• LCrawford (unregistered)

    Depending on the application's needs, it can be possible to custom develop DB functionality that is better in the ways needed by that application. For example, back in the day there were few or no NoSQL models. Those are unusual cases however, and it sounds like this would not have been a good candidate for a custom DB.

    The real WTF is the architect continuing to work on a DB replacement for months after the app already had a working solution.

    And for going behind the architect's back on his frist project, how did this affect their future working relationship?

  • RLB (unregistered)

    Erm, guys... your SSL certificate is @#$&ed.

  • guest (unregistered)

    "Our manager never did do anything to stop him from wasting time. "

    I would not be surprised to find out that your manager was happy to have him working on something pointless so that he was too busy with that task to pick up other side projects that might screw things up elsewhere. It would be a good excuse "Well, I know you'd like to completely redesign the frontend on this, but you're working on that Sybase replacement and we just can't justify two side projects like that, so we'll stick with the current design for now and you can get to it once you have that Sybase replacement done".

  • MediDev (unregistered)

    "Our manager never did do anything to stop him from wasting time. "

    Think we've all seen this sort of behavior. Unfortunately, in one case I have seen he was "decoupling" the application and broke in it two - broke being the correct word. Not a thought to the serialization issues that would occur by changing all the namespaces in one half. Not a thought to the huge increase in development time as it wasn't "decoupled" correctly at all. Not a thought to the amount of regression testing that would be required. The manager just allowed this madness to be checked in and went with it.

    He has out lasted the manager... whether the company with out last him is something else.

  • Zadkiel (unregistered)

    The real WTF is having to manually re-craft every URL on the site to http because your SSL certificate is so borked i can't get any browser to accept it.

  • (nodebb)

    Did you let your SSL certificate expire again or is this an intentional test to see if we can figure out how to get around these kinds of things (level 0.5 hacking).

  • DrPepper (unregistered)

    The most valuable line of code is the code you DO NOT WRITE. Always.

  • Brian Parker (google)

    The humility to know when to use an existing product over rolling your own, or when to ask a technology specialist (the DBA) to pick up a piece you aren't as comfortable with... that's a valuable soft skill, right there.

  • Kashim (unregistered) in reply to DrPepper

    I'd say that the most valuable line of code is the code that someone else wrote for you that does exactly what you need it to do, that you can access easily and understand well enough to know that you need to use it, and how to use it.

    Whenever anyone over-designs a project like this, I always say, "Look at this language I found! It's called C, and it is infinitely extensible! Within a relatively limited period of time, you can use it to write any program that you want!" In the end, it's always about time. The IT department once asked a member of my department if he could write a new filesystem that would provide quicker and easier access for our document management system because it was so annoying working with the filesystems that already existed, and then they could write their own security and permissions. They were disheartened when he said that yes, he could do it, if they provided him a research team of 20 guys and 15 years, but that it would almost certainly come out very buggy when he was done.

  • Chronomium (unregistered) in reply to guest

    I would not be surprised to find out that your manager was happy to have him working on something pointless so that he was too busy with that task to pick up other side projects that might screw things up elsewhere. It would be a good excuse "Well, I know you'd like to completely redesign the frontend on this, but you're working on that Sybase replacement and we just can't justify two side projects like that, so we'll stick with the current design for now and you can get to it once you have that Sybase replacement done".

    This is all well and good, but maybe three years from now Overclever Architect Fellow will actually have something done and demand it get used because he put so much effort into it.

    Now maybe the ultimate plan is to take the "no and that's final" path so he quits in disgust. But you risk him winning a fight with the upper boss morass.

  • OTee (unregistered)

    Ahhh, this brings memories: As a DB and PL/SQL developer in early 2000 I was part of a team tasked with developing a content DB for mobile device media when everywhere were still just proprietary file formats and features.

    We ended up with the most horrible abomination of 'abstracted to death' tables along the line of device_atribute_type, device_atribute_value, file_atribute_type, file_atribute_value, device_file_requifement, etc. just to build the part that had to find out what version of a ringtone should be sent to a Nokia or a Siemens when receiving "Summertime 12". Not to mention the blob columns storing the actual files, preview pics, test-soundbits for the webpage.

    It was ugly as shit an slow as a stone. We even knew then but a "pointy haired boss" said it would be 'cheaper' then buying existing CMS since he "paid us anyway"

  • Steve (unregistered)

    Chronomium: that is a good guess, so here's a counterattack: allow the replacement to be used, after it passes a few carefully designed stress and scalability acceptance tests. (along with tests for all the actual features) Getting those to pass will keep the architect busy for another few years :)

  • Developer Dude (google)

    Reminds me of a job interview (whole dev team save one was interviewing me) where the subject of the C++ STL came up. The manager of the dev team stated that they didn't use the STL because it was too complicated and they did not understand it. Instead their "architect" had somehow written an in-house version of the STL.

    I crossed them off my list of places I would accept a job offer from.

    Also, my first job out of college was an outfit that used obsolete Data General Novas as their platform for a wireless WAN that bounced signals off meteorite trails (look it up - it is called 'Meteor Burst'). The head of R&D invented his own macro language to run on top of the assembly language of the platform - a mish mash of other languages. The devs hated it.

  • Sole Purpose of Visit (unregistered) in reply to Chronomium

    That one is actually quite easy. "We've got a greenfield project starting up, and it needs a radical new approach to databases. "Now, the preliminary architecture to this is going to be a thin-layer CRUD API to the business logic, and we'd like to evaluate a bog-standard commercial piece of rubbish against your cutting-edge technology. "We're really excited about building this. Plug your version in, and we'll have a world-beating product!" Then, when it doesn't work, fire the git.

  • Zenith (unregistered)

    You're right. Nobody should ever write a custom solution. Just use what other people wrote...or didn't, because they, too, were told to defer to somebody else (who in turn deferred to somebody else ad infinitum).

  • 04012018 (unregistered)

    https://bit.ly/2INFdKO https://bit.ly/2qtiIDj

    Nothing new under the sun.

    Peace for planet earth. Let's cure blindness. A neurological pathway treatment is in sight if we all just calm down.

    I see everyone marching towards the end of a tunnel. That tunnel leads to a well manicured clearing, with an empty Disney World.

    I just want to make sure I'm on it, and not usurped by a 7' cosmological serial killer 100 years from now!!!

  • (nodebb) in reply to Sole Purpose of Visit

    Then, when it doesn't work, fire the git.

    Out of a cannon. Into the sun.

  • bvs23bkv33 (unregistered)

    what he had written we know now as SQLite

  • 🤷 (unregistered)

    "Sybase sucks"

    As someone who has to work with sybase on a daily basis, I second that statement? Could I build something better? No. But it still sucks! >:(

  • Mr A (unregistered)

    I had one genius architect who replaced all Java with NodeJS. He then set about writing the code release, unit testing, database driver and monitoring frameworks himself because at the time NodeJS wasn't even a beta release. Except he never quite finished any of it, or understand the theory behind all this. 6 years on and they still haven't finished migrating the Java code. The Node stuff crashes regularly and they can't find any devs dumb enough to work there. A complete management blindspot

  • Guntank (unregistered)

    "Several months later..."

    Sounds like a male version of Paula.

  • Radius Theme (unregistered)

    Thanks for taking the time to discuss this.I thought that it was exceptionally helpful

  • Gumpy Gus (unregistered)

    Yeah, rewrite-itis is endemic. Very very long ago a girlfriend of mine wrote a program to process like five named files. She had code to sort the file names by date, but it was a slow bubble-sort. She kept asking for permission to write a faster sort for these five dates. Never mind that once the file names were sorted, the program went into huge gyrations to process the huge files. And never mind that this code ran on the fastest computer model in the world, a CDC 7600. So give her points for wanting efficiency, and take away many more points for missing the whole point.

Leave a comment on “To Suffer The Slings and Arrows of Vendor Products…”

Log In or post as a guest

Replying to comment #:

Log Out

« Return to Article