bob wrote: ↑
Mon Feb 17, 2020 5:27 am
chrisw wrote: ↑
Sun Feb 16, 2020 10:03 pm
Then show a list of bugs “that have been there for a while”.
Do you not realize how silly that sounds?
Still on the NPD put-downs? Can't help yourself?
And no, it doesn't sound silly at all. Someone said there's a list of bugs, so I responded, ok show us the list of bugs. Why would you think that was "silly", when it's just a request for actual evidence of an assertion? Well, it isn't silly, but you needed to get a rudeness/put-down into your first sentence, we've spoken about this on your part in the past, I think?
If they can't show a list, the program is therefore proven correct? Pick up ANY software engineering textbook and read the chapter on testing.
You can stop the patronising telling me to read XYZ as well. Is just another gratuitous put-down, and you know what those mean about you.
If there's no list (it was asserted a few posts ago that there is a huge list) then that is good evidence for 100% bug free over the time the list was compiled. Stockfish is a remarkably solid program, and that's down to management and attitude and testing.
Having a list of bugs fixed in the last year does not show the program is now bug free, nor does it show it is not bug free. It just shows that all detected bugs have been fixed. The set of all existing bugs is guaranteed to be a super-set that includes the set of known bugs). The problem is finding those in the super-set but not in the known set. Hard to find that which can't be seen, until they produce an artifact that can be seen. Just because you can't see an obstruction in a dead straight road does not mean you can drive your Lambo at 200 miles per hour for an hour with no problem. You can see about 7 miles ahead due to curvature of the earth. What bugs lie beyond your line of sight? How can you know until you actually _get there_?
I doubt you can find a single chess programmer that is actively working on their chess program, who has not discovered an unknown bug in their code while looking for something unrelated. Multiple times in a year.
Obviously. When something is being worked on there will be bugs of one sort or another. By the time its considered finished and been through QC and testing through to release, the goal, the requirement is 100% no bugs. That's entirely doable for something (a chess program) relatively simple that's come out of a long history of computer chess. No excuse at all for anything to be broken. That's the publishing requirement, and those of us who have worked to 100% no bug contracts know how to deliver.
Bugs that have had zero impact on the program (so far as we knew prior to discovering the bug). This is just a part of writing complex software. Shoot, even the "almighty" could not produce a bug-free program in our DNA. If he could not, we certainly can not.