Persistent levels

[Nick]

Originally Posted by kaypy

Word of Recall seems to keep x,y location fixed, which may leave you stuck in a wall or worse yet off the valid map area for a savefile wrecking crash.

I'll go poke around in the source now and see if I can come up with a patch for it.

Problem is in prepare_next_level(), which assumes the player is arriving via stairs. Feel free to fix it, otherwise I will before too long

[kaypy]

Oops, hadn't noticed a new post had come in (it was on the next page)

There should be a first pass patch on github now

It should probably use a similar approach to the teleport code, but I wanted to get something up quick...

(my first thought was to call into the teleport routine, but its kinda linked into the player-activation stuff, also the range isnt really relevant for level creation time...)

[kaypy]

Originally Posted by kaypy

BTW, I am seeing a fair amount of code like

level_by_depth(p->depth + 1)

in the new code.

Does that cooperate with the stair_skip constant?

I double checked this and the stair connections break if stair_skip is set to != 1

There is also the matter of z_info->max_depth.

dungeon_get_next_level is almost what you want here, but unfortunately it is set to also block on quest levels...

...

Hmm. If create stair is disabled, does killing Sauron work?...

[Nick]

Thanks, I'll look at this properly soon.

[Sky]

got another unrecoverable crash.

recalling back to town gave me a "invalid player location" error, and a broken save file.
https://drive.google.com/open?id=0B9...2tvY2tmRTZVVnM

[kaypy]

Originally Posted by Sky

got another unrecoverable crash.

This is the one I just put up a patch for...

(The patch will stop the initial crash- I dont *think* it will recover the file though...)

[kaypy]

I've got a seg fault here if I recall.

savefile.zip

The stack trace points at something in the sdl font handling, but I doubt that is where memory is getting first mishandled...

I am running with the recall patches, but they dont do anything interesting with memory, so I woudn't guess the problem is there?

Hmm. I wonder what happens if I wander up to the top-left where my code may not be called at all? (This might also be a way to test against unpatched code without the out-of-bounds crash interfering)

...

it still crashes on the top-left area...

[Nick]

Originally Posted by kaypy

I've got a seg fault here if I recall.

Attachment 1586

The stack trace points at something in the sdl font handling, but I doubt that is where memory is getting first mishandled...

Found it, it's bad string handling in savefile loading. I'll get a fix up soon.

EDIT: Correction, that's a small problem that doesn't cause the crash. There's a bigger issue I haven't nailed down yet.

[Nick]

OK, latest build fixes the problem, that savefile should work fine now.

[kaypy]

That got it. Thanks