8
wojtek322
108d

How are you supposed to fix bugs in your program where:
1. it has been investigated before and was not found how to reproduce it
2. it cannot be tested at all
3. never happens on the test environment

fml

Comments
  • 8
    If a bug is not reproducible, it does not exist
  • 6
    Add more logs so if it pops up keep adding more logs until you figure out what happened
  • 4
    I used c realloc the wrong way everywhere in my programming language and it never faulted. Only if you would execute it thousands of times like in batch it COULD go wrong 😁 Found out later, never destroyed an evening 😁 And yes, the language was well tested in its own language
  • 1
    @retoor im really curious as to how you misused it now... i also miss good ol' c
  • 1
    @awesomeest no memset / zerofill after it so I could randomly contain a zero somewhere is my theory
  • 2
    @retoor we had a "squishy" bug in our codebase for 6 months. I thought I had solved it by upgrading the library version. It came back later. Every time I debugged it (when I could reproduce it) it would appear in unrelated code. I finally was going through some code for a different reason and saw some funky pointer shit being done. I cannot remember exactly what the issue was, but a pointer was being used for another context which was invalidating the original context. So it was leaving a fucked up pointer behind. It would eventually time bomb later and the debugger would think it was other code causing the problem. This was the "squishy" part. It moved all over the place. In hindsight the way we reproduced it should have given me a better clue. I found out later a lot of the code I was dealing with was written in a weekend in a rush for some reason. I spent months sorting out all the issues in the code base and making it less shitty to maintain.
  • 1
    @Demolishun wait until you see my meme post 😁
  • 0
    @Demolishun can't post yet. That one time in my life I had two things to rant about in 2h..

    aaa
  • 1
    @retoor im shit at saving/backing up my code (even worse at it being decipher-able for others... seriously, im great at writing crypto and compression algos but naturally dont even need them) but base level storage in C is one thing ive kept snippets of for over 2 decades. Im a data architecture nerd for sure... with a horribly invalid frame of reference to what is "normal". That said, i appreciate the insight.
  • 0
    @jestdotty And then you get a 5000$ bill from Azurem, becaouse you logged 5 TB of data per day withj "detailed logs".

    Not viable in production. Production logs should be as minimalistics as possible.
  • 2
    @retoor well? we're waiting!

    Be retoor did, not retoor would.
  • 3
    @NoToJavaScript this so why I don't make myself dependent on Goliath companies that are not at all invested in my success wtf

    We have strayed so far from functional. Seriously. Cloud logging services? Lol
  • 3
    @Demolishun I did it but you didn't upvote 🙁
  • 1
    @retoor what? I thought you were going to post another rant? I don't see it in your rants.
  • 0
  • 1
    @retoor I did upvote though. lol
  • 3
    @Demolishun now I'm happehh 😁 I'm happehh in general. Got an operation today. May leave the same day :) Happehh
  • 1
    @retoor did they give you happehh pills or something?
  • 1
    @Demolishun no, but I slept so good under narcose. Just wow
Add Comment