Looking for a simpler PC Gen...

[mattcolville]

But asking people in this thread to help fix it is exacerbating the problem.

Why?

Very next sentance;

The people in this and other threads are not experienced tools programmers.

I'm not a tools programmer. I'm one of the people the tools programmers has to occasionally say "no" to. I don't have that education or experience.

This is why I don't think we're going to see a mass-market D&D character editor until a large company with the right experience steps up and signs a deal with WotC. I think this requires a Cathedral, not a Bazaar. Maybe if there were 30 million D&D players, you'd have a wide enough potential developer base, but that's not the case.

 

[kingpaul]

I'm not a tools programmer.

And who said help had to come from just programmers? Design specs were being discussed as well, and how the GUI was not appealing. I don't code for PCGen, yet I help out with the project.

 

[DethStryke]

So, if I had stated the docs were set up in a tree/sub-tree format, that would have been better?

My apologies, that was not clear... I meant if that format in Windows was presented as a standard. My mistake. :/

But seriously, as I stated earlier, overhauling the GUI is something thats been attempted in the past, but those GUI monkeys disappear on us.

That's because to be interested in doing GUI work, you have to be an ar-teest; hence the Graphic end of that.

We all know them artists be flaky!

 

[labyrinth]

If you want an example of character creation software that has an intuitive UI that even a non-techie can figure out quickly, go check out Hero Lab at www.wolflair.com. For a V1.0 product, they sure seem to have done a lot of things right on the first try.

Maybe WotC should go talk to those guys about a license....

Or maybe the PCGen team should take a look and take some notes. Supposedly, Hero Lab was written by only TWO guys. Perhaps the trick is to have a small team that actually understands usability (like Matt opined) instead of an army of coders working in a vacuum. Yes, I tried PCGen and no, I was not impressed. Hero Lab makes intuitive sense from a user's perspective, while PCGen is a confusing mess to me.

Hero Lab has a free demo, so there's no excuse for not checking it out. That includes the PCGen team.

 

[Taurendil]

Are you saying the users are loading to many datasets, or that PCGen is grabbing more datasets then is intended? Its a known issue that loading more datasets slows the program down, and the coders are looking at that as well.

No, I'm under the impression that loading a dataset in pcgen is an all-or-nothing thing. I had the impression that if i load the Dragonlance set (from CMP) to use the mystic class, all spells, feats, races, etc. are also loaded. So when you only use one thing from a dataset, a lot of redundant data gets loaded (taking up system memory).
Looking at the .lst files I also guess that you always load an entire file, even if you only need one thing from that file.
So if you use a feat from source A, a class from B, some items from D and E and a monster from F, you've got a whole lot of redundant info floating around in memory.

I think you could allow users to specify which parts of a dataset to load (e.g. load the classes, but not the items). Although I can imagine that some parts of a dataset are dependent on each other (class a needs feat z to function or some other thing).
Loading only a part of a file is probably not possible with the current .lst files and would probably require some kind of rdbms (sqlLite or other?).

 

[kingpaul]

No, I'm under the impression that loading a dataset in pcgen is an all-or-nothing thing. I had the impression that if i load the Dragonlance set (from CMP) to use the mystic class, all spells, feats, races, etc. are also loaded. So when you only use one thing from a dataset, a lot of redundant data gets loaded (taking up system memory).

Ah, I understand you now. Yes, currently, there's no way, at least none that I know of, to have PCGen pick and choose only the items that you want out of various datasets. I know some users have created custom *.pcc files so they only call the files they need.

 

[mosat]

I think you could allow users to specify which parts of a dataset to load (e.g. load the classes, but not the items). Although I can imagine that some parts of a dataset are dependent on each other (class a needs feat z to function or some other thing).
Loading only a part of a file is probably not possible with the current .lst files and would probably require some kind of rdbms (sqlLite or other?).

There is a Customise button in the Source Materials Tab which will do this to an extent. You can create your own Source (.pcc file) which will load any specific .lst file in PCGens data folder. You are right about potencial problems with unincluded dependencies, that is a risk. And also this method does not allow you to pick out single objects, however I feel this is where PCGen shines, since the data is all in text files you don't need anything more than a text editor to start digging in any creating this level of customization for yourself.

 

[soulcatcher]

If it were marketed and presented as a "set-up", then your logic would stand. As it is, every new version of windows since 3.1 has introduced wizards made to automate just about every single thing any end user would ever want to do with a PC. Millions of people drive cars every day and don't know what a clutch is or how it works, despite every car having one - it is done for you in automatics (the majority in America).

As mattcolville has pointed out (and I did in a little rant a few posts before this) it fails in that many people can't stand to use it. *IF* it ran less like a C-64 game with the loading times and did not require above average knowledge of software/computer use to make it function, then it would be the best thing ever and this thread wouldn't exist. It would literally be the original post, a link to PCGen and a bunch of people reading it. Maybe another few posts comprised of "Bump!" and a request to have it stickied.

If you have the cure for cancer but you chisel the knowledge on a stone tablet in a dead language, what good is it going to do anyone?

I just want to point out here, PCGen may be very complicated, but I think it's fair to say that what it's trying to do is *absurdly* complicated. If people want a character generator for just the SRD, our solution may well be far to complicated. But we didn't set out to create a character generator that can just do SRD. We set out to build a character generator that will allow people to use rules from essentially any source (yes, we can't distribute data for everything, but if a person wants to do something with their character, odds are, PCGen can do it). So far as I'm aware, there just isn't another application that can model as many rules as PCGen can. Can we make the GUI better? Yes. Can we improve perfomance? Certainly.

Improving PCGen is something we have had to put a lot of thought into. Writing an app that can do what we can with the data *and* performs *and* has a simple gui is not the trivial problem you all seem to think it is.

I'm open to any suggestions, but the simple fact is, writing an app that will let a person create an epic psionic vampire wereboar, and do so accurately is in itself a difficult problem. Coming up with an interface that makes this easy borders on the impossible.

Simply put options increase complexity, and complexity increases perceived difficulty. It is *inherently* at odds to allow a person to create an epic psionic vampire wereboar, and at the same time consider interface simplicity to be paramount.

Now, I'm *not* saying that we can't make pcgen easier. But making pcgen *easy* is at odds with our chosen goal of making pcgen capable of performing as many rules as possible. (Really think about this. Really just sit back and consider that there are hundreds of OGL books on the market, and each has subtle new rule changes, and outright conflicting rules and rule changes. The actual possible combinations that these rules can make when used on a character probably number in the billions of billions.) There are plenty of applications that can make building a Player's Handbook character, there are less that can do the whole SRD. There (so far as I'm aware) is *one* that can handle almost any book thrown at it. The SRD only market clearly isn't our target market. I don't intend to be snarky at all when I say this: If you only need the rules from the PHB and DMG, PCGen was not designed for your needs. There are indeed simpler and faster tools out there for this simple subset of the rules. But when you buy that shiny new book from Green Ronin, and want to use it (or some of it) in your game, almost none of the other apps out there won't meet your needs anymore. Few of them will let you even enter new rules. You can enter the rules in PCGen. It's not simple to do so, but it does work, but likely you won't even have to, because PCGen may even ship with the book you are now using.

I invite you all to *actually* look at pcgen 5.10.1, because a lot of thought was put into it to make the interface more consistent, and for it to offer the user more clues to the user. I suspect many of you have used PCGen here and there over the past 5 years. The application has had it's ups and downs, but it has continued to improve, continued to accept more and more rule variations, and *continued to get faster and easier*. Before 5.10.1 shipped, we put a lot more polish on it then any previous release. We standardized the windows in it, added search boxes just about everywhere, added clear tips as to what was unfinished with your character, and a number of other UI improvements. *yes* it is still the same basic UI, but it it a hell of a lot cleaner then 5.8 or any previous version. And in the future, 5.12 will be cleaner still, and on to 6.0.

And to those who scoff at us asking you to come help, You don't have to be a programmer to help. You don't have to edit lst files to help. You don't have to document to help. Even filing bugs, or pointing out specific elements that are flawed helps. Come over, join our lists, and throw stones constructively. Like products put out by businesses or not, all of you have unparalleled access to the people who make and maintain PCGen, and we do it *for you*. If you want a better character generator, you have the forum and the access to the people to make it happen.

Devon Jones
PCGen BoD
Code Architecture Silverback

 

[soulcatcher]

I do care that it's cross-platform since I use windows less and less. That being said, I don't think Java's to blame for the speed issues. Having looked at the source for pcgen, it's main problem seems to be it's domain model with lots of data being needlessly copied and passed around and too many classes inheriting from one class. This means that something like a language object can be assigned follower objects and that makes a language object bloated.

I think another problem might be that it loads to many data at startup, even if it never gets used. It maight benefit from some form of lazy loading, but I guess that it would need a db backend for that.

Java certainly isn't the fastest language out there but refactoring the program could do a lot to alleviate the speed issues. So I wouldn't throw out the cross-platform-ness of pcgen.

I don't have anything to do with pcgen (I just use it in my campaign and had a look at the source), but I'm guessing some of the developers are working towards a beter domain model, some documents on the wiki seem to suggest that much.

Yes, the object model is a complete mess. We have been designing a new backend for the app that should resolve most of this. You are right, java isn't the problem, the code is the problem. That being said, PCGen is *way* faster then it was 2 years ago. We have been fixing and polishing the code for some time, and it is certainly paying dividends.

Here's the problem with lazy loading - you never know which rules will impact others. The gaming system wasn't designed to be 'loaded' piecemeal, but we don't in general notice, because humans are really good at dealing with this kind of thing. That being said, we are moving towards a somewhat different paradigm - you may still have to load all the rules you want when you make a character, but we are working on a method to allow the character to load *only* the data the character uses. This has a number of benefits, including the future ability to load characters from multiple game modes at the same time.

Where we are going has been my aim for some 2 years now, but it took a long time for us as a team to get the code up to a level where we could make this kind of architecture change. It's still not quite there, but its *much* closer.

As for UI speed, we are looking at a number of new ideas in that space. We may well dump swing at some point in the future, and use something else a little speedier. Swing is a genuine speed issue. It doesn't have to be, but there are simpler ways to make a gui that performs then swing.

Devon Jones
PCGen BoD
Architecture Silverback

 

[DethStryke]

I just want to point out here, PCGen may be very complicated, but I think it's fair to say that what it's trying to do is *absurdly* complicated. If people want a character generator for just the SRD, our solution may well be far to complicated. But we didn't set out to create a character generator that can just do SRD. We set out to build a character generator that will allow people to use rules from essentially any source (yes, we can't distribute data for everything, but if a person wants to do something with their character, odds are, PCGen can do it). So far as I'm aware, there just isn't another application that can model as many rules as PCGen can. Can we make the GUI better? Yes. Can we improve perfomance? Certainly.

Improving PCGen is something we have had to put a lot of thought into. Writing an app that can do what we can with the data *and* performs *and* has a simple gui is not the trivial problem you all seem to think it is.

I'm open to any suggestions, but the simple fact is, writing an app that will let a person create an epic psionic vampire wereboar, and do so accurately is in itself a difficult problem. Coming up with an interface that makes this easy borders on the impossible.

Simply put options increase complexity, and complexity increases perceived difficulty. It is *inherently* at odds to allow a person to create an epic psionic vampire wereboar, and at the same time consider interface simplicity to be paramount.

Now, I'm *not* saying that we can't make pcgen easier. But making pcgen *easy* is at odds with our chosen goal of making pcgen capable of performing as many rules as possible. (Really think about this. Really just sit back and consider that there are hundreds of OGL books on the market, and each has subtle new rule changes, and outright conflicting rules and rule changes. The actual possible combinations that these rules can make when used on a character probably number in the billions of billions.) There are plenty of applications that can make building a Player's Handbook character, there are less that can do the whole SRD. There (so far as I'm aware) is *one* that can handle almost any book thrown at it. The SRD only market clearly isn't our target market. I don't intend to be snarky at all when I say this: If you only need the rules from the PHB and DMG, PCGen was not designed for your needs. There are indeed simpler and faster tools out there for this simple subset of the rules. But when you buy that shiny new book from Green Ronin, and want to use it (or some of it) in your game, almost none of the other apps out there won't meet your needs anymore. Few of them will let you even enter new rules. You can enter the rules in PCGen. It's not simple to do so, but it does work, but likely you won't even have to, because PCGen may even ship with the book you are now using.

I invite you all to *actually* look at pcgen 5.10.1, because a lot of thought was put into it to make the interface more consistent, and for it to offer the user more clues to the user. I suspect many of you have used PCGen here and there over the past 5 years. The application has had it's ups and downs, but it has continued to improve, continued to accept more and more rule variations, and *continued to get faster and easier*. Before 5.10.1 shipped, we put a lot more polish on it then any previous release. We standardized the windows in it, added search boxes just about everywhere, added clear tips as to what was unfinished with your character, and a number of other UI improvements. *yes* it is still the same basic UI, but it it a hell of a lot cleaner then 5.8 or any previous version. And in the future, 5.12 will be cleaner still, and on to 6.0.

And to those who scoff at us asking you to come help, You don't have to be a programmer to help. You don't have to edit lst files to help. You don't have to document to help. Even filing bugs, or pointing out specific elements that are flawed helps. Come over, join our lists, and throw stones constructively. Like products put out by businesses or not, all of you have unparalleled access to the people who make and maintain PCGen, and we do it *for you*. If you want a better character generator, you have the forum and the access to the people to make it happen.

Devon Jones
PCGen BoD
Code Architecture Silverback

I was not trying to insinuate what the program was trying to accomplish is trivial or the work you are doing to harness the system was easy. Regardless, the bottom line is that if you want to have a successful product there are certain markers/goals you have to meet/accomplish/overcome. Those are the items I was pointing out. I'm not saying that it isn't something that would take a huge amount of time, brilliant programming and very reliable people. Perhaps this is why some have brought up the question as to whether what would need to be done could be accomplished in a Open-Source manner.

I have used 5.10. The UI changes and improvements you mentioned are a noted improvement from 5.8 and previous versions. I was dismayed that the WotC sourcebook would only work with 5.8 and that the contract problems they had with WotC would mean that no other sets would be forthcoming. I even tried manually adding just the feats and such for what I needed, but you could only make custom magical items per the DMG with the official sourcebooks... so I was stuck with the older version. That's just my experience. If I'm being limited to an older version, what does it matter if the new version is all new and shiny?

That being said, both versions still ran sluggish with a huge amount of available memory allocated to it. It is a resource hog in the extreme. You point out that PCGen's "intended market" is not the basic SRD / Corebook crowd. Well, that's great. When you load more than 1-2 sources it sucks up so much memory and processor resource that you can't do anything in less than thirty seconds per click. Gods help you if you want to work on more than one character at a time, or have followers/familiars/companions/etc. I would say that this is a show-stopper for your intended audience. If you have show-stoppers like that in existence, then you have not fulfilled your intended goal for the project. To say otherwise is marketing and blowing smoke. I would recommend polishing that stone A LOT more before it is a viable product that should see wide-spread use.

You use the term "clues" as to what to do in the GUI; that is a troublesome thing. I should not have to decipher the code of your GUI like a scavenger hunt, never knowing where I should step next to arrive at my destination.

While this sounds harsh, please consider that the original question was "what is a simple PC Generator program for D&D?" Since PCGen is free (yay!) and fairly well known having been around for a few years, it of course came up.... but not as a very widely backed suggestion. The above, and I'm sure a few others, are the reasons for that. Now you can say all that you did to justify why these problems exist in the program, or you can take control of them and work towards a better tomorrow for us all, but the end result will still be that a majority of people have this view of the program - a view that will not change until the program does on those points. You can even write off those of us who think this, claiming that we're not your intended customer... but that won't change my view on it.

I, for one, wish you great luck and speed in making the program better. I think you and your co-monekys have the best chance in the market today of making it happen! However, sugar-coating the truth of the matter and cooing over flash in the pan when there are these 800-lb gorillas in the way doesn't strike me as something that helps anyone here. The point of the thread was asking for suggestions and critique of available generators for D&D. I kept all of my points to things that are specific, accurate and personally experienced - I'm not trying to make things up just to slander or belittle the product and team here. As I've said before, I actually do like and have used the program... the mentioned cons simply outweighed the pros.

Edit- outweighed the pros.... for now. I'm always willing to check out future releases and changes. Hopefully these benefits and changes you've mentioned do see the full light of day in all their glory, and then I will praise PCGen for overcoming those hurdles and use it all the time.