[Fixed] Crash w/ ultramarine.wad

Biff · Post by **Biff** » Tue Aug 05, 2003 6:28 pm

I repaired the map with DeePsea, there are unused sectors, sidedefs, overlapping lines, etc. WA can't even open it. Node rebuild took a portion of a minute. It now plays with zdoom, no lags, no lockups. I don't like this map at all, lots of work in it but gameplay?

The Ultimate DooMer · Post by **The Ultimate DooMer** » Tue Aug 05, 2003 6:40 pm

Biff wrote:Yeah, the crash may have corrupted my system, it's still win98 and therefore weak.

Could this be the same crash that has screwed up my win98 system? (although I ran zennode on it before playing and got the lockups then but not the 'very fatal error' thing. I never had any lockups on any other wads before I did this) I've not opened it up in any editors except Wintex.

I ran scandisk and reinstalled ZDoom 47 (with the 47i exe) and it still locks up on any wad.

BTW, can you send me a copy of the fixed wad anyway? (I want to get it in newstuff this week if I can get it working)

Edit: Panic over - I reinstalled ZDoom 47.cab and it seems to work fine now. But it does mean I can't use any future versions of ZDoom until that far off day I get a new PC. (Either that, or Randy improves the code to make this problem disappear in 48.cab when it comes)

arcticwolf · Post by **arcticwolf** » Tue Aug 05, 2003 6:58 pm

Kappes Buur wrote:
The Ultimate DooMer wrote:After a looooooooooooooong zennode build ...
No kidding

Mmm. Time's actually not that much a problem IMO; what surprises me more is that ZenNode is currently using more than half a gig of main memory on my machine rebuilding those nodes. I don't have any experience with ZenNode, though (only did a few levels myself back in 1994, and used bsp for those), so... is *that* normal?

Biff · Post by **Biff** » Tue Aug 05, 2003 8:43 pm

[/quote]
BTW, can you send me a copy of the fixed wad anyway? (I want to get it in newstuff this week if I can get it working)[/quote] OK, it's sent to your hotmail account. I hope there are no bugs, I went at it aggressively with DeePsea. Deleted unused sectors, unused sidedefs, set one sided lines impassible...etc etc.

Edit: Sigh, I couldn't handle a quote within a quote.

HotWax · Post by **HotWax** » Tue Aug 05, 2003 8:53 pm

Graf Zahl wrote:You should have used zennode -rz to skip the reject build. This alone probably took 17-19 minute of your time...

I'd say more like 19.92268333 minutes, myself... Just a hunch.

Post by **Graf Zahl** » Wed Aug 06, 2003 2:23 am

arcticwolf wrote:
Kappes Buur wrote:
The Ultimate DooMer wrote:After a looooooooooooooong zennode build ...
No kidding

Mmm. Time's actually not that much a problem IMO; what surprises me more is that ZenNode is currently using more than half a gig of main memory on my machine rebuilding those nodes. I don't have any experience with ZenNode, though (only did a few levels myself back in 1994, and used bsp for those), so... is *that* normal?

For Zennode, yes! This program has utterly horrendous memory management and makes some pretty bold assumptions about the contents of a WAD. It frequently allocates way too much memory and if some values are too far off of what it expects it crashes. The biggest problem is that it starts its memory calculations from the least reliable resource in the whole WAD - the SECTORs. This can create lots of problems with WADs that do a lot of sector merging like the 10-sector-WADs for example. It frequently crashes on them.

arcticwolf · Post by **arcticwolf** » Wed Aug 06, 2003 6:12 am

Graf Zahl wrote:
arcticwolf wrote:
Kappes Buur wrote: No kidding

Mmm. Time's actually not that much a problem IMO; what surprises me more is that ZenNode is currently using more than half a gig of main memory on my machine rebuilding those nodes. I don't have any experience with ZenNode, though (only did a few levels myself back in 1994, and used bsp for those), so... is *that* normal?

For Zennode, yes! This program has utterly horrendous memory management and makes some pretty bold assumptions about the contents of a WAD. It frequently allocates way too much memory and if some values are too far off of what it expects it crashes. The biggest problem is that it starts its memory calculations from the least reliable resource in the whole WAD - the SECTORs. This can create lots of problems with WADs that do a lot of sector merging like the 10-sector-WADs for example. It frequently crashes on them.

Mmmm - so much for the better nodebuilder. Not that I don't like ZenNode - I do -, but it kinda makes you wonder why there hasn't been an update to it in over a year, fixing these problems.

Well, at least it's GPL'ed, so someone more knowledgeable than me might be able to take care of that themselves (TaoNode? <tt>:)</tt>). I might take a look myself, too, but more for curiosity than anything else; I never was good with low-level languages like C etc.

Getting back to the original topic, though, the level works for me now (and the complete rebuild of everything, including the REJECT tables, only took 5:30 minutes, too - go fast computers <tt>;)</tt>), so I'll finally be able to check it out. It looks promising so far...

The Ultimate DooMer · Post by **The Ultimate DooMer** » Wed Aug 06, 2003 7:14 pm

Biff wrote:

OK, it's sent to your hotmail account. I hope there are no bugs, I went at it aggressively with DeePsea. Deleted unused sectors, unused sidedefs, set one sided lines impassible...etc etc.[/quote]

Got it, spent ages playing it too. Can you upload it to your site so I can link to the fixed version instead of the buggy original? (for the review). Or maybe see if Ty will accept the fixed version on the archives?

Biff · Post by **Biff** » Wed Aug 06, 2003 7:21 pm

I'll ask Ty what his policy is and send it to him if he's agreeable to just replace it. Otherwise, I could put it on my web space for a while, I'll let you know, and if Ty won't take it, I'll upload it to my space and state a link to it here by Friday evening.

Post by **Ty Halderman** » Thu Aug 07, 2003 1:08 pm

Since this is a brand-new upload (just cycled out of newstuff, in fact) I'd rather not update it without the original author being involved. I scanned through the thread here and didn't see where anyone had discussed all this with "Fausto" yet.

I would doubt he'd have an issue, but I probably shouldn't allow other people to alter wads that are current in the archives. There's a lot of older stuff where the author doesn't exist or can't be contacted that I make exceptions for, but this one's from May (June for the English text).

Biff, your email to me pretty well explained the situation--why don't you send something like that to him asking permission to update (CC me if you would) and we'll take it from there.

/knew he'd get karma-smacked if he left the wad with a long name

Biff · Post by **Biff** » Fri Aug 08, 2003 8:15 pm

Ty, thanks for the suggestion. I've done that (mailed him) today with the proposal and the fixed wad.

Ultimate Doomer - for now, I'll put the repaired version here so you can do the review:

http://home.earthlink.net/~mjblakely/da ... marine.zip

I'll take it down if Fausto agrees to allowing an update of his version in the 3d archives.

Biff · Post by **Biff** » Sun Aug 10, 2003 9:49 am

Status:

1. Fausto has agreed to using the repaired ultramarine.wad to overwrite the copy he put in the 3d archives.

2. Ty probably hasn't seen the permission yet and hasn't put the repaired file in the archives yet.

3. The review on doomworld points to the broken 3d archives version for some reason, instead of to the repaired file at http://home.earthlink.net/~mjblakely/da ... marine.zip
so don't use the link on doomworld.

Post by **Enjay** » Sun Aug 10, 2003 11:01 am

I haven't seen this mentioned in the thread: The Zdoom internal builder seems to cope OK with the WAD.

I noticed in the text file says:

"IT DOESN´T RUN WITH DOOM2.EXE, TRY OTHER SOURCE PORTS, BUT WITH ZDOOM GL IT RUNS PERFECT, A LEAST IN MY COMPUTER"

So I'm guessing ZdoomGL is building the nodes (which I think happens automatically).

Anyway, here's what I got at the console using Zdoom 47i

Map has 72 unused sidedefs
BSP generation took 2.202 sec (32233 segs)
REJECT is 5714 bytes too small.
Sector 2033 (tag 0) has no lines
Sector 2853 (tag 0) has no lines
Sector 2854 (tag 0) has no lines
Sector 2855 (tag 0) has no lines
Sector 2856 (tag 0) has no lines
Sector 2857 (tag 0) has no lines
Sector 2858 (tag 0) has no lines
Sector 2859 (tag 0) has no lines
Sector 2860 (tag 0) has no lines

After that, the map seemed to play OK, although, IMO it would have been better as a few smaller maps. It looks impressive, but it's not the most fun to play map I have tried.

The Ultimate DooMer · Sun Aug 10, 2003 11:49 am

Biff wrote:3. The review on doomworld points to the broken 3d archives version for some reason, instead of to the repaired file at http://home.earthlink.net/~mjblakely/da ... marine.zip
so don't use the link on doomworld.

There is a link to your version at the end of the review, the main link is pointing to the archives, where the fixed version should be soon.

Biff · Post by **Biff** » Sun Aug 10, 2003 12:10 pm

The Ultimate DooMer wrote:

There is a link to your version at the end of the review, the main link is pointing to the archives, where the fixed version should be soon.[/quote]

:O I missed it, thanks. The review basically says "It crashes", which you realize after you check the "screenshot", so I couldn't imagine the review would say that and yet point to a repaired version.

Edit: Oops and my apologies, UD. My lameness is now revealed. That was last week's review stuff I was reading and I didn't even realize it, not having looked into the reviews for quite a while. Of course I see your new submittals now, they weren't up at the break of day here. Ah yes, I see whose fault that is.