Page 1 of 2
TEST: server code changes
Posted: 21 Feb 2013, 17:16
by o11c
I've pushed a bunch of code changes and built them on the server, after local testing seemed to indicate nothing major broken.
Please test by doing all quests, spells, and combat.
Still haven't fixed the biggest performance problems, but I've finally finished nearly all the steps I have to do before that.
Also, *somebody* did some local changes to the server-data again, so I had to force-reset it ...
---
Fixed:
- accreg.txt lost
- off-by-one buffer overflow in timestamping
- walktimer
- itenplz realloc
- party.txt_*.tmp
- skill level
- #chipchip verification
- crash on invalid @command
- healing rate
- chest armor
- crashes relating to status timers, such as weight over 50% or 90%, poison (shock sweet), ...
Reported: None
Re: TEST: server code changes
Posted: 22 Feb 2013, 02:46
by o11c
Whoops, I forgot to 'make install' ... it's a good thing, too: I accidentally wiped accreg.txt (which has all the bank accounts). Well, it *is* the test server, but still ...
Re: TEST: server code changes
Posted: 24 Feb 2013, 00:59
by o11c
Fixed a buffer overflow; this is now running on the test server.
Please test everything! I need to be able to verify this a stable so we can get rid of some lag.
Re: TEST: server code changes
Posted: 25 Feb 2013, 15:29
by v0id
When changing of chest armor / unequipping chest armor, the previous chest sprite isn't removed. However, the equipment change is effective regarding stats.
Re: TEST: server code changes
Posted: 25 Feb 2013, 15:49
by Chicka-Maria
also when i try to equip another shirt it seems the top layer is always my black short tank top...even with the dress it looks weird lol
edit: also other items like the dress..i put a screenie.
regards,
Re: TEST: server code changes
Posted: 25 Feb 2013, 15:56
by Frost
The map-server process on the testing server has been crashing very often. We should solve that before we push anything to the main server.
Re: TEST: server code changes
Posted: 25 Feb 2013, 17:07
by o11c
I'll look into the equipment issue.
Frost wrote:The map-server process on the testing server has been crashing very often. We should solve that before we push anything to the main server.
I would have expected the reverse, but this is definitely the biggest general change. It will need at *least* a major content release to be tested on it.
Re: TEST: server code changes
Posted: 25 Feb 2013, 18:45
by v0id
The character's focusing skills and magic schools has been all reset to 0, however the client's F5 window was showing level 9 in focusing skills and magic schools to 2.
Regarding the recent crashes, I ran a session in gdb and got this:
Code: Select all
map-server: src/map/pc.cpp:2562: void pc_walk(TimerData*, tick_t, int, unsigned char): Assertion `sd->walktimer == tid' failed.
Program received signal SIGABRT, Aborted.
0x00110416 in __kernel_vsyscall ()
Missing separate debuginfos, use: debuginfo-install glibc-2.12-1.80.el6_3.7.i686
This happened when a character entered in my character's line of sight.
Upon restarting the server, the focus skills and magic skills were back to 0 again.
Note magic experience is safe.
Re: TEST: server code changes
Posted: 25 Feb 2013, 19:03
by o11c
Okay, I know how to fix walktimer ... no clue about skills though.
Edit: walktimer should be fixed. It was a real bug, merely exposed by the newly fatal assertions.
Re: TEST: server code changes
Posted: 26 Feb 2013, 13:54
by v0id
Some more testing:
Healing is abnormally slow (+2HP on each heal tick, which is afaik the bare minimum). (reported by Monomu)
Using #itenplz in town crashes map-server (outside town is fine). I didn't test all spells but e.g #flar or #ingrav are fine outside or inside town.
For this crash, I got:
Code: Select all
SYSERR: realloc failure: Operation now in progress
Program received signal SIGABRT, Aborted.
0x00110416 in __kernel_vsyscall ()
Re: TEST: server code changes
Posted: 26 Feb 2013, 16:56
by o11c
v0id wrote:Using #itenplz in town crashes map-server (outside town is fine).
For this crash, I got:
Code: Select all
SYSERR: realloc failure: Operation now in progress
Program received signal SIGABRT, Aborted.
0x00110416 in __kernel_vsyscall ()
Hm, that's a generic error when realloc() returns NULL ... huh, I guess it doesn't set errno, so "Operation now in progress" is probably a previous error. The SIGABRT, at least, is intentional after the previous error ... I have no idea why realloc() would return NULL though. I definitely need a backtrace for this one.
(Sigh, I'll be working on fixing the code for a while ...)
Re: TEST: server code changes
Posted: 26 Feb 2013, 17:05
by Big Crunch
Keep at it o11c. There is no shame in failure, only shame in not attempting. You are doing valuable work for the project, amigo. I'm sure you will get these issues ironed out.
BC
Re: TEST: server code changes
Posted: 27 Feb 2013, 21:48
by o11c
Also getting a bug with party.txt never being de-tmp'ed
Re: TEST: server code changes
Posted: 27 Feb 2013, 22:43
by o11c
I seem to have solved V0id's issue with #itenplz.
I've also solved an issue regarding parties (and anything later - storage and registers) not being saved.
Re: TEST: server code changes
Posted: 28 Feb 2013, 05:57
by Nard
Maybe it could be possible to use the second testing server to avoid mixing possible bugs coming from new content and those coming from sever code change?