View Single Post
Posts: 521 | Thanked: 296 times | Joined on Sep 2009
#42
Originally Posted by daperl View Post
Okay, let's ask me.

Easy to Reproduce Crashes: I can easily fix those in less than an hour.

Random Crashes: How random? How often? How catastrophic? Depending on your answers and assuming it's a software issue, the time-to-fix usually ranges from 5 1/2 hours to ∞ - 1. But this topic is too big for this discussion.

Have you ever been part of a successful software project? Good developers don't freak out like you're suggesting. In your context, absolutes like "never end well" are FUD. WEEKS is a long time, and it doesn't frighten a good development group if they're in bug-fixing mode. Sh*tty management is what usually frightens a good development group.

If it's your code, bug fixing shouldn't be difficult. Good design is where all the real work is, right? So, children, the moral of the story is: Worry less about bugs and worry more about design, then things will always (okay, mostly) end well.

Back to the topic:

Nokia needs to find out what happened, having a high profile shoot-from-the-hip blogger as an extended part of your QA department doesn't seem like a good thing. That's what we're here for. And my prototype should be arriving any day now...
I guess I have differing opinions to you about software development. I have been part of several successful projects in group of 10's which the software will sell annually 200-400M dollars to hundred of customers.

A large project which exhibits these behaviors WEEKS before release is in trouble. Even though the bug can be fixed in hours, it introduces risk, QA cycles have to be redone. You cannot be sure your changes have not affected anything else. To think otherwise is "hacker" mentality and not a "software development".

Yes, the developer never freaks out, but its a sign that its too early for release and should be set back. No matter how good QA is, QA cannot fix bad code.

Large projects with only a handful of customers and a limited set functionality may get away with it but I'm talking about O/S type code which is open for any combination of functionality and integration.

Anyway, hoping again the problems are only visible in Eldar's weird unit and not issues being "withheld" by fanboy reviewers (check out how bad reviewers of the N97 were crucified by fanboys!)
 

The Following User Says Thank You to bugelrex For This Useful Post: