they might not have existed, but really?

    No, it wasn't a wooden table, it was a drafting board.

  • secundum (unregistered) in reply to herby
    herby:
    No, it wasn't a wooden table, it was a drafting board.
    I thought they did away with the drafting board after Vietnam?
  • Calli Arcale (unregistered)

    Having worked in CM for about ten years now, the words that went through my head as I read this story was "Oh, lord...."

    I have encountered my share people who just don't trust the system well enough to let it actually enforce what it's designed to enforce, but this is a whole new level of that.

  • Bananas (unregistered) in reply to evilspoons
    evilspoons:
    This has got to be the dumbest thing I've read on this site in a *long* time. Holy crap.
    And not a moment too soon. I was despairing of ever seeing a real WTF on this site ever again.
  • Chelloveck (unregistered)

    A similar situation happened to me in reverse. I was doing development for an embedded device, and the docs people needed to illustrate the dot-matrix display. So they meticulously tried to recreate it using Word or Photoshop or whatever it is tech writers like to use, but of course it looked nothing like the actual display because they didn't have the custom fonts, icons, etc. When I first saw a draft of the manual my jaw dropped; it had a dozen or so of these. "You know, I can just dump the display RAM and hand you a pixel-perfect BMP, or GIF, or whatever format you want. Takes me all of 30 seconds instead of hours for you to do it by hand." Fortunately they were smart enough not to let the "procedure" get in the way.

  • garaden (unregistered) in reply to Hannes
    Hannes:
    There has to be a better solution: Frist, create script in test environment. Then, print it to paper. Scan it to PDF, extract the PDF pages as image, run OCR on those images, diff the result with the script. If there are no differences, repeat the process again, but this time put the OCR result in production.

    Reminds me of a Frontalot song:

    "Get your most closely kept personal thought: put it in the Word .doc with a password lock. Stock it deep in the .rar with extraction precluded by the ludicrous length and the strength of a reputedly dictionary-attack-proof string of characters (this, imperative to thwart all the disparagers of privacy: the NSA and Homeland S). You better PGP the .rar because so far they ain’t impressed. You better take the .pgp and print the hex of it out, scan that into a TIFF..."

  • garaden (unregistered) in reply to garaden

    Also this WTF is awesome. Bravo.

  • (cs) in reply to herby
    herby:
    All of this reminds me of an (over 15 years ago) episode of "This Old House" (American public TV). They went to a design studio and entered the design into some cad program, and plotted it out (nice sized sheets). Then because the data was needed as input to another program, they used an absolute positioning mouse (graticule, etc.) to re-enter the data into the second program. A wonderful manual step (probably prone to error) just because someone didn't know the export/import commands to the programs. Of course they might not have existed, but really?

    No, it wasn't a wooden table, it was a drafting board.

    Norm and all the old guys? Yeah, I could see it. They wouldn't be such amateurs on Hometime, though.

  • (cs)

    OCR lands somewhere between PHP and Java on the WTF scale.

  • n_slash_a (unregistered) in reply to ubersoldat
    ubersoldat:
    Snowman25:
    commands to build the application.

    There's your first WTF. If I have to write nothing more than "$./deploy" something is broken in my deployment process.

    This. Right here. Maybe put "deploy qa" or "deploy prod" if you have specific servers or run-time checks you don't want active, but never the entire command. I know that our build command is huge, several dozen parameters, but all that is scripted and we type 2 or 3 into the command line.

Leave a comment on “The System”

Log In or post as a guest

Replying to comment #435772:

« Return to Article