[Roma Loom]

also launching mapster32 in fullscreen on dual-monitor system makes the secondary desktop look like "8bit corrupted palette total fuckup". On another note - there's a switch on the ceiling in e4l3 you have to shoot at which is invisible due to maphacks haven't been implemented in polymer.

[The Commander]

Marked, on Apr 20 2010, 11:38 PM, said:

I set it back to 4 only when I am actually placing and adjusting some light effects. Then its right back to 3.

That is what everyone should be doing IMO, trying to run Mapster in constant Polymer mode will be asking for trouble.

[Sobek]

Well I sorted out most of my map's performance issues - though I still don't understand what the problem was anyway (had to delete some random sprites as they were destroying the frame-rate throughout the entire map).

There is however one old 'issue' that I still struggle with... Mapster32 only uses about 200-300mb of ram at any given time (this system has 1gb of ram with most of it free, being a really slimmed down XP install), and yet even when there's a good 400mb free, it seems to start hammering my hdd every 5 seconds or so, absolutely killing the frame-rates and causing a brief pause each time it does. We talked about this once a while back but I don't recall any solution ever being found.

It'll start doing this when I'm just sitting still, doing absolutely nothing... It's like a memory leak only, not. And when I quit after this, my system acts like it's been hammered for hours on end by some incredibly intensive app - mapster32 takes forever to exit, the hdd is just going nuts and even opening an explorer window is extremely slow for a decent while after. It's like when you've been playing some 'heavy' game for hours on end, and your system needs a short while to recover afterwards.

Only this is Mapster32, and it's barely touched my system resources.

[Micky C]

It doesn't seem like anyone else has mentioned this pretty damn big error:

A while back I discovered mapster32 now has an undo function by hitting control-z in 2D mode. The problem is if I use it, it randomly deletes every single sprite in the entire map. Most of the time it works correctly, but when this error happens, it's fatal.

It must have happened at least 5 times: 3 of those were easily fixed by reverting to autosave.map, but twice I didn't notice it in time before saving. Luckily one of those times I had a decent backup, and the other time it was relatively early stages of a new map (but still a big setback)

Has anyone else noticed this? The undo is a handy feature and I'd hate to avoid using it, or use any other work around.

[Gambini]

I did know that undo thing was unsafe. I think i´ve used it once and never more for the reasons you just mentioned.

I rather save often and go back to the saved file when i commit a mistake.

[TerminX]

Helixhorned, I noticed a couple of issues:

Quick map cycling is broken
F1 help browser in 2D mode is broken

Just those two for now.

[Helixhorned]

Fixed 'em, but found some new ones in the process (specific to my system, which is 32-bit Ubuntu 10.04):
- Sound seems to be broken: it's very "buzzy" and has an echo attached.
- Various SDL and/or malloc-related crashes: 640x480 and 800x600 don't work at all for me, giving the following:

mapster32: malloc.c:3096: sYSMALLOc: Assertion `(old_top == (((mbinptr) (((char *) &((av)->bins[((1) - 1) * 2])) - __builtin_offsetof (struct malloc_chunk, fd)))) && old_size == 0) || ((unsigned long) (old_size) >= (unsigned long)((((__builtin_offsetof (struct malloc_chunk, fd_nextsize))+((2 * (sizeof(size_t))) - 1)) & ~((2 * (sizeof(size_t))) - 1))) && ((old_top)->size & 0x1) && ((unsigned long)old_end & pagemask) == 0)' failed.

Also crashes when changing to 3D mode, although I can't reproduce them reliably.

[The Commander]

Quitting Mapster32 in 3D Mode always seems to result in an error.

[The Commander]

Clicking "Cancel" on the start up window for Mapster32 always causes a "Mapster32 has encountered a problem bla bla bla"

[Forge]

Snapshot 1657. So far things seem to work fine except: I can't change the gridextent. I change it to 393216 in mapster32.cfg, but when I launch mapster, it automatically reverts it back to default.

[The Commander]

Forge, on Jun 8 2010, 12:31 AM, said:

Snapshot 1657. So far things seem to work fine except: I can't change the gridextent. I change it to 393216 in mapster32.cfg, but when I launch mapster, it automatically reverts it back to default.

262144 is the highest it will go for me now, but that is higher than the default of 131072.
The biggest size I think that it could allow was 524288 from looking at an older build.

That said, I don't think anything was done in the r1657 shot mapster related as it still errors when you quit from 3D mode or click cancel in the start up window for mapster.

[TerminX]

Yes, editorgridextent was lowered. For a reason. It didn't need to be so large that errors cropped up based on where exactly you decided to put your sectors. 1657 has multiple fixes for crashes that actually occur within the editor, not for bullshit like deciding to cancel at startup.

[Forge]

That's cool. I really didn't need all that extra space. The only issue I really had was I had started a map that was out of bounds for the new limit. I re-installed an older version, moved the map back within limits, re-installed the new version. No big deal. I didn't know if the limit was on purpose or a bi-product of something else and needed to be addressed.

btw: I'm not experiencing the errors about shutting down or canceling a launch that have been reported by other users in either full screen or windowed mode.

[Forge]

I'm not hex code savy, so maybe somebody can tell me what the value is supposed to be.

snapshot 1657

In 2d mode both KP4 & KP6 lengthen sprites. KP4 is supposed to shorten sprites.

mapster.cfg

KP4=4B and KP6=4D

[Helixhorned]

4b=4*16+11=75
4d=4*16+13=77
but you should never need them since you enter hex values into the remap string (if that's what you're concerned about).

[TerminX]

Speaking of that bug Forge mentioned, "svn blame" says you introduced it in r1644. I just looked at it and it's caused by shifting 2 places on line 6590 and 6592 instead of 1.

[Helixhorned]

Ah, then take my hex calculations with a grain of salt

[Forge]

r1662 & r1663

Issue: SE (xx, 12) Switch activated lights
&
SE (xx, 3) Ceiling lights flicker when shot

SE12 works normal as long as the SE12 and the switch are in the same sector. If there is an SE12 in a different sector than the switch, then the first time the switch is activated (lights on), it keeps the sector shaded as if the lights were off. If the switch is activated again to make the lights go off, the the SE will light up the sector as if the switch were on. Somehow the SE12 that is outside the sector of the switch skips a cycle then gets on opposite cycles of what the switch is on. (There was only one wall switch to activate the lights).

SE3 works randomly. Sometimes it works just fine. Sometimes if the light is shot, the sector will dim, but the sector may or may not flicker.


have not tested SE (xx,4), but the "C" cycler works flawlessly

Further testing results:

SE12: after flicking the light switch several times the SE12 that is not in the switch sector goes into "turned on" mode and stays there no matter what position the light switch is in.

SE3 is a little more complicated. To simplify it picture a flat, straight hallway with three sectors. The first sector is where the player enters. The second sector has a touchplate switch to activate explosions that break the lights in the second and third sector. The second sector has a normal SE3, the third sector has a pal SE3 to accommodate for the color of the fire.
With the touch plates in the second sector, the second sector's flickering will not work, but the third's will. If the touch plate is put into the first sector the second sector will flicker, but the third will not. If the touch plates are removed and the lights are shot out, they will not flicker. Somehow flickering shot out lights are tied to switches. The switch and the shot out light can't be in the same sector or two sectors apart, but must be in separate, adjoining sectors.

Further experimenting proves that it doesn't matter which is where and what's is it what sector. One time all flickerings will work, other times only one will work, and sometimes none of them work. This is an eduke quirk and not a mapster issue.

[bioman]

Fail to open 'NAMES.H' with Mapster32 rev 1707 / Linux 64 bits.
Not a case issue, was ok with Mapster32 rev 1671.

mapster32.log :

Quote

Mapster32 2.0.0devel 20100521
Failed to open NAMES.H
Using /usr/share/games/eduke32/ for game data
Using /home/bioman/ for game data
Using /home/bioman/.eduke32/ for game data
Initializing SDL system interface (compiled against SDL version 1.2.14, found version 1.2.14)
Using "x11" video driver
Loading "m32help.hlp"

[Helixhorned]

Yep, that was truly a bug. It should have affected anyone having their names.h not in the same directory as the executable.

I'd like to crack down at the not Polymer-related Mapster crashes. Can anyone experiencing them try this exe (SHA1 checksum: ca6cd2b9516cc40f507ac9ac41a8417ee5eb342d) and report backtraces after the crashing? The build is made without stack protectors, so you might actually get less crashes than with the official ones. Those cases should be reported too since they indicate serious corruption. I think the stack protector library prints out diagnostics, so then you could start the official Mapster32 from a console (for example cmd.exe) and report the output printed there. The mapster32.log file is pretty useless.

Edit 20150218: made URL point to an invalid location.

[Stabs]

major problem i have been having with newer mapsters lately is it has an instant CTD when i use a or z to zoom out or drag right click or generally doing anything when it comes to moving around in 2d mode

[Stabs]

did alt + kp1 get changed for modding the x y z vel values?

also could be a problem with my numpad, when i change the color with num 9 its also pgup on my keyboard and the sprite gets raised

[Helixhorned]

Yeah, Alt-KP1 should work. The other issue should be gone with toggling NumLock.

Since nobody reported backtraces I guess people are reluctant to
1. run unofficial binaries
2. install GDB
(or both)?

[Mark]

I probably don't know of or use a whole lot of the features in Mapster so I am not doing some of the things that crash it for others.

When Polymer is disabled, I can't remember the last time I had a problem. And with Polymer on, I had a one-time glitch which corrupted my map and the occaisional closing of the program for no apparent reason.

So its been way more stable for me these days. I hope I haven't jinxed it by saying so

[Stabs]

if you want to give some detailed instructions on using GDB i will use it

[Hank]

Helixhorned, on Oct 13 2010, 03:16 PM, said:

Since nobody reported backtraces I guess people are reluctant to
1. run unofficial binaries
2. install GDB
(or both)?

Yes and no. I got the GDB, but whenever Mapster crashes, my Visual Studio mouths’ off. And I need this IDE right now undisturbed. - Meanwhile, where I am, winter is coming, during the first major snow storm, I'll download the source code page by page, (it does not come in a lazy man's zip or tar ball), compile it and run it through the mill. I still know shit about C, I am only good in C++, so I need to fudge it a bit, but I can post the debugging info, and you’ll take it from there.

[Helixhorned]

DanM: all you need to do besides installing GDB is to follow these simple instructions. So, start cmd.exe, cd to Eduke32 directory, type "gdb --args ./mapster32.exe", type "r", wait for the thing to crash, type "bt" and copy that mess to the clipboard.

Hank: maybe try disabling Windows' interception of crashes? It's in the "My Computer" properties, about the same place where you can set environment variables. At least in XP...
What do you mean by "page by page"? Duke does come in an SVN repository nowadays...

[Hank]

Helixhorned, on Oct 14 2010, 11:26 AM, said:

Hank: maybe try disabling Windows' interception of crashes? It's in the "My Computer" properties, about the same place where you can set environment variables. At least in XP...
What do you mean by "page by page"? Duke does come in an SVN repository nowadays...

I went to http://sourceforge.n.../eduke32/files/

IDE says: The OPTIONS responce bla bla bla URL is not WebDAV-enabled. (No clue what this means!) What command is the server looking for? Am I in the right URL? Do I need to be a member at sourceforge? - Also, to make Windows 7 to shut-it, is like getting blood from a stone. b.t.w. the new mapster does rock! I can live with the crashes for now, it's like in the old Build days and ME

[Plagman]

That's the URL you're looking for, click on "Download GNU tarball".

http://eduke32.svn.s...viewvc/eduke32/

[Hank]

Plagman, on Oct 14 2010, 11:50 PM, said:

That's the URL you're looking for, click on "Download GNU tarball".

http://eduke32.svn.s...viewvc/eduke32/

I think I need new glasses.

update added 16:15 15.10.2010:
Eduke32 can be compiled and played with, using Visual Studio Pro 2010, Microsoft DirectX SDK (June 2010) and oggvorbis on Windows 7. The catch – move the entire eduke32 source codes to C:\user\yourName[admin]\ so the compiler can have free reign reading and writing DirectX lib routines. (Yes, there are alternatives)

Now let’s see how long it will take for the old fart (me) to learn C.

p.s. I remember trying to compile Build into Watcom, I gave up, after like one million error messages, and crashes. Now - what a difference, one afternoon, and still time for a beer. Wow! Thanks again guys.