Castle Paradox

Bob the Hamster · Posted: Fri May 04, 2007 7:15 am Post subject: Bug of the Week (Vote!)

As you all may know, the upcoming stable release of the OHRRPGCE, 'ubersetzung', has been delayed and delayed and delayed. However, ubersetzung is so featureful, that most OHR developers are already using it-- which makes it all the more important to stomp out the remaining bugs and get it officially stablized. There are a number of bugs holding it back but some of them especiall need the help of the community-- and that means you, non-programmers!

Until the release, we will be declaring "Bug of the Week" each friday. These will normally be determined by popular vote, but I am unilaterally declaring '' . bug_title('315') . '' to be our first BotW. Go there, read about it, and see what you can do to help.

Although ubersetzung is already released, I would like to keep bug-of-the-week going. I have enabled the voting feature on bugzilla, so you can vote for your favorite bug if you log-in there. You can also see a list of what people have voted for.

Reply to the bug with any information that might help solve it, and reply here with non-bugfixing-related comments and votes on next week's Bug of the Week.

Fenrir-Lunaris · **WUT** Joined: 03 Feb 2003 Posts: 1747

I can't comment on that particular bug, but I've encountered a potentially GAME BREAKING glitch that occurs sporatically in the 5-4-2007 default nightly build related to .Wav files apparently. I've had this error repeat somewhat inconsistently, but it seems to crop up with ogg and wav sound effects in battles without any preference for either.

The bug consists of having an attack with a sound effect, which when triggered automatically ends the battle as an undesired effect. Upon returning to the map, the player is THEN unable to access their menu (preventing them from saving or quitting, though save-state works), and any future battle/random encounter/preset battle simply fades to black, then returns as though no battle took place at all.

I've also provided a test-case, which also makes for a convenient excuse to plug my epic RPG I presumably dropped two years ago to work on Vikings: Shameless advertisement

Ysoft_Entertainment · Posted: Sun May 06, 2007 6:25 am Post subject:

I don't know if its related or anything but both ohr multi-pack and SCHBW crash whenever I try to play it, I checked the ohr multipack it contains ogg files, however I am unable to verify if they are the ones that crash the game, it would be good idea to look into the scripts.

The nightly build that I am using is 4/28 fb native
_________________
Try my OHR exporter/importer.
OHRGFX
Striving to become better pixel artist then Fenrir Lunaris. Unfortunately the laziness gets in the way of my goals.

Moogle1 · Posted: Sun May 06, 2007 7:51 am Post subject:

SCHBW has an MP3. Try using the included EXE, which is slightly more recent. Also, what OS are you using? You're the first to complain about crashes, so it may be Linux-specific if you're running that.
_________________

Slime Salad :: Independent Game Design

Ysoft_Entertainment · Posted: Sun May 06, 2007 8:37 am Post subject:

I am running xp pro
I tried playing it with the included, however the game crashes when the song after the title screen plays. I happens when I press the any buttons after the title screen.
_________________
Try my OHR exporter/importer.
OHRGFX
Striving to become better pixel artist then Fenrir Lunaris. Unfortunately the laziness gets in the way of my goals.

Moogle1 · Posted: Sun May 06, 2007 10:27 am Post subject:

Interesting, so it doesn't crash if you don't push anything after the title screen?

It is almost definitely not the script for SCHBW. All the script does is palette changes, hero swapping, and some tricky math. It could be the palette junk, but that seems very unlikely.
_________________

Slime Salad :: Independent Game Design

Uncommon · Posted: Sun May 06, 2007 3:04 pm Post subject:

There've been a lot of random crashes relating to sounds. It's what kept me from playing Eyes, I'm assuming. I ran into a few while making Mazes Of Persistence, and JSH had troubles while making Mending Wall and had testimonies of it crashing when sending out early copies.

The biggest problem with tracking this bug to provide a testcase is it appears to be inconsistent between computers. I haven't heard any problems with Mazes Of Persistence post-release, but it's just weird. But maybe it has something to do with mp3s?

Ysoft, you run into any crashes with Mending Wall or Mazes Of Persistence? Or Eyes even? Most of the TGC entries use sound, so they could be the best testcases we've got. JSH I think cut all the mp3s out because they were crashing on him, but Mazes has one on the title screen and at the end, if you can hold out long enough to get to it.

msw188 · Joined: 02 Jul 2003 Posts: 1041

In reference to Bug 315:

First off, what exactly should we do? Should we download the latest nightly, and just try importing wavs into custom? Should we report on Bugzilla whether or not we get crashes, or should we only report if we DO get a crash?

I read on there that you thought there might be a memory overwrite problem of some sort. Is there any way to test this particular theory? I don't know anything about computers, but it seems that if one were to make an RPG file that was really really full (as in, using all of the available pictures, text boxes, etc) it might be easier to spot overwritten memory. Or am I really off-base here?

After typing this I will download the nightly and try importing some wavs. If you think that a super-full RPG file might actually help, reply and I will make one, but I'm not sure how long it will take (just the thought of making 35,000 some enemies with all of their naming and statistical slots filled makes my existence hurt).

EDIT: Okay, I downloaded the default nightly, imported two wav sound effects (actually the same wav twice), went back and forth with the browsing menus several times, and got no crashing. The only thing that seemed strange was that when you back out of a sound effect browse menu, the sound effect in question stops playing, but when you back out of a song browse menu, the song does not stop playing until either you switch what number song you are affecting, or go back to the main menu. Now here is a question: do you want these kinds of replies on Bugzilla, or do you only want us to reply if we can confirm the existence of the bug?
_________________
My first completed OHR game, Tales of the New World:
http://castleparadox.com/gamelist-display.php?game=161

This website link is for my funk/rock band, Euphonic Brew:
www.euphonicbrew.com

Bob the Hamster · Posted: Mon May 07, 2007 7:37 am Post subject: How to help

The biggiset thing we need here is simplified test cases to reproduce the crash on-purpose.

But both positive and negative results are worth reporting. I would like to know:

* Which music backends did you test? (native/native2/sdl/silence)
* Which ones crashed?
* Which ones didn't crash?
* Can you reproduce the crashes consistently?
* Were the crashes on sound effects or music?
* What formats crashed? (bam/midi/wav/mp3/ogg/other)
* What formats didn't crash?
* What operating system are you using? (98/xp/vista/linux/other)

You can post all that stuff in the bug report.

Uncommon · Posted: Mon May 07, 2007 10:36 am Post subject:

One thing'd be immensely helpful is if, on crash, game.exe wrote what crashed it into an output file, like custom.exe did that time we figured out that ohrrpgce.fnt was missing. All we're seeing on our end is that it's crashing, but if we could give back more information, you'd be better equipped to fix the problem.
All I'm saying is help us help you.

Like for instance, I can tell you that the game Eyes consistently crashes on me at a certain point in the intro over music_native on Windows XP, but the cause of which would be purely speculation because what ever's triggering the crash crashes while it's being triggered itself. Though perhaps someone for whom the game works could tell you what happens after the hero walks up to the first dead girl upon entering the house.

Bob the Hamster · Posted: Mon May 07, 2007 12:59 pm Post subject:

A file called g_debug.txt in the same folder as game.exe will contain some crash messages, but not everything will be written there. Many types of crashes aren't really practical to capture in that way.

I'll try to think of some way to get better debugging information-- ideally we would all test through a debugger that does a stack trace on all crashes-- but that ain't easy to set up :(

Moogle1 · Posted: Mon May 07, 2007 2:12 pm Post subject:

I saw a g_debug today and opened it up - it was about 1.7k of lines that looked like this:
math 18
math 3
math 0
etc.

I stuck my thumb on the PgDn key for a few seconds, then closed the file. You may want to suppress the "math" message, unless it's somehow useful and I'm missing something.
_________________

Slime Salad :: Independent Game Design

msw188 · Joined: 02 Jul 2003 Posts: 1041

I put the rest of my results (which were pretty much nothing) on Bugzilla. I hope that Chaotix sees this and is able to report whether or not he/she is still seeing this bug.

I know it is not listed as a blocker, but Bug 178 seems like it ought to be such a simple fix that I would nominate it for a Bug of the Day (and if a test case is needed, I will provide it). But if we are restricted to the blockers as defined by Bug 255, I think I'd vote for Bug 304. I feel like I caused extra delay on that one, but really I think any of the solutions proposed near the end could work (except the bit about breaking the overhead bitset when using color zero), we just need to pick the one that makes most sense. And it shouldn't be that much of an issue anyway (besides backwards compatability, as in my exception above), since all passability can be set manually if need be.
_________________
My first completed OHR game, Tales of the New World:
http://castleparadox.com/gamelist-display.php?game=161

This website link is for my funk/rock band, Euphonic Brew:
www.euphonicbrew.com

TMC · Posted: Tue May 08, 2007 1:54 am Post subject:

Unfortunately, using a debugger with FB is difficult anyway, as it doesn't seem to export debug information for module level or global variables, a lot of type information seems to be missing, and of course FB arrays have to be indexed manually.

What we really need is array bounds checking! Maybe FB0.17 offers some hope.

Rinku · Joined: 02 Feb 2003 Posts: 690

If this is bug 315, and you guys fix a bug a week, that's... 6 years before Ubersetzung (assuming there's no bug after bug 315)!
_________________
Tower Defense Game