4.1 feature branches

[Nick]

I was more concerned about carrying my monster knowledge over, but I've realized now that Vanilla doesn't suffer the same portability problem with that that the other variants I've played have had. Excited to try it out, thanks.

Excellent, this was precisely the point of separating out monster memory.

[debo]

Excellent, this was precisely the point of separating out monster memory.

That's a bold statement.

[nikheizen]

Trying this out (the new master branch) and ran into a crash. Running on windows via cygwin, compiled in terminal.

Went to check my artifact knowledge, all the revealed artifacts show up in the menu as (nothing), though the description(and thus name) of the artifacts I have in my inventory are still available. If I look at an artifact which I left behind, however, the game crashes.

I managed to get this error line even though the crash totally messes up my terminal:

assertion "obj->known" failed: file "obj-identify.c", line 547, function: object_know_all_but_flavor

And then the aborted, core dumped line.

Luckily I can restart from the last save. Infinitely reproducible on my system.

Here is my save file and a screen shot, zipped together:

[Nick]

Here is my save file and a screen shot, zipped together

Thanks - easy fix. I'll post again when rebuilds incorporating the fix are done.

[nikheizen]

Should I post bug reports (+ core dumps if a crash) with the master branch here, or is elsewhere better in future. I had a couple of unexplained crashes recently.

[Nick]

Should I post bug reports (+ core dumps if a crash) with the master branch here, or is elsewhere better in future. I had a couple of unexplained crashes recently.

Here is good.

[nikheizen]

Here is good.

Alright, first, saving with ctrl+S freezes the entire terminal. Not sure what's going on there. The save process doesn't complete, nor do I get a core dump since I have to forcibly end the process. Likely cygwin-specific, I'd think.

Second. I had an unexplained crash while clearing a greater vault. Was walking down a hallway(by holding h) when it happened. Dump attached. I had already been on the floor because of the former. No message displayed.

Also: I think this is just a general 4.0.4 bug, but ^H doesn't seem to be bound to +4 like the other ctrl+movement keys are set up (to +1, +2 etc.) in the roguelike keyset.

Tell me if I can provide any more information I might not have thought of.

[Nick]

Alright, first, saving with ctrl+S freezes the entire terminal. Not sure what's going on there. The save process doesn't complete, nor do I get a core dump since I have to forcibly end the process. Likely cygwin-specific, I'd think.

Ctrl+S suspends a lot of terminals, with ctrl-Q un-suspending. So I get this behaviour with a linux terminal and the gcu port - I think this is just a limitation of playing in a terminal.

I haven't looked at the crash yet.

[Nick]

Here are links to the latest builds - Windows and OS X.

Second. I had an unexplained crash while clearing a greater vault. Was walking down a hallway(by holding h) when it happened. Dump attached. I had already been on the floor because of the former. No message displayed.

I'm afraid I have no idea what to do with that stackdump, as I don't recognise the format. So I think we're just going to have to wait for more of those crashes to turn up, hopefully reproducibly.

Also: I think this is just a general 4.0.4 bug, but ^H doesn't seem to be bound to +4 like the other ctrl+movement keys are set up (to +1, +2 etc.) in the roguelike keyset.

This may be another terminal thing - ctrl-H is usually mapped to backspace.

[yyt16384]

For the Ctrl-S thing, I think there are some ways to do that like using raw(). PosChengband does it well, for example.

I think cygwin terminal has some broken terminfo about Ctrl-H and backspace. I was having the same problems with other programs there.

[Pete Mack]

Yes, gcu should use raw. main-win.c doesn't have the problem. What OS are you using to see the issue?

[nikheizen]

I think I figured out what may have caused my prior crash.
On another level with a greater vault (had this happen twice on the same level!):
I blink and Tselakus summons ringwraiths
Move away from an Istar
Crash
assertion "obj" failed: file "obj-pile.c", line 220, function: object_lists_check_integrity

Okay, glad it's just a problem for my terminal then.

Yes, gcu should use raw. main-win.c doesn't have the problem. What OS are you using to see the issue?

Windows Cygwin.

[Nick]

I think I figured out what may have caused my prior crash.
On another level with a greater vault (had this happen twice on the same level!):
I blink and Tselakus summons ringwraiths
Move away from an Istar
Crash
assertion "obj" failed: file "obj-pile.c", line 220, function: object_lists_check_integrity

Right, thanks. If you get a savefile where this is repeatable, I'd like to see it.

[bdgamer]

I just tried playing the new build, but the screen is messed up, font-wise, from the start screen, and far worse when changing to no graphics. Tried multiple fonts, and all of it looked bad. Anything I can do?

Here are links to the latest builds - Windows and OS X.



I'm afraid I have no idea what to do with that stackdump, as I don't recognise the format. So I think we're just going to have to wait for more of those crashes to turn up, hopefully reproducibly.



This may be another terminal thing - ctrl-H is usually mapped to backspace.

[Pete Mack]

Nick--
I'll start running in gdb. I hate doing it, because the moment I start paying attention to code is the moment I lose attention to the character and die :/