transition states not saved and/or recalled correctly

[BryanFRitt]

Murphy in transition state not saved on or recalled from tape?
Pause in the middle of moving Murphy to new a position, save tape state, exit level, reload tape, Murphy will be paused at the new position instead of paused in transition to the new position.
You can also do this without pausing too, if you save tape quick enough.

This may result in a segmentation fault in some cases, like when something might have fallen after Murphy moved to the the new spot, or this is done several times.
This probably related to the other segmentation fault issue I posted. Segmentation fault after 2nd reload from tape

This seems to affect other stuff too. Like I tried this out with Level 17 of SOKOBAN. Started the game. didn't move 'Rockford', 'paused' the game, 'saved' the tape. and guess what? Some times after doing this, on restoring this tape 'Rockford' is moving! Not sure why, but something makes me thing this moving is related to previous tape save(s)? [Update: Can't seem to consistently recreate this again. Same version of R'n'D.]
Well I notice If I do this while 'Rockford' is moving, on restore 'Rockford' seems to be in the right place but 'Rockford' is sometimes not in the same body position.

Earlier I was playing around with this, and if saved right after Murphy moves(not just in transition), which way Murphy was facing wasn't being saved/recalled correctly, always going back to the default facing.

[filbo]

Upload a tape which you've observed that loading it causes a segmentation fault. First make sure it reproduces the fault when loaded into a new run of RnD.

[BryanFRitt]

Upload a tape which you've observed that loading it causes a segmentation fault. First make sure it reproduces the fault when loaded into a new run of RnD.

I suppose you can check the linked to post for that tape. I'm not getting segmentation faults when playing right now, but I'm wasn't trying to recreate the issue from that post. So I'll try those tapes...

In trying the ~/.rocksndiamonds/tapes/sp_infotron/048.tape from the linked post, after loading the tape and unpausing, 'Murphy' goes up, and I can't move him. Then 'Murphy' gets killed by 'Snik Snak'. No segmentation faults.

In trying the ~/.rocksndiamonds/tapes/sp_infotron/043.tape from the linked post, after loading the tape and unpausing, 'Murphy' goes down, and I can't move him. Then 'Murphy' gets killed by 'Snik Snak'. No segmentation faults.

Same with the '/.rocksndiamonds/tapes/Sulps levels/040.tape' one. 'Murphy' goes down, and I can't move him. 'Murphy' gets killed by 'Zonk'. No segmentation faults.

I should be able to move 'Murphy' in these situations, but can't.

Testing with rocksndiamonds-4.3.0.0-rc5

[filbo]

Testing with bad040.tape + 4.3.0.0RC5, I can move after the tape is loaded, and was able to finish the level. I used F2 to get to the starting point, hit Alt+2 to go to ultra-slow-mo, then played it out. Murphy makes part way into the exit before being exploded, which (according to the game engine at that moment) constitutes a win.

bad+finished040.tape

(183 Bytes) Downloaded 203 times

(Alt-2 might only work for me due to my config; but you should be able to go to Setup > Game & Menu, turn on Game Speed Extended List, and choose from several excruciatingly slow speeds...)

[BryanFRitt]

Testing with bad040.tape + 4.3.0.0RC5, I can move after the tape is loaded, and was able to finish the level. I used F2 to get to the starting point, hit Alt+2 to go to ultra-slow-mo, then played it out. Murphy makes part way into the exit before being exploded, which (according to the game engine at that moment) constitutes a win.bad+finished040.tape

(Alt-2 might only work for me due to my config; but you should be able to go to Setup > Game & Menu, turn on Game Speed Extended List, and choose from several excruciatingly slow speeds...)

I retried it again... Even at 1fps, still can't control 'Murphy' after loading the bad tape* (can't 'snap', move or face 'Murphy' in a different direction), and 'Murphy' is auto moving down, ...
*until the level is exited, ...

The tape 'filbo' posted makes it to the end / solves the level when I play it back here. Although, it's not showing a high score for anybody but me for any of these 'Sulps levels', at least so far. Double checked, I am playing rocksndiamonds-4.3.0.0-rc5. Has anybody but me, completed at least one of these levels and submitted it? Read the comments for 4.3.0.0-rc3 viewtopic.php?f=2&t=3283 The latest version as of this post is 4.3.0.0-rc5 viewtopic.php?f=2&t=3285 There maybe later versions check viewforum.php?f=2

[filbo]

I don't think my live play is being uploaded to the server yet, as I've been building with the TESTING flag that directs it to a test server.

Also, I'm not sure how the server handles receipt of scores for levels / levelsets it doesn't know about.

[Holger]

I must admit that I am not totally sure what this thread is really about, or what kind of bug happens.

I don't think my live play is being uploaded to the server yet, as I've been building with the TESTING flag that directs it to a test server.

There is currently no other score server than the test server, so please always build with TESTING when testing against the new score server. (The "production" score server has very outdated code and currently does not accept score uploads at all.)

Also, I'm not sure how the server handles receipt of scores for levels / levelsets it doesn't know about.

It accepts them, stores them in the "scores" table, creates a new level set entry in the "levelsets" tabel, but the verification script later invalidates these scores as the corresponding level set does not exist on the server yet.

I get a report every night so I can see which level sets are missing, and can then add them to the server. But there are still some problems to be solved, as, for example, level sets with spaces in their directory names cannot be processed by the verification script (as "canonical" level set identifiers do not contain spaces -- just think of two level sets "cool set" and "cool set 002", and how to validate a tape "002.tape" for these sets). But a solution for this problem is already on its way...