Post by Radio PARALAX in Castlevania AGA comments

Viewing post in Castlevania AGA comments

I updated the original version from February 20th (at the time of preparation for the broadcast) to the latest version during the livestream after two failed attempts. Funnily enough, the version was updated during our stream, so that the version from the last attempt was only 45 minutes old, see position 04:32:53. Unfortunately, this didn't help and I couldn't use the CD32 version because the trainer didn't work (keys didn't work during the game). I will definitely take another look at the game later and adjust my rating accordingly. It is and remains a damn good and technically very high-quality game. And sometimes a livestream like ours can be very helpful for troubleshooting, as our example shows. It would not be the first time that another update follows shortly afterwards. 🙂

The game was tested with WinUAE 5.2.0 (not the most recent version, as this apparently contains changes and a few small errors in sound playback). This runs with a realistic, cycle-accurate A1200 standard configuration with HDD, Kickstart 3.1, 2 MB Chip- and 8 MB Fast-RAM without a turbo card in order to simulate the most realistic setting as possible. The color palette under the Workbench (OS 3.1) was set to 256 colors by default, which requires a bit more Chip RAM and there are actually games like "Bunnito's Feast" that need more to start. However, based on previous knowledge, it was not to be expected that the error would also occur during gameplay.

Edit: The bug also occurs in the (now) current version from 26.2., even with the color palette reduced to 2 colors under OS 3.1. To be on the safe side, it was tested again with the current WinUAE 5.3.1, and as expected, there were no differences. I would be happy to send you my WinUAE configuration that causes the problem so that you can troubleshoot it. If you are interested, just send me an email using our contact form: https://www.radio-paralax.de/kontakt/

Dante Retro Dev8 days ago (1 edit)

Oh, it was emulation, that's a relief. My focus is on crashes on real hardware. Emulation will never be 100%, for example, you can configure UAE to have 8MB of chipRAM, which is physically impossible on real world (you'll need to redesign the Amiga chipset and add extra addresses).

So, it enables strange stuff like that, resulting in strange and unexpected results that we developers can't manage to foresee.

Bunnito's Feast programmer was clever, I should have done that memory check too.

As the error keeps happening, I would please ask you to test the following:

Download the 'Castlevania AGA HDF' version from this page. Follow this quick tutorial that I just recorded for you on how to load the HDF file on your emulator (it's very simple).

This is basically like the CD32 version, but running on a disk image rather than a CD ISO. Think of it as an ADF floppy image. Don't worry, it has all the level skip shortcuts that you need to quick test the place with problem.

If that HDF works for you, in the way shown in the video tutorial, I know what's happening. And again, as a disclaimer, we developers can't guarantee to fix everything for all possible configurations around (as they're almost infinite), so, our focus is first in real hardware, then direct emulator configurations (like CD32 or the HDF shown above, getting rid of the Workbench which just jeopardizes things). The best Amiga games from back in the day also removed the system out of the way to squeeze the best performance and to have less troubles! Thanks.

Edit: my WinUAE is version 3.4.0, installed in 2016, and the game works here from start to end without any issue. As they say, "if it ain't broke, don't fix it". ;-)

Radio PARALAX8 days ago (4 edits)

I have just tested the game again as described with the HDF version and only one disk drive, unfortunately with the same result. I also noticed that the game does not run without additional memory of at least 64 KB (Z2 Fast). For the test I had set 1MB, a completely normal configuration for memory extensions at the time. I have attached the corresponding error message here as an image.

I would also like to add that I only played such games with 8 MB Fast RAM usually, just to avoid memory bottlenecks with WHDLoad and HDD games. Of course, the chip RAM was only preset to the usual 2 MB, see the Workbench memory status in our video. As far as compatibility goes, I rely on the statement of "Toni Wilen", the author of WinUAE. For years I have not found an OCS/ECS game or demo that does not run with it, which is why I now assume that it is 100% emulated. The AGA games and demos that I have tested over the last 4-5 years also ran perfectly with it, but of course I couldn't test all releases. However, there are sometimes configurations that do not correspond to the same setting of the computer on which they were programmed. That is why I am happy to offer to provide you with the configuration I use. 🙂

Dante Retro Dev8 days ago (1 edit) (+1)

Excellent, thanks, the error message pinpoints exactly what happened.

So, I did a special version for you that should work (let's cross our fingers). In this link.

It's a very rare issue! You can check on YouTube now, many longplays and live streams that didn't had that problem, like this one, for example.

Also, the problem doesn't seems to happen on FS-UAE, it only happens on WinUAE.

So, what I did was to reduce the number of candles and objects on the level by about 10 to 20% (the error message is about that: max number of objects/limit).

If it still crashes, I'll reduce even more! Don't give up! ;-)

Radio PARALAX7 days ago (5 edits)

Thank you very much for the quick update. 🙂 Unfortunately, the problem still occurs with exactly the same error message (tested with the HDF version). Have you really uploaded the new version to Google? Anyway, I used exactly the same configuration and teleported to world 5 with "F5" as before after starting a new game. Then I refreshed my energy with "H". Perhaps there is a problem related to this or with any driver like AHI that is preset in my OS 3.1. (I still can't really believe that it's because of WinUAE, since it supposedly works for you, even with an older version). This evening I will test the game in the regular download version on my real A1200 with Blizzard 1260 (68060 @ 50 MHz, 64 MB RAM), as this is also bothering me, and then complete this text accordingly. I would also like to send you my WinUAE configuration. However, I need your email address for this.

Dante Retro Dev7 days ago (3 edits) (+1)

I did another version, this time I increased the actor count. The error message says that the limit of 32 has been breached, so I increased that limit considerably. You'll see the number on the score panel at the top, right next to the word 'enemy' (it's temporary, just to make sure you're on the right version and can see the actor count). Link here.

(Edit: added new modifications around 03:55 AM GMT, please download again if you got it before).

I had reports about other issues from other players, but nothing on that stage. They're more related to CD32 version, which isn't your case. But let's see what happens with that increased number of actors on this new version I'm sending now.

Your WinUAE config isn't needed now, because I was finally able to make it crash on my configuration too. I also recommend that you avoid leaving too much candles and objects behind (you must destroy as much as possible) because they accumulate on the level and cause the limit to be more easily breached. If nothing works, I may try to compile the project on other versions of the engine, older or newer, until we find a version that works for you. Thanks again!

Radio PARALAX7 days ago (5 edits)

Unfortunately no changes. It still crashes with the same error message. Only the number in the "PC:" line changes after each crash. I have now tested the game in the regular, current download version (LHA) on my real A1200 via a PCMCIA CF adapter and... what can I say? Exactly the same crash at almost the same point in the first level of the 5th world. So it is definitely not a WinUAE problem.

I started the game on my real A1200 with only 2 colors in the Workbench from the CLI and got the same error message as on my WinUAE setting, see here:

Since I actually tend to avoid certain enemies when using trainers and not destroy everything on the screen, it seems that my gaming style has revealed this bug . 😉 In any case, I wish you the best of luck in fixing the bug and am happy to be available as a tester again at any time. And I also stand by my word when it comes to the final rating of the game. 🙂

Dante Retro Dev7 days ago (1 edit) (+1)

Definitely not your Wb or chipRAM, that were assumptions I took based on the most common errors described. Like an airplane accident, it's generally caused by a conjunction of factors: the limit on number of actors, your play style, and now a new discovery:

The bones launched by the red skeletors were not being removed when they go off screen, so, they were *also* accumulating. So, that was the modification I did on the post edit. But to confirm it further, one more test, this time I removed them completely (applying the Muntzing method):

https://drive.google.com/file/d/15F00WcMoPLme680Ers_JTh0AcENxTuiG/view?usp=shari...

Also in the plane accident analogy, at this point I'm completely exhausted, have put many sleepless hours on this issue, and I'm prone to commit errors at this point. Like exhausted pilots do when under absolute stress. So, I may even put wrong files up for your testing (this time is just the WHDLoad on the link above, to make sure). Sorry, but doing an Amiga game of this scope, for free as a single programmer, no matter how advanced are the current tools, is like swimming against a very strong tide, several miles from any beach on open ocean. All this for a free game on a very niche platform. That's important to have in mind when it's about game rating.

Radio PARALAX7 days ago (1 edit)

I understand that very well and have great respect for so much programming skill, but in the end I always have to base my overall vote on the finished product that I have at the time of the review during a broadcast. Take your time when looking for and fixing errors and give yourself a few days of rest. As you say, this is a hobby project for a free game and there is absolutely no time pressure, especially since the game is actually already finished. 🙂 Apart from that, even in the boom years of the Amiga there were serious bugs in commercial full-price games that could not be fixed with a patch without starting a costly exchange campaign (e.g. Turrican 1). I am very happy that there are websites and projects like this one where the community can provide active help, not just error tests and suggestions for improvement. Otherwise such great games and our live streams would not be possible in the first place.

Back to the game: Unfortunately nothing has changed with the current update, despite the missing red skeletons. The game crashes again in the first level of the 5th world. 🙁

Dante Retro Dev6 days ago(+1)

Well said!

But about the game status, it's "feature complete", that doesn't means it's "complete" as a full release. Perhaps there's a misunderstanding going on. It's technically a beta (all the content is in it, but it's still untested and has issues to solve).

Also worth saying, that the internet today is very biased *against* the Amiga. You can see issues and glitches on full-price commercial console games released (search for "Super Castlevania SNES moonwalk" on YouTube for example) but the internet seems to turn a blind eye to that - if that was on a free Amiga game, its author would be harshly criticized! Because there's a sort of herd mentality, mostly negative, regarding the Amiga today (internet "revisionism"), which is very frustrating and disheartening.

Finally, check the status posted on the devlog, your error has been reported there, work continues:

https://danteretrodev.itch.io/akumajou-dracula-aga/devlog/876838/status-and-repo...

Thanks!

View more in thread