[10.4] convoy spacing bug with ships

[ӔO]

for some reason, sometimes ships will get stuck by ignoring minimum load and wait times.
It's a bit odd, but the ship will get a waiting time of 2000+ when there is only 3:12 of time each month.

http://dl.dropbox.com/u/17111233/Turnkey.sve

[wlindley]

I have seen this as well, but I never use the spacing controls, just the minimum load and wait times; does it happen for you even without spacing?

[ӔO]

aren't spacing and wait times the same thing in experimental?

[jamespetts]

Thank you for the report. I have seen similar issues reported, but have not been able to reproduce them. Are you able to posit steps to reproduce the problem reliably? If so, I can have a good crack at fixing it when I get back home.

[ӔO]

Well, I'm not certain, but they seemed to have occurred while the game was in fast forward.
They might have gotten stuck after adjusting the spacing, but it took a while for me to notice that they were stuck because I was working on other lines at the time.

[jamespetts]

Can you see whether you can reproduce this reliably? It would be of great assistance to me if you could. Thank you for your help!

[dustNbone]

I've noticed some connection between this and the depot being directly adjacent to the dock.  Same thing seems to happen to trains being deployed from a depot directly adjacent to the first destination.

[inkelyad]

aren't spacing and wait times the same thing in experimental?

No. It is close, but not the same.

for some reason, sometimes ships will get stuck by ignoring minimum load and wait times.

I can't test/debug this right now too. Can you build custom packs?
Bug ships sometimes have big loading and reverse time. Try to change it.

[ӔO]

It seems to happen when convoy spacing is set to 4 or lower, the convoy starting point is right next to the depot and the convoy is still partially inside the depot.

the East Indiaman does it quite frequently when deploying.

[jamespetts]

AEO,

may I ask where on the map the problem occurs? The screenshot shows boats of a type different to those which I can find operating anywhere.

[jk271]

I had a problem with ships that seems to have something common with this.

I build two ports (one is near a coalmine, another near coal power station).
I made a ship line connecting these two ports. Ships has to be fully loaded to be able to leave the first port.
When ship is loading big time amount is shown.
At the moment of leaving port, segmentation fault is received.

I am using  simutrans experimental 10.4 (simutrans-exp-2011-12-19-17f9c1b) with pak.64-experimental.
OS  32-bit linux, CPU 32bit too.

gdb backtrace:

Program received signal SIGSEGV, Segmentation fault.
0x082347df in convoi_t::increment_odometer(unsigned int) ()
(gdb) bt
#0  0x082347df in convoi_t::increment_odometer(unsigned int) ()
#1  0x082efb2e in vehikel_basis_t::fahre_basis(unsigned int) ()
#2  0x08235c9d in convoi_t::sync_step(long) ()
#3  0x082bdfc7 in karte_t::sync_step(long, bool, bool) ()
#4  0x08272afc in interrupt_check(char const*) ()
#5  0x082d3446 in karte_t::interactive(unsigned int) ()
#6  0x0827c28f in simu_main(int, char**) ()
#7  0x08314161 in main ()
(gdb)

(Too) Long time, which is dislayed, could be caused by reading some part of memory, that was not intented to be read.

P.S. Sometimes it worked so I used electric fence to reproduce it.
(LD_PRELOAD=/usr/lib/libefence.so)

P. P.S. There can be also bug in a compiller, gcc 4.4 and 4.5 has a bug - it sometimes swapped instructions, that should not be swapped.

[jamespetts]

Thank you very much for your report. Can you upload the saved game in which this can be reproduced?

[jk271]

I have compiled compiled simutrans experimental from latest 10.x branch. No segmentation fault now received. Newly created games (pak.64-experimental) work well.
Older game (created with 10.4 ) has the strange inscription (showing enormous waiting time) still.
But works ok, no enourmous waiting time despite the label.

I will upload the game. I am just looking for some place.

https://rapidshare.com/#!download|849tl4|2152990522|lode.sve|293|R~E4F48B7D78EB86FCA39F5EFF31C5585B|0|0

Edit: Unfortunately the link is whole string. I will search for some better place.


Edit2: This should work better:
http://simutrans-germany.com/files/upload/lode.sve

[jamespetts]

Thank you very much for confirming this - this is very helpful. From testing your saved game, the erroneous waiting time only appears when ships first dock, and, as you say, is now an error in display only: the ships do not observe the incorrect waiting times. On subsequent dockings, the ships appear to show the correct time in any event. In the circumstances, I think that one can call this bug solved, probably as a side effect of solving this bug.