Bananas: cleanup, categorisation, all-in-one packs...

[Felix Atagong]

Recently there has been a question if OpenTTD has become too complex and another thread asked what where the most popular add-ons.
In my opinion OpenTTD is not too complex (but I have grown into it since over a decade now) but if one may say one thing about Bananas is that all this add-on stuff is rather messy and unorganised.

The AI section has got AI that
- work fine and are still regularly updated
- used to work more or less but are no longer updated (some of these grew into other projects)
- never properly worked and have been abandoned (I think the RocketAI was just a school project from some students, for instance)

The above remarks go for the alternative road, rail, air & sea vehicles, on top of that some NewGRF are incompatible with others, etc..., etc...

The problem is that ALL add-ons are put on the same 'level' and even I get confused and I can imagine that newbies are frightened by the list of hundreds of add-ons. I can well understand that a newbie gives up after he has tried 3 or 4 AI or NewGRF that 'crash' or do not work. (Please do not react as follows: one should consult the proper forum post before installing an add-on. With the hundreds of add-ons it will take weeks of reading before you can start playing the game.)

My proposition (to be done by a set of testers?):

divide the add-ons in playability groups, perhaps with color or number codes or in separate 'folders'
active add-on (still being developed, etc...): green light
working add -on, not developed any more: orange light
dead add-on, buggy, obsolete, etc...: red light
(another idea could be to put some add-ons in a dustbin.)

Second propostion: make pre-configured add-on packs for newbie use and lazy people like me. I know this will not be simple and open for a lot of discussion, but why not give it a chance.

For instance:
1800-pack, using vehicles and AI that will use the horse-driven vehicles, early trains, sailing ships...
realistic or geographic pack: the realistic vehicles that go together: North-Amerian pack, Japanese pack, etc...
any other category you can think of (but not too much, obviously)...

Any ideas?

[Lordmwa]

I like a lot!

If all the GRF makers could agree you could in theory combine lots of Newgrf files into one for example a UK pack with UKRS(2), Uk road, UK houses etc etc

I also agree it is a little messy and hard to use

As a simple solution just making it expandable so you only see scenarios or AIs would be an improvement

[planetmaker]

While I see the problem you mention, the big question is rather by what standards should it be judged which add-on is a good one and which isn't? Which is outdated which isn't? Who shall judge and who shall not? How do we avoid abuse, that person A marks the work of person B as junk, unmaintained and not working? How do you define 'being worked upon'?

Different people have different ideas of what constitutes a 'good' trainset, a good vehicle set, a good AI, a good scenario, a good heightmap etc and there's no universal standard for that, nor can there be.

Any suggestions on how to answer these questions?

[dihedral]

presets via bananas? could be an interesting thing - question would then be who may upload them, without hitting an overly large sum of presets people could download

[lugo]

the big question is rather by what standards should it be judged which add-on is a good one and which isn't?

Well if i get OP right he wants to filter/categorize results from Banana. Since this filter could/should be toggable i think it lies in the hand of the user if he/she trusts the opinion of whoever made that filter/categorization...maybe such a filter could be tweaked individually too

Any suggestions on how to answer these questions?

Conventions, maybe...

plus there could be a folder 'not sure' for projects which are in an unclear state of development (is BROS actually being worked on right now? )

[andythenorth]

One option could be a simple 'verified' status for newgrfs.

'Verified' would be determined using a similar approach to getting patches into trunk in ottd and other open source projects: small group of knowledgeable people (could be as few as one) review stuff to see if it's good enough.

Criteria would be mix of some subjective community standards, and some objective measures for project quality, such as:
- does it have a valid license + no copyright issues? (GPL, CC etc.)
- is it generally free from known quality problems (flashing pixels, bad palette, wrong or missing sprites, error output in ottd console, missing strings)
- does it take reasonable steps to play nicely with other newgrfs, e.g. approach to disabling / mixing with other sets, compatibility checks
- does it use action 14 min version or grfid to handle savegame compatibility
- is there a readme or other obvious manual that explains any needed parameters, version constraints etc.
- is it done? i.e. is there clearly a >1.0 or other stable version
- does it have a consistent style that fits into TTD

One thing that wouldn't be valid is "reviewer just doesn't like it"

It would be approximately the same rules as patches: newgrf developer has to ask for review, there's no obligation for anyone to review or verify a grf, decision of reviewers is final, whining / flaming makes no difference, if you don't like it ask for your money back.

There are ~189 newgrfs on Bananas, my bet would be that ~20% would pass the standards above.

[Lordmwa]

How about an optional collumn for user rating. If you like it you can add a point?

Not at all scientific but would bring the more complete sets to the top almost certainly

[Muzzly]

User uploaded presets with difficulty settings + user rating would be interesting combination !!!

[Elcamo]

I like it! Instead of adding the green light, red light, orange light, you could just have a 5 star rating system. (5 is a must have, 4 is very good, 3 is working but not amazing, 2 is sloppy/poor, 1 is not working.)

[lawton27]

I like Andy's idea of criteria for a verified section that way users could browse files which would definitely work fine, it would also encourage GRF authors to meet these standards.

User ratings would be useful too in my opinion, perhaps after 10 years of play on a game users could be prompted to rate the GRFs they are playing with (that they haven't already rated) to make sure they give an accurate reflection of the GRF quality. It would need to prompt again after file updates ect for course.

[Muzzly]

Just an idea. Maybe openttd should collect information about what grfs are used mostly, what map type and size is played and send this data to bananas server. Transmission could be made when user starts online update. It would be also kind of newgrf rating

[Hyronymus]

Just an idea. Maybe openttd should collect information about what grfs are used mostly, what map type and size is played and send this data to bananas server. Transmission could be made when user starts online update. It would be also kind of newgrf rating

Ever heard of privacy issues?

[Muzzly]

Ever heard of privacy issues?

Have you heard about interactive installation steps ?

[Lordmwa]

Anonymous usage information could be really interesting.

I bet there are a lot more people playing OTTD than we think

[dihedral]

I guess there are more people than you think who do not check that box

[Lordmwa]

Im pretty sure some programs have it auto ticked

[Eddi]

i'm pretty sure that is against the law in some countries.

[Lordmwa]

quite possibly....

Im not saying it should be done at all!

[Zephyris]

I would be interested in seeing it as an option though - usage stats on various things would help all developers (i.e. "real" ones and us grf makers)...

[Rubidium]

But what is "usage"?

For example, there's a NewGRF on Bananas called "Fix TTD grf bugs". It's totally pointless for OpenTTD as those bugs are fixed by openttd.grf (OpenTTD's part of the original base graphics), or non-existing in OpenGFX. Yet, many people "use" it.

Likewise one could ask what a good measure of usage is. Is it downloads? Are Bob's random vehicles better than say ECS? That's what happens when you look at the downloads of the current file. However, when you look at the download of all files the old stuff has a major advantage. Though it doesn't make them better.

You could look at the servers, but how representative are they for what is used on average? What about a someone owning a lot of servers and configuring that the same?

Finally, I'm personally against OpenTTD "phoning" home to tell us what NewGRFs you're using. It's only something that will break, get odd questions about firewalls asking to connect to the internet when in single player and such.

Nevertheless, if you want some statistics you can always look at the ones from content.openttd.org and/or the ones you can gather from querying the servers.