Friday Facts #232 - PAX, Bugs, Graphs

[Jap2.0]

Impressive decline in crashes.

Or just people learning what not to do because then the game crashes.

Which would be a weird result of the automatic log uploading.

You also notice that decreases generally correspond with a new version. I find it interesting that there are still several people playing the old experimental versions.

[mrvn]

Impressive decline in crashes.

Or just people learning what not to do because then the game crashes.

Which would be a weird result of the automatic log uploading.

You also notice that decreases generally correspond with a new version. I find it interesting that there are still several people playing the old experimental versions.

Never change a running system. Although if the game crashes is it running?

What I commented on was that after the initial spike of bug reports for a new version it seems to be falling. Since the version doesn't magically become less buggy the reason must be that people trigger the bugs less.

But looking again at the various versions I wonder what the total amount of reported bugs is. I guess the first spike and drop of in .24 is when people noticed the upload and opted out. But then .25 came out and people switched to that. I would have expected a big drop in reports for .24 at that time. Or is the .24 line the sum of .24 + .25 reports and only the visible colored region is the amount for each version? Is the graph additive or overlapping?

[5thHorseman]

I read it as that when a new version comes out, people who aren't playing load it up to just fiddle a bit, then stop. This makes that version spike early. Others are playing games and don't want to update right away for some reason, which is why there's a long tail. Others just play whatever version comes out when it comes out, and that's the bulk of the entire graph.

The significant thing to take from this graph in regards to number of crash reports is merely that they're going down as version numbers go up. This could be due to people opting out in droves, but I suspect that's not the case and Wube should be able to find out how many individuals are submitting these reports, and therefore at least get a speculation as to if the numbers indicate less total bugs, or less total people submitting reports.

[mrvn]

Also don't forget that people that had a bug will try the same thing with the next version to see if it was fixed.

[Zeblote]

Is the graph additive or overlapping?

It's probably additive, otherwise the reports per day would be increasing massively - which makes no sense.

[AnarCon]

u devs are simply the best!

[posila]

can't think of anything to write about

There is one topic that I'd like to hear about.

Yes, it's additive.

17th and 18th was weekend, it dropped a little on Monday 19th, and next day it dropped much more because next version with fixes for the most common bugs came out. Next spike is another weekend 24/25.

[Alice3173]

I find it interesting that there are still several people playing the old experimental versions.

Could be for a variety of reasons. A few versions ago I updated and had an error (seemed to be related to a mod I think) where my save wouldn't load because of some tile error. I forget the exact error but it was something about trying to allot more than 255 tiles or something like that so I skipped the next few versions so I didn't risk having to start my save over again. I also frequently have an updater error regarding some Entities.html file or something like that (a file which doesn't appear to even exist as far as I've seen?) which requires me to manually download the update and overwrite the old files myself. When that happens I typically end up waiting a few updates before bothering.

[seePyou]

You are going to Boston, but you could not make it to PC Gamer Weekender? I'm sad... I've been waiting for you guys to show up there for three years, and every year you are not there There is always hope for 2019's Weekender...

[Sander_Bouwhuis]

You also notice that decreases generally correspond with a new version. I find it interesting that there are still several people playing the old experimental versions.

I do that. Better the bugs you know so to speak. I only update my server when I see something in the change log that I'm interested in. I don't want to come home tired and find out one of the mods is not working, so we can't play because I have to spend half an hour dicking around trying to get it fixed. After that, I'm often too tired to actually play.
With the high frequency of updates I simply update every 5 or 10 versions or so.