A few bugs in the options management, and I shed light on the lingering mismatching config (see X in the config menu, game does Y). We'll start with the easiest.

---

Wrong default out of box.

1. Freshly extract Exp 6 from your full release.
2. Run NL.
3. Play a level (e.g., the included Mazulems level 1).
4. Fast-forward into death.

Expected: We get the panel message to rewind or quit.
Observed: We exit to postview.

IIRC, we agreed that the default should be to exit if we have won, and show the panel message when we lose. This is the most appropriate default. We had newbies (Darkshoxx in one of his streams) shout in surprise when the level exited on losing.

---

File contains three options, not one

1. Freshly extract Exp 6 from your full release.
2. Run NL.
3. Open the options menu (click on the cog in the main menu).
4. Press OK to save options.
5. Exit NL.
6. Open in a text editor: ./settings/settings.ini

Expected: One of the lines is for the end-of-level behavior.
Observed: Three of the lines are for the end-of-level behavior, e.g.:

AlwaysEndGameplay=1
EndIfLevelPassed=0
NeverEndGameplay=0

This invites to put nonsensical values in the file, e.g., set them all to 0, or set several of them to 1.

I recommend to store one line, e.g., with values 0, 1, 2. Call it, e.g., ExitToPostviewBehavior, or, if you'd like 0 to mean always-exit and want to keep that nuance in the text file, call it, e.g., StayInGameAfterLemmingsDied, although that's contrived.

Certainly, the user can still put nonsense values even when we use only a single line. E.g., he can text-edit the line to put negative values, or 3 or greater. It's probably okay to do whatever you want then, as long as it means the same as exactly one out of 0, 1, or 2. Maybe fix it under the hood, e.g., convert it to 0, 1, or 3 at options-loading time?

In particular, avoid the following mismatching bug for such nonsense inputs. It's the uncanny bug from months ago:

---

Mismatch from more than one 1

1. Produce a settings file with the Exp 6. (E.g., do all steps (1-6) from previous bug: File contains three option lines for the end-of-level behavior.)
2. Text-edit the file: Make all three lines end in 1 (none in 0).
3. Run NL.
4. Play a level and lose, e.g., kill all lemmings.
5. You automatically exit to postview.
6. Return to the NL main menu.
7. Open the options menu and go to the interface tab.

Expected: "Always Exit to Postview" is selected, matching our step 5.
Observed: "Never Exit to Postview" is selected.

Certainly, the user had to do something nonsensical to get here, and you can probably treat the garbage as what you like. But be consistent across game and options dialog.

-- Simon

All right, 17:00 UTC on Saturday is it!

-- Simon

We link both the 32-bit and the 64-bit executables with lld-link. These executables will try to write all exceptions flying out of main() to stderr, which is closed because we suppress the console. Therefore, we see Lix terminate seemingly without error.

When we link a test project with Optlink (instead of with lld-link) and again suppress the console, we get the desired error box for exceptions flying out of main().

Is it possible to route the uncaught exceptions to error boxes even when we link with lld-link? I've looked through lld-link's command-line switches, and none of them look related. I'll ask on the D forums these days.

-- Simon

I call a replay oneframediff if and only if both of the following hold:

• During mass-replay verification in NL 12.12.5, it shows the same high-level result (pass, fail, undetermined, ...) as during mass-replay verification in nl-2024-04-08.
  
• Mass-replay verification in NL 12.12.5 takes exactly one physics update longer to run this replay than the mass-replay verification in nl-2024-04-08.
  

See my earlier long post for why I expect oneframediff replays.

Now:

• All 208 replays in Lemmings United (excluding Bonus rank) are oneframediff.
  
• All 120 replays in Lemmings Introduction pack are oneframediff.
  
• I have 3 more Icho replay collections that I'll test for oneframediff-ness. Expect more results in the next 0-3 days.
  

Thus: Excellent so far!

Here is a Bash script that takes two NL output files (pass them as argument 1 and argument 2 to this script) and tells me if the replays are oneframediff:

#!bin/bash
paste "$1" "$2" \
    | sed 's%[^(]* (%,%' \
    | sed 's% [^(]*(%,%' \
    | sed 's% .*$%%' \
    | grep ^, \
    | grep -v "none" \
    | grep -v "TALISMAN" \
    | awk 'BEGIN { FS="," } { print $2 - $3; }'

The script prints the difference (see item 2 in my first enumerated list in this post) in run frames of the replay between the two NL versions. For a oneframediff replay, the script will print 1. If everything is oneframediff, you get a long stream of 1s.

The NL replay results format isn't the nicest to parse. But this script isn't particularly smart either; I should have relied on the consistently-appearing "... frames" in the NL replay format. Instead, the script cuts at open parentheses (, which will occiasionally appear in the level names. The script won't halt on such extra parentheses, but will produce garbage results over 1,000 or under −1,000 that we must examine by hand. All were fine.

-- Simon

Right.

First, #488: Logging (user/log.txt): Exceptions during init aren't logged. This will be easy to fix. I'll fix this in the next release, Lix 0.10.23. I've already put the fix in my unstable repo's master.

Second, on Windows, years ago, the D runtime used to generate a dialog box for unrecoverable errors (exceptions/errors flying out of main). Only after clicking away that dialog box, Lix would exit. Nowadays, we don't get this dialog box anymore. I don't know why we don't get the dialog anymore.

Third, on Windows, we suppress the console. This is correct. We'll ask people to look in user/log.txt for errors. The above #488 has prevented this during initialization, it won't prevent it from Lix 0.10.23 on.

---

We can investigate why there is no dialog box (above "Second, on Windows ..."). Possible relation with suppressing the console? Although we have suppressed the console for years, too.

-- Simon

Quote from: WillLem on April 18, 2024, 11:47:38 AM

We should catch up again properly about this soon, let me know when you're available.

I'll be busy this weekend. Which of the following suits you best?
[...]
Wednesday, April 24, 20:00 UTC

I can't do much more than offer these times. I expected you to agree on one or propose something else.

Friday, 26th, 17:00 UTC I can make. Otherwise it will have to be next week.

-- Simon

Thanks. I'll call that executable nl-2024-04-08.

Icho has sent me his replay coverage for Lemmings United. I've looked at his 208 replays for what is not in the bonus rank of United. These 208 replays are mostly 1 per level, sometimes 2 per level.

I've ran those 208 Lemmings United replays separately through each of the following:

• NL 12.12.5 stable,
  
• nl-2024-04-08 with the option: Always Exit to Postview,
  
• nl-2024-04-08 with the option: Exit if Save Requirement Met,
  
• nl-2024-04-08 with the option: Never Exit to Postview.
  

Find attached NL's text output. First findings:

All 208 replays pass (solve their level) in all of the 4 runs.

Your nl-2024-04-08 produces output independent of the 3-way option. In more detail: When I run mass replay verification in the nl-2024-04-08 with the option to Exit if Save Requirement Met, I get the same output (i.e., identical text file) as when I run mass replay verification in the nl-2024-04-08 with the option Never Exit to Postview, or with the option Always Exit to Postview. This is good.

Your nl-2024-04-08 produces different output than NL 12.12.5. In NL 12.12.5, most (all?) replays run for exactly 1 physics update longer than in nl-2024-04-08. It's drudgework to check this claim for all 208 lines, I haven't written a script to verify that claim for me.

For example, NL 12.12.5 produces:
Pacifism 8:  10000 B.C..nxrp   (1882 frames) LvV 0000000000000000 / RpV: 0000000000000000
nl-2024-04-08 produces:
Pacifism 8:  10000 B.C..nxrp   (1881 frames) LvV 0000000000000000 / RpV: 0000000000000000
The only difference here is that NL 12.12.5 runs for 1882 physics updates ("frames") and nl-2024-04-08 runs for 1881 physics updates before both conclude that the replay passes its level.

We can explain this with the reworked end-of-level behavior: NL 12.12.5 must start a new physics update before it can conclude that the map is over. nl-2024-04-08 can test for completion in between two physics updates. Therefore, nl-2024-04-08 needs one physics update fewer. Thus, I believe: There is nothing to worry here. Do you agree?

The high-level result (pass or fail) is identical across NL 12.12.5 and nl-2024-04-08. All levels pass.

This doesn't conclude the testing of mass replay verification yet. Reason: I've only tested solving replays. I haven't tested a big bucket of failing replays, of indeterminate replays (that run for too long after final skill), of weird nukes in the replays, ...

And I have replays for 4 more Icho packs. That's for next week.

-- Simon