Cargo Distribution

[Hamilton2007]

I got a couple of crashes today playing the latest build that Petert posted yesterday. I upload one of the crash reports.

You should also upload the crash save or if possible a save before it crashed

Ok, done

[petert]

I got a couple of crashes today playing the latest build that Petert posted yesterday. I upload one of the crash reports.

Quoting the last thing that happened before the crash:

Code: Select all

Tick 8020: game loaded\par
     New game mode: 1 landscape: 0\par

[NekoMaster]

Ok I have a crash peterts most recet build

[petert]

Debian binary, for r18152. Don't worry NekoMaster, I'll make you your win32 build.

[NekoMaster]

Debian binary, for r18152. Don't worry NekoMaster, I'll make you your win32 build.

yeah, sorry that you have to keep building binaries :\

[fonso]

Thanks for all the feedback.

The crash, at least for Auge (and probably for the report on page 42), is happening when that vehicle preview window pops up where you can select if you want to try a vehicle before it gets released. There is some arcane magic with StringIDs in that code and I don't quite see through that, yet. My current guess is that I have added some string in english.txt at the "wrong" place and now it's selecting a different string than originally planned for something in that preview window. Then it's using an uninitialized DParam for a cargo entry which may or may not be within the bounds of CargoSpec::array. If it's not, you get the assertion. An interesting fact is that in FormatString it sometimes checks the cargo ID for validity and sometimes it doesn't. For example in strings.cpp:619

Code: Select all

StringID cargo_str = CargoSpec::Get(GetInt32(&argv))->units_volume;

but in strings.cpp:754

Code: Select all

StringID cargo_str = (cargo == CT_INVALID) ? STR_QUANTITY_N_A : CargoSpec::Get(cargo)->quantifier;

So, I still don't fully understand what's up there, but I know more than yesterday ...

[Auge]

Hello

The crash, at least for Auge (and probably for the report on page 42), is happening when that vehicle preview window pops up where you can select if you want to try a vehicle before it gets released. ...

That said, I've noticed in this window, that the operating costs (only in this window) are shown as 0.

Tschö, Auge

[petert]

I guess Rubidium read about your problem and he has fixed it in r18176. Here is the diff for it. Also, for NekoMaster, here is a win32 binary. This binary will still have the crash, we have to wait until fonso syncs with at least r18176.

[Hamilton2007]

Today i got a crash with the r18053 Win32 build.

[fonso]

I guess Rubidium read about your problem and he has fixed it in r18176. Here is the diff for it. Also, for NekoMaster, here is a win32 binary. This binary will still have the crash, we have to wait until fonso syncs with at least r18176.

Well, yes, he read about it from my comments on IRC ... when I found the crash's cause

Now I'm just merging all the commits from trunk. There are some conflicts in the GUI code. It may take an hour.

[petert]

Then I will wait to make the build, thanks fonso.

[NekoMaster]

I hope to see a new build soon, for now im going to play a unpatched nightly on a network game with a friend

[bigos]

I see that Rubidium is changing all GUI so better to wait when he finish...

[petert]

I get an error during make:

Code: Select all

make[1]: Leaving directory `/home/Peter/CargoDist/objs/lang'
make[1]: Entering directory `/home/Peter/CargoDist/objs/release'
[SRC] Compiling and Linking endian_check.exe
[SRC] Testing endianness for target
make[1]: *** No rule to make target `C:/msys/1.0/home/Peter/CargoDist/src/linkgraph/demands.cpp', needed by `Makefile.dep'.  Stop.
make[1]: Leaving directory `/home/Peter/CargoDist/objs/release'
make: *** [all] Error 2

[fonso]

I have merged the changes from trunk now. The bug Auge and Creat reported should be resolved now. I suspect at least some of the other crashed have been caused by the same bug, so please recheck with a newer version or apply the hotfix petert has posted. I have also commited a fix for the time reset problem Beday has posted. Components are merged now not only when they are due, but also when their join date is very far in the future. Like this the number of components shouldn't grow too much, even if you reset the time. I include a hotfix for that, too - in case you have a problem with link graph components from the future eating your RAM and you don't want to give up the game ...

Now, as the assertion has been found to not stem from corruption of game state, the question about Creat's desync arises again. So, how many of you have been playing network games lately and did you see any desyncs? With which versions?

petert: The demands.cpp file is in the patch as well as in git. Maybe you need reapply the patch on a clean trunk or check "git status" and "git log" as well as maybe "git diff origin/cargodist" to see what's wrong there.

[petert]

Ok, here is the r18177 bundle for cargodist win32, happy gaming Neko, and other win32 users.

[Hamilton2007]

I think there is a bug with the depots in the latest build Petert?

[Hamilton2007]

Ok, when i make the depot window a little bit larger then the vieuw is normal..

[Creat]

Hamilton2007: Even though I haven't checked I'm pretty sure this is due to Rubidiums GUI overhaul and not a Cargodist problem. Can you confirm that this happens only in Cargodist and not the corresponding nightly build?

[petert]

I haven't seen any revisions that would cause this, but what do I know. Wait for Rubidium to give you a precise answer.