Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[EmeraldWind]

I made one these a while back and posted it on Reddit when people were complaining that DF versions always take forever to come out.

Mostly to show people that in general DF versions don't take very long and the few that did are outliers. Now that 0.40.01 is out I decided to clean it up and repost it with the complete data on it.

I only put labels on the bars with more than 30 days of devtime on them. And I'm missing a bunch of version numbers just to keep the chart a little less cluttered. I might eventually do something similar to data labels where I only have significant version numbers and version numbers of the outlier columns or something.

[Footkerchief]

Those "outliers" account for like 70% of all dev time since the first release.  That kinda makes them not outliers.

[EmeraldWind]

Those "outliers" account for like 70% of all dev time since the first release.  That kinda makes them not outliers.

Yeah, but at the time people were complaining that every version had an average devtime of 6 months to a year
when in reality there were only 7 major releases with more than 3 months of devtime. Or 12 releases that had more than a month of devtime out of the 80 approx. releases.

It was just to point out the absurdity of the claims... which frankly a lot of newcomers were starting to believe were true. That said... the only release I couldn't pin-point a devtime for was the original release.

Granted outliers might not have been the best term for what I pointing out. I was looking for a word that meant points of data that are significantly different from other points of data in a given set.

Edit: I should also mention I have a timeline chart that shows the process as a span marking out different versions and groups of versions (which I refer to as series, IE 0.31 series). Which was to show what you mentioned about them also contributing the bulk of the time in dev time, but I got a lot of improvements to make to that chart before I post it again... because I want to show more detail on it than I current have. Plus some of the colors are misleading due to a design choice I made at the beginning before realizing my design didn't work well for the information I was attempting to get across.

[glopso]

The average time for a release isn't actually that long, but that isn't a meaningful measurement because the time to a release and the amount of work done for a release does vary so much. A more useful graph might be one that plots (release time / change in version number) against version number. I don't know how you would work bugfix/minor releases into the graph though, since their version numbers advance in a different way from the major releases. I would just put them in a separate graph, especially because I would expect bugfix releases to have different factors affecting release time than major releases.

[palu]

The only release I couldn't pin-point a devtime for was the original release.

Dwarf Fortress started October 2002

EDIT: fixed quote.

[EmeraldWind]

The only release I couldn't pin-point a devtime for was the original release.

Dwarf Fortress started October 2002

Now how did I miss that?

Edit: Thanks for that. Now I have that as well.

[Kokabiel]

Mostly to show people that in general DF bugfixes don't take very long and the releases do.

FTFY

[thvaz]

The only release I couldn't pin-point a devtime for was the original release.

Dwarf Fortress started October 2002

Now how did I miss that?

Edit: Thanks for that. Now I have that as well.

In have to account for the fact that Toady used to work only pqart time in DF. I don't remember when he started working on DF full time, but it was quite a few months after the inital release.

[miauw62]

I feel you should make a graph showing the progress of the version number over time.

[EmeraldWind]

I feel you should make a graph showing the progress of the version number over time.

Yeah, I could probably put that together as well. It might not be all that interesting though.

Also DF seemed to use a different numbering system prior to 0.31.01, so I'm not sure if those would work as well in showing that kind of progress.

On flipside, it would be an interesting bit of data.

[Thief^]

The first number is the same as before, the letter at the end of the old number corresponds to the number at the end of the new system. The other numbers from the old system can be ignored mostly.

In other words, 0.28.181.40d is roughly equivalent to 0.28.04 under the new system.

[cephalo]

I think it's good to have long periods of version stability once the larger bugs are fixed, because it helps modders and utility makers to focus more on content than version compatibility, and it also helps in making very large fort projects when you don't have to worry about save compatibility. However, this last time was too long, and left too many bugs in before it started.

[thvaz]

I think it's good to have long periods of version stability once the larger bugs are fixed, because it helps modders and utility makers to focus more on content than version compatibility, and it also helps in making very large fort projects when you don't have to worry about save compatibility. However, this last time was too long, and left too many bugs in before it started.

It is because Toady and Threetoe aren't enough for playtesting. He has to playtest, find the bugs and fix them. Now mostly he just fixes them, sometimes the community even pinpoint the cause of the bugs for him.

[cephalo]

It is because Toady and Threetoe aren't enough for playtesting. He has to playtest, find the bugs and fix them. Now mostly he just fixes them, sometimes the community even pinpoint the cause of the bugs for him.

I don't mind playtesting, and I think most of us quite enjoy it, but before he goes on a two year coding binge, I prefer if he could get the bug count on the tracker down to 500 max. We were stuck with around 2000 bugs last time for two whole years! A lot of those bugs are 5 years old.

Footie and Dwarfu do awesome work at keeping that tracker organized and identifying which bugs are real and which are duplicates and which simply became non-issues. That's a lot of work! Toady needs to honor their sacrifice and squish em.

[reality.auditor]

I think tendency to have longer and longer periods between major releases is visible. If I was to bet, next major release will be longer than 323, but shorter than 763 days - but ANOTHER release after that will beat 763 days. Nice definition of "outlier" designed to downplay and deny this tendency noted.

I hope I am mistaken and my prediction will not happen.