View Issue Details

IDProjectCategoryView StatusLast Update
0004702Spring engineGeneralpublic2015-03-25 23:16
Reporterlamer Assigned ToKloot  
PrioritynormalSeveritymajorReproducibilityalways
Status resolvedResolutionfixed 
Product Version98.0.1+git 
Target Version99.0Fixed in Version98.0.1+git 
Summary0004702: Broken pathing
DescriptionPathing was broken somewhere between 98.0.1-398 and 98.0.1-426
Now units try to go directly through unpassable areas.
Additional InformationMade quick demo, trying to move commander around cliff at 0:10-2:00
Good pathing: http://zero-k.info/Battles/Detail/337736 (98.0.1-398)
Bad pathing: http://zero-k.info/Battles/Detail/337740 (98.0.1-426)
TagsNo tags attached.
Checked infolog.txt for Errors

Activities

Google_Frog

2015-03-14 06:42

reporter   ~0014171

This bug is easy to reproduce so I am testing intermediate versions.

98.0.1-407-g8c94d36 works.

Google_Frog

2015-03-14 07:22

reporter   ~0014172

"save ~1MB in pathfinder (thx Picasso)" is broken.

https://github.com/spring/spring/commit/d120caeabba0a3e2e1326f41b4be3fbaaeb7d026

Google_Frog

2015-03-14 08:03

reporter   ~0014173

"TestBlock(): remove an unused arg" has the bug too.

https://github.com/spring/spring/commit/1e4342db2f6ba5e4572705e5d05b733e6a23c485

abma

2015-03-16 01:39

administrator   ~0014175

Last edited: 2015-03-16 01:42

did you delete pathcache each time you run the test?

i don't see a bug in

https://github.com/spring/spring/commit/1e4342db2f6ba5e4572705e5d05b733e6a23c485

this could also lead to desyncs when the pathcache version wasn't changed when pathing was changed.

Google_Frog

2015-03-16 12:37

reporter   ~0014176

A new cache was built each time. The time between commits was too short for the automatic build so I had to branch from some versions and build them. Each branch had its own cache.

What do you mean by "I don't see a bug"? The version prior to this one did not compile for me so there are two pathing related commits in what I built. Were you not able to reproduce the bug but were able to reproduce it for later versions?

Google_Frog

2015-03-16 14:28

reporter   ~0014177

Regardless, 98.0.1-426 has the bug. I trust my bisecting but it is not worth getting hung up on.

abma

2015-03-16 23:47

administrator   ~0014180

side-note: 98.0.1-426 is not a valid commit / version. due to force pushes / merges / tags this number isn't a static reference, so please use either the short tag, like 98.0.1-426-g1aad565 or the commit hash, like https://github.com/spring/spring/commit/1aad5655fadd5b191f6aab359202bec3ac96049d

https://github.com/spring/spring/commits/develop

as "TestBlock(): remove an unused arg" is broken, very likely it was introduced in "TestNeighborSquares(): save a few expensive ::IsBlockedNoSpeedModChec":

https://github.com/spring/spring/commit/2d4ffe4c1e66471a2023a7f630da53a722cb33ab

abma

2015-03-16 23:48

administrator   ~0014181

> What do you mean by "I don't see a bug"?

the bug wasn't introduced in this commit.

abma

2015-03-16 23:50

administrator   ~0014182

can you verify that the bug was introduced in https://github.com/spring/spring/commit/2d4ffe4c1e66471a2023a7f630da53a722cb33ab ?

lamer

2015-03-17 16:27

reporter   ~0014186

here's how it goes for me:
* works: https://github.com/spring/spring/commit/736df0c543091907a3d890a7c20ed8cd52665e79
* won't compile: https://github.com/spring/spring/commit/2d4ffe4c1e66471a2023a7f630da53a722cb33ab
* bug: https://github.com/spring/spring/commit/1e4342db2f6ba5e4572705e5d05b733e6a23c485

Google_Frog

2015-03-18 12:25

reporter   ~0014187

" can you verify that the bug was introduced in https://github.com/spring/spring/commit/2d4ffe4c1e66471a2023a7f630da53a722cb33ab "

No. I said:

"The version prior to this one did not compile for me so there are two pathing related commits in what I built."

abma

2015-03-18 13:28

administrator   ~0014188

Last edited: 2015-03-20 03:06

verify = make it compile & test it

or, if possible: revert the commit and test then


goal is to reduce the ammount of code where the error was introduced or maybe already fix the error.

cleanrock

2015-03-20 17:51

reporter   ~0014199

I made 2d4ffe4 compile with and without the removal of "// check if square is outside search-constraint" block in CPathFinder::TestBlock, to make a long story short:
jK, you broke pathing, start a NullAI game on 2_Mountains_Battlefield and see what happens when you try to walk your com down the mountain when standing close to the edge.

Issue History

Date Modified Username Field Change
2015-03-13 00:34 lamer New Issue
2015-03-14 06:42 Google_Frog Note Added: 0014171
2015-03-14 07:22 Google_Frog Note Added: 0014172
2015-03-14 08:03 Google_Frog Note Added: 0014173
2015-03-16 01:39 abma Note Added: 0014175
2015-03-16 01:42 abma Note Edited: 0014175
2015-03-16 12:37 Google_Frog Note Added: 0014176
2015-03-16 14:28 Google_Frog Note Added: 0014177
2015-03-16 23:47 abma Note Added: 0014180
2015-03-16 23:48 abma Note Added: 0014181
2015-03-16 23:50 abma Note Added: 0014182
2015-03-16 23:52 abma Target Version => 99.0
2015-03-17 16:27 lamer Note Added: 0014186
2015-03-18 12:25 Google_Frog Note Added: 0014187
2015-03-18 13:28 abma Note Added: 0014188
2015-03-18 13:29 abma Note Edited: 0014188
2015-03-20 03:06 abma Note Edited: 0014188
2015-03-20 17:51 cleanrock Note Added: 0014199
2015-03-21 18:49 Kloot Status new => resolved
2015-03-21 18:49 Kloot Fixed in Version => 98.0.1+git
2015-03-21 18:49 Kloot Resolution open => fixed
2015-03-21 18:49 Kloot Assigned To => Kloot
2015-03-25 23:16 jK Relationship added duplicate of 0004704
2015-03-25 23:16 jK Relationship deleted 0004704