Feelings about recent "development"

[Timo Pietilä]

Unsurprisingly, I disagree. Balance is actually really really difficult to see.

You got my words backwards. Change in gameplay is easy to see (does it work or not). It's possible effect to balance is a different matter completely and needs longer testing.

[ChodTheWacko]

Timo stated that you don't want to "release/finalize" a release without 6 months of testing.
He said 1 year but 6 months should be plenty. The point is you don't have a full idea how balanced something is until people have played all the races/class combos through to the endgame. I don't think anyone can deny the logic of that.

The reason we aren't getting that is we have the release builds and the dev build.
I've worked in software development for a while, and imho, the failing is the lack of an intermediate build - the functionally locked beta test build.

( DISCLAIMER: If I am wrong and we DO have a beta build (labeled #2 below), then the rest of this post is invalid and should be ignored ).

We should have:
1) Fully stable/playable build. This is 3.2
2) Functionally locked build for bugfixes ONLY. And required balance fixes. (the leather shoes bug - stuff blatently wrong).
3) The pre-alpha build for testing new features and stuff not ready for prime time.




The problem is that the dev build is avoided because new features always contain major bugs. I'd love to try it but everytime one bug is fixed another is added. Pval errors. Shards not working. Crashes due to the new rooms. Food not working. That will NEVER stop until you functionally lock down the product. Every single major software product must go into a functionally locked/alpha/beta/release candidate/Final release cycle.

Timo's post incorrectly implied that development should stop for 1 year. That's not correct.
Development never stops - but it happens in codeline #3. #2 is stabilized until it becomes the new #1. Bugs discovered in #2 should take top priority. Once #2 is declared 'good to go' it becomes the new #1 and then stuff that is 'close to ready for prime time' gets pushed into #2 for the next round of playtesting.

Developers and people who want to be bleeding edge can downloaded and test #3.
The general public who wants to play something new and generally stable play #2.
Players who won't tolerate any weirdness/crashes at all can play #1.

I think this is sane and fairly standard.
Let's look at Firefox - last I checked there was the 'Generally available' 3.5, 3.6 beta, and the nightly builds for 4.0. (i'm sure the version #s are bumped up now, but you get the point)


- Frank

[Magnate]

This almost rubs me the wrong way. It almost sounds like "You'll know we've achieved balance when we tell you that we've achieved balance." I almost feel obligated to add the smiley face. Ehh, what the hell! .

I honestly wasn't grinding any kind of axe - I was just trying to say "the better you know something, the harder it is to adjust to changes to it". So when I watch an episode of Babylon 5 or Star Trek or whatever, I'm not at all fussed if they've done some massive retcon, but if I were an aficionado who knew all the backstories, I'd be royally pissed.

@Timo - sorry, I did indeed parse your words backwards. So we agree once more!

[Magnate]

I think this is sane and fairly standard.
Let's look at Firefox - last I checked there was the 'Generally available' 3.5, 3.6 beta, and the nightly builds for 4.0. (i'm sure the version #s are bumped up now, but you get the point)

Yes, this is how Debian does it too - Stable, Testing, Unstable. If I've understood takkaria correctly, we will be moving towards something close to this model. 3.3.0 will be the "Stable", and there will be a "bugfix" branch for 3.3.[1+] which will be bugfixes only. The main staging/master development branches will be your #3, which will eventually become 3.4.0 ... and so on.

Crucially, one of the devteam will have a responsibility to cherry-pick bugfixes and update docs for the bugfix branch, and to do bugfix releases. This enables the rest of the team to get on with the next version, without anyone feeling guilty about bugs in the major release version remaining unfixed.

[Timo Pietilä]

We should have:
1) Fully stable/playable build. This is 3.2
2) Functionally locked build for bugfixes ONLY. And required balance fixes. (the leather shoes bug - stuff blatently wrong).
3) The pre-alpha build for testing new features and stuff not ready for prime time.

Timo's post incorrectly implied that development should stop for 1 year. That's not correct.

I did mean that development should continue for bug fixes so that people get time to actually get to know the game and all the new features. Currently we don't have #1 at all, we have few #2 stage versions that are treated like #1 (both 3.1 and 3.2) and #3 that is changing like mad.

Also the bottom truth in angband is that it is or more correctly was a ready product. It just didn't need huge changes and lots of development. Changing it rapidly for sake of change just breaks things. Frog-knows with modern UI would still be a great game. 2.8.3 still is a great game. 3.0.x with JLE changes required balancing that didn't happen. From that point onward balance of the game has got worse in every version. MoD with 6 blows for warriors, 8 if it is extra blows. That kind of stuff should not happen.

(BTW your FF example wasn't the best possible example. There will be huge major changes in that within this year, they move from 4 to 7 in just few months. I hope they get it stable, but that fast I don't think they will have huge success in it).

[Magnate]

Also the bottom truth in angband is that it is or more correctly was a ready product. It just didn't need huge changes and lots of development. Changing it rapidly for sake of change just breaks things. Frog-knows with modern UI would still be a great game. 2.8.3 still is a great game. 3.0.x with JLE changes required balancing that didn't happen. From that point onward balance of the game has got worse in every version. MoD with 6 blows for warriors, 8 if it is extra blows. That kind of stuff should not happen.

And indeed, in 3.0.6, it didn't.

What you don't seem to have understood is that nobody wanted to *maintain* Angband after Robert. Julian didn't, and neither did takkaria - that's why we have a *development* team instead of a maintainer.

3.0.6 hasn't gone anywhere - it's still there if you want to play it. So is frog-knows.

You seem to want the non-gameplay stuff to be improved (UI etc.) without anyone touching the gameplay. You are of course welcome to do that, or try and find other people to do it, but all of the current devteam want to contribute changes, and that's what takkaria is managing.

There is nothing stopping you taking 3.0.6, or f-k, or anything in between, and updating it to use all the modern UI improvements and fixes. Nothing. I know you are not a coder yourself, but there's nothing stopping you leading a team of coders to do it.

You will recall that I actually offered to create a branch that re-created the exact gameplay of frog-knows, but you declined. That tells me that you actually want excuses to keep on complaining.

[Derakon]

Two things, Timo.

1) The previous versions of the game still exist. It's not like the existence of 3.2 means that 2.8.3 is no longer available. I'll grant it's harder to find, but it's not gone.

2) You're not going to find a maintainer who's willing to not touch the game. Why be a maintainer then? Just so your name's in the splash screen when the program launches?

[Antoine]

Yes, this is how Debian does it too - Stable, Testing, Unstable. If I've understood takkaria correctly, we will be moving towards something close to this model. 3.3.0 will be the "Stable", and there will be a "bugfix" branch for 3.3.[1+] which will be bugfixes only. The main staging/master development branches will be your #3, which will eventually become 3.4.0 ... and so on.

Crucially, one of the devteam will have a responsibility to cherry-pick bugfixes and update docs for the bugfix branch, and to do bugfix releases. This enables the rest of the team to get on with the next version, without anyone feeling guilty about bugs in the major release version remaining unfixed.

This sounds very good

A.

[Antoine]

As Derakon and AnonymousHero have both helpfully pointed out, dev versions are published precisely to increase the amount of gameplay testing of changes. The devteam make sure that there are no obvious crash bugs [cough]usually[cough], but since most of us are not very proficient players, it is not sensible to hold back dev versions while we playtest.

I think your tags are wrong - did you mean [/cough] there?

I guess the risk is that players get put off the nightlies by bugs (or misfeatures) and then it becomes hard for you to get the level of feedback you want.

Would it make sense to have an intermediate build - weekly, fortnightly, or monthly perhaps? - which people can play if they want to test interesting new features in advance of the next version, but don't want "a lot of bugs or weird stuff"?

Does something like this exist now?

A.

[buzzkill]

What you don't seem to have understood is that nobody wanted to *maintain* Angband after Robert. Julian didn't, and neither did takkaria - that's why we have a *development* team instead of a maintainer.

... but all of the current dev team want to contribute changes, and that's what takkaria is managing.

I not sure of the opinions of the others who seem to be on my side of the fence, but this is exactly my beef, and no one on the dev team seems to "get it". Vanilla should be maintained. If small changes creep in during the course of maintenance, so be it, some things are inevitable. If it's your aim to "contribute change" then, then you really should be working on a VARIANT of Vanilla, not Vanilla itself.

Some valuable innovation has been put forth by the current dev team, but in addition to this they've independently (of necessary maintenance) and fundamentally changed Vanilla, and have done such by following a game plan which shouldn't really be in the maintainer playbook.

[kaypy]

Vanilla should be maintained.

"Should" doesn't count for much if no one is willing to do the shitty soul-destroying job of maintainance coding.

So your choices are either development (as-is) or stagnation (vanilla gets renamed takband or something and 3.2 sits untouched).

[Nick]

I not sure of the opinions of the others who seem to be on my side of the fence, but this is exactly my beef, and no one on the dev team seems to "get it". Vanilla should be maintained. If small changes creep in during the course of maintenance, so be it, some things are inevitable. If it's your aim to "contribute change" then, then you really should be working on a VARIANT of Vanilla, not Vanilla itself.

This is a valid viewpoint, but as I understand it it isn't shared by takkaria, or anyone on the devteam, or most of the people here. Timo said upthread that he thinks the game is improving; the complaints he has (again, if I understand correctly) are about rate of change.

I think this comes down to a question of ownership. "Angband" is not really owned by anyone, but several years ago there was a discussion on rgra which led to broad agreement that takkaria should be maintainer; at about that time, he expressed his view about the future of Angband. Anyone else at any time can claim to be maintainer and set up in opposition, or whatever - there's no divine anointing. Currently what most people agree is "Vanilla" is what is produced by the devteam. You don't have to agree with that, but turnabout is fair play

[Magnate]

This is a valid viewpoint, but as I understand it it isn't shared by takkaria, or anyone on the devteam, or most of the people here. Timo said upthread that he thinks the game is improving; the complaints he has (again, if I understand correctly) are about rate of change.

I think this comes down to a question of ownership. "Angband" is not really owned by anyone, but several years ago there was a discussion on rgra which led to broad agreement that takkaria should be maintainer; at about that time, he expressed his view about the future of Angband. Anyone else at any time can claim to be maintainer and set up in opposition, or whatever - there's no divine anointing. Currently what most people agree is "Vanilla" is what is produced by the devteam. You don't have to agree with that, but turnabout is fair play

This is really important. There is nothing stopping Buzzkill, or Timo, or anyone else, from making a unilateral declaration of maintainership: "I/we will take [insert favourite version here] and *maintain* it, fixing bugs and improving the UI but preserving the gameplay". Nobody would mind that, and lots of people would like it. IMO it is even likely that takkaria would agree to renaming the version that the current devteam is working on, in order that this new "maintained" version could take the Vanilla title.

But nobody can ask the current devteam to do that. If you want it done, you find someone who is willing and able to do it.

[Magnate]

I guess the risk is that players get put off the nightlies by bugs (or misfeatures) and then it becomes hard for you to get the level of feedback you want.

Would it make sense to have an intermediate build - weekly, fortnightly, or monthly perhaps? - which people can play if they want to test interesting new features in advance of the next version, but don't want "a lot of bugs or weird stuff"?

Does something like this exist now?

No - though it has been discussed on a few occasions when the nightlies have been buggy. The basic problem is that we are a small team - between two and eight of us are active at any one time, and that's not really enough personpower to manage three different builds. Takkaria's original plan was for the entire 3.1.x series to be "beta" releases, and for 3.2.0 to be a new "stable" release. But we went wrong for a couple of reasons: the years between 3.1.0 and 3.2.0 were too long for the concept of a beta, so 3.1.1 and 3.1.2 came to be judged as non-beta versions even though they weren't meant to be - and when we did release 3.2.0, we did it without a code freeze which meant that too many late changes were not properly tested, so it wasn't really any less beta than 3.1.x

I think the new strategy (of releasing 3.3.x as bugfix releases, and eschewing betas in favour of dev versions roughly weekly) will be better for everybody. Yes there is still a danger that people will be put off by bugs in dev versions, but to be honest that's ok - in a normal environment the vast majority of users would stick to the mainline version, with only a small minority playing the dev versions. In this community it's inverted, which IMO has been part of the problem.

[Timo Pietilä]

And indeed, in 3.0.6, it didn't.

What you don't seem to have understood is that nobody wanted to *maintain* Angband after Robert. Julian didn't, and neither did takkaria - that's why we have a *development* team instead of a maintainer.

3.0.6 hasn't gone anywhere - it's still there if you want to play it. So is frog-knows.

It would be enough if there would be maintainer for common sense. Not to fix things that are not broken. Look at the past what did work.

You seem to want the non-gameplay stuff to be improved (UI etc.) without anyone touching the gameplay. You are of course welcome to do that, or try and find other people to do it, but all of the current devteam want to contribute changes, and that's what takkaria is managing.

Gameplay can be touched, but carefully, not with a sledgehammer. If thing is already balanced and working turning it upside down every now and then just plain cannot give you as balanced or working result than the original was.

There is nothing stopping you taking 3.0.6, or f-k, or anything in between, and updating it to use all the modern UI improvements and fixes. Nothing. I know you are not a coder yourself, but there's nothing stopping you leading a team of coders to do it.

Yes there is: I don't have that team to back me up. Team that would like to do those changes without tweaking everything. Or time for it.

You will recall that I actually offered to create a branch that re-created the exact gameplay of frog-knows, but you declined. That tells me that you actually want excuses to keep on complaining.

I declined because I don't believe you would actually do it. It is too much work for any single person. But prove me wrong and go ahead and do it anyway. First thing you need to do is write a console version of the game, because extra term windows give away too much information. If you want more modernized version of it look at sangband. Then deal with colors, targeting, LoS, dungeon generation, monsters, items, item activations, rarities, depths. Pretty much everything has changed.