0001229: Random artifacts are not indexed properly

(0003430)
Warmonger (administrator)
2013-03-04 17:40

I didn't find anything that would explain this bug in ArtifactID, adventure object subID or def list indexed by int.

It is possible, however, that this bug holds also for other types of objects, such as creatures.

(0003431)
Ivan (developer)
2013-03-04 18:15
edited on: 2013-03-04 18:19

I'm not sure that this is related to 256 - list in mod.json does not have anything to do with load order: artifacts in one mod are always loaded alphabetically and load order of mods is (somewhat) random - hashmap in filesystem.

Check log file to see real load order ("loaded artifact XXX" messages)

For example one of arts you mentioned have index < 200 and I still have the crash.

This one looks more like desync (likely during map initialization).

UPD: got exactly same crash even after disabling dosen of arts from one of mods. Not 256 problem.

(0003432)
Warmonger (administrator)
2013-03-04 18:37

Strange, as I noticed issues just today and only with recently added artifacts.

Something could also be wrong with the mod itself, but I added lots of arts already with no problems :?

(0003434)
AlexVinS (developer)
2013-03-04 21:00
edited on: 2013-03-04 21:02

In witchkings mod "imglarge.json" is not emplty but its empty in H4 mod. May be an issue or it`s unused?

(0003441)
Ivan (developer)
2013-03-12 16:14

Mysteriously solved? No longer reproducible for me. At least not with rev. 3289, these 3 mods + testy 4 map

(0003442)
Warmonger (administrator)
2013-03-12 19:41
edited on: 2013-03-12 19:45

Bug still persists.
One thing I noticed: it does not seem to happen in FIRST game you run after opening application, but does in next ones when you return to main menu.

(0003470)
Warmonger (administrator)
2013-03-29 11:42

This part of code is suspicious:

        case Obj::ARTIFACT:
        case Obj::RANDOM_ART:
        case Obj::RANDOM_TREASURE_ART:
        case Obj::RANDOM_MINOR_ART:
        case Obj::RANDOM_MAJOR_ART:
        case Obj::RANDOM_RELIC_ART:
        case Obj::SPELL_SCROLL:
            {
                int artID = -1;
                int spellID = -1;
                CGArtifact * art = new CGArtifact();
                nobj = art;

                readMessageAndGuards(art->message, art);

                if(defInfo->id == Obj::SPELL_SCROLL)
                {
                    spellID = reader.readUInt32();
                    artID = 1;
                }
                else if(defInfo->id == Obj::ARTIFACT)
                {
                    //specific artifact
                    artID = defInfo->subid;
                }

                art->storedArtifact = createArtifact(artID, spellID);
                break;

If artifact is not set or is not a scroll, new artifact with ID = -1 is created. The code is not correct even if it should never trigger (which I assume). Will investigate this part.

(0003471)
Warmonger (administrator)
2013-03-29 13:38
edited on: 2013-03-29 13:46

Debugging shows that real artifact stored in objects is different than what def shows. For example, both subtype of object and artifact ID are 55 (Vampire's Cowl), but adventure map def is Logbook of The Master Sailor (new art).

Game crashes when hero tries to equip artifatc in different slot than it should.
There must be some desync on game end / map reload, but I can't find anything.

(0003474)
Ivan (developer)
2013-03-29 18:59
edited on: 2013-03-29 19:11

Desync is in CArtHandler::getArtSync. That's well-chosen name :)

Don't know where it comes from exactly but it definitely here - parameter rand is same on both client and server but returned values are not.

Probably due to internal state (erasePicked flag) but can't say for sure - I'm not familiar with that code.

EDIT: found it

(0003476)
Ivan (developer)
2013-03-29 19:17

Should be fixed, rev 3298

(0003613)
Zamolxis (viewer)
2013-05-31 06:30
edited on: 2013-05-31 06:30

@Ivan/Warmonger: is this the same with what I still see in 0.92c, namely all WoG artifacts showing wrong image when I pick them up? (see screenshot)

Very easy to reproduce by loading the VCMI_Tests_2012 map and check artifacts around Sir Mullich.

(0003615)
Ivan (developer)
2013-05-31 11:26
edited on: 2013-05-31 12:50

In future - it is better to either set this to "feedback" or open new report. I often have resolved/closed filtered out to see only unresolved reports.

On topic - unrelated but will check this out. I think I know where this one is coming from.

UPD: fixed.

Issue History
Date Modified	Username	Field	Change
2013-03-04 17:36	Warmonger	New Issue
2013-03-04 17:40	Warmonger	Note Added: 0003430
2013-03-04 18:15	Ivan	Note Added: 0003431
2013-03-04 18:19	Ivan	Note Edited: 0003431	View Revisions
2013-03-04 18:37	Warmonger	Note Added: 0003432
2013-03-04 21:00	AlexVinS	Note Added: 0003434
2013-03-04 21:02	AlexVinS	Note Edited: 0003434	View Revisions
2013-03-12 16:14	Ivan	Note Added: 0003441
2013-03-12 19:41	Warmonger	Note Added: 0003442
2013-03-12 19:45	Warmonger	Note Edited: 0003442	View Revisions
2013-03-29 11:42	Warmonger	Note Added: 0003470
2013-03-29 13:38	Warmonger	Note Added: 0003471
2013-03-29 13:40	Warmonger	Note Edited: 0003471	View Revisions
2013-03-29 13:46	Warmonger	Note Edited: 0003471	View Revisions
2013-03-29 16:35	Warmonger	Summary	Artifacts over 256 are not indexed properly => Random artifacts are not indexed properly
2013-03-29 18:59	Ivan	Note Added: 0003474
2013-03-29 19:11	Ivan	Note Edited: 0003474	View Revisions
2013-03-29 19:11	Ivan	Assigned To	=> Ivan
2013-03-29 19:11	Ivan	Status	new => assigned
2013-03-29 19:17	Ivan	Note Added: 0003476
2013-03-29 19:17	Ivan	Status	assigned => resolved
2013-03-29 19:17	Ivan	Fixed in Version	=> 0.93
2013-03-29 19:17	Ivan	Resolution	open => fixed
2013-05-31 06:28	Zamolxis	File Added: 2013-05-31_HelmBoots092c.jpg
2013-05-31 06:30	Zamolxis	Note Added: 0003613
2013-05-31 06:30	Zamolxis	Note Edited: 0003613	View Revisions
2013-05-31 11:26	Ivan	Note Added: 0003615
2013-05-31 12:50	Ivan	Note Edited: 0003615	View Revisions
2014-05-30 17:43	beegee	Status	resolved => closed

Notes
(0003430) Warmonger (administrator) 2013-03-04 17:40	I didn't find anything that would explain this bug in ArtifactID, adventure object subID or def list indexed by int. It is possible, however, that this bug holds also for other types of objects, such as creatures.

(0003431) Ivan (developer) 2013-03-04 18:15 edited on: 2013-03-04 18:19	I'm not sure that this is related to 256 - list in mod.json does not have anything to do with load order: artifacts in one mod are always loaded alphabetically and load order of mods is (somewhat) random - hashmap in filesystem. Check log file to see real load order ("loaded artifact XXX" messages) For example one of arts you mentioned have index < 200 and I still have the crash. This one looks more like desync (likely during map initialization). UPD: got exactly same crash even after disabling dosen of arts from one of mods. Not 256 problem.

(0003432) Warmonger (administrator) 2013-03-04 18:37	Strange, as I noticed issues just today and only with recently added artifacts. Something could also be wrong with the mod itself, but I added lots of arts already with no problems :?

(0003434) AlexVinS (developer) 2013-03-04 21:00 edited on: 2013-03-04 21:02	In witchkings mod "imglarge.json" is not emplty but its empty in H4 mod. May be an issue or it`s unused?

(0003441) Ivan (developer) 2013-03-12 16:14	Mysteriously solved? No longer reproducible for me. At least not with rev. 3289, these 3 mods + testy 4 map

(0003442) Warmonger (administrator) 2013-03-12 19:41 edited on: 2013-03-12 19:45	Bug still persists. One thing I noticed: it does not seem to happen in FIRST game you run after opening application, but does in next ones when you return to main menu.

(0003470) Warmonger (administrator) 2013-03-29 11:42	This part of code is suspicious: case Obj::ARTIFACT: case Obj::RANDOM_ART: case Obj::RANDOM_TREASURE_ART: case Obj::RANDOM_MINOR_ART: case Obj::RANDOM_MAJOR_ART: case Obj::RANDOM_RELIC_ART: case Obj::SPELL_SCROLL: { int artID = -1; int spellID = -1; CGArtifact * art = new CGArtifact(); nobj = art; readMessageAndGuards(art->message, art); if(defInfo->id == Obj::SPELL_SCROLL) { spellID = reader.readUInt32(); artID = 1; } else if(defInfo->id == Obj::ARTIFACT) { //specific artifact artID = defInfo->subid; } art->storedArtifact = createArtifact(artID, spellID); break; If artifact is not set or is not a scroll, new artifact with ID = -1 is created. The code is not correct even if it should never trigger (which I assume). Will investigate this part.

(0003471) Warmonger (administrator) 2013-03-29 13:38 edited on: 2013-03-29 13:46	Debugging shows that real artifact stored in objects is different than what def shows. For example, both subtype of object and artifact ID are 55 (Vampire's Cowl), but adventure map def is Logbook of The Master Sailor (new art). Game crashes when hero tries to equip artifatc in different slot than it should. There must be some desync on game end / map reload, but I can't find anything.

(0003474) Ivan (developer) 2013-03-29 18:59 edited on: 2013-03-29 19:11	Desync is in CArtHandler::getArtSync. That's well-chosen name :) Don't know where it comes from exactly but it definitely here - parameter rand is same on both client and server but returned values are not. Probably due to internal state (erasePicked flag) but can't say for sure - I'm not familiar with that code. EDIT: found it

(0003476) Ivan (developer) 2013-03-29 19:17	Should be fixed, rev 3298

(0003613) Zamolxis (viewer) 2013-05-31 06:30 edited on: 2013-05-31 06:30	@Ivan/Warmonger: is this the same with what I still see in 0.92c, namely all WoG artifacts showing wrong image when I pick them up? (see screenshot) Very easy to reproduce by loading the VCMI_Tests_2012 map and check artifacts around Sir Mullich.

(0003615) Ivan (developer) 2013-05-31 11:26 edited on: 2013-05-31 12:50	In future - it is better to either set this to "feedback" or open new report. I often have resolved/closed filtered out to see only unresolved reports. On topic - unrelated but will check this out. I think I know where this one is coming from. UPD: fixed.

Relationships