4.1 feature branches

[Pete Mack]

Glad you mentioned that. I was getting ready to give it a shot. Well done!

[bdgamer]

Bug found

I received an Assertion failed! error

File: obj-identify.c
Line:579
Expression: obj->known

Savefile with error is attached.

[PowerWyrm]

Fuzzy detection doesn't update object knowledge when something has disappeared from cave (player casts sense objects, monster picks up something, player casts sense objects again). Maybe it should, unless "fuzzy" also includes player knowledge...

[Nick]

I received an Assertion failed! error

File: obj-identify.c
Line:579
Expression: obj->known

Savefile with error is attached.

What were you doing when you got the error, and what had happened from where you made the savefile?

[PowerWyrm]

Fuzzy detected objects probably have the wrong prefix when displaying the object list.

if (entry->object->marked == MARK_AWARE)

was changed to:

if (entry->object->kind == entry->object->known->kind)

which is the new code for "seen" objects. This line should probably be:

if (entry->object->kind != entry->object->known->kind)

[PowerWyrm]

Objects wiped by effects out of LOS of the character are noticed, which kinda opposes the whole knowledge stuff. Like objects picked up/crushed by monsters, they should stay in the player's knowledge until in LOS or unrevealed via detection.

Function project_o() still states this clearly:

Hack -- effects on objects which are memorized but not in view are also seen.

And the code does:

			/* Effect "observed" */
			if (obj->known && !ignore_item_ok(obj)) {
				obvious = true;
				...
			}

"known" should probably been replaced by "seen"...

[Nick]

			/* Effect "observed" */
			if (obj->known && !ignore_item_ok(obj)) {
				obvious = true;
				...
			}

"known" should probably been replaced by "seen"...

So I want

			/* Effect observed */
			if (obj->known && !ignore_item_ok(obj) && cave_isseen(cave, y, x)) {
				obvious = true;
				...
			}

and then only give messages and become aware of mimics if obvious is true, right?

[Pete Mack]

BTW: "unreavealing" an object via detection is a overkill. I'd much rather have the spell do "object belief" rather than "object knowledge."

[PowerWyrm]

So I want

			/* Effect observed */
			if (obj->known && !ignore_item_ok(obj) && cave_isseen(cave, y, x)) {
				obvious = true;
				...
			}

and then only give messages and become aware of mimics if obvious is true, right?

Exactly that.

[PowerWyrm]

Fuzzy detection doesn't update object knowledge when something has disappeared from cave (player casts sense objects, monster picks up something, player casts sense objects again). Maybe it should, unless "fuzzy" also includes player knowledge...

Nevermind... the current behavior is correct. It's consistent with ignored objects, which are sensed by fuzzy detection, but disappear from view only when in LOS.

[PowerWyrm]

Slay cache is broken again in the latest version. The old code was doing a check vs known brands/slays:

if (brands_are_equal(obj->brands, slay_cache[i].brands, TRUE) && slays_are_equal(obj->slays, slay_cache[i].slays, TRUE))...

The new code does:

if (brands_are_equal(obj->brands, slay_cache[i].brands) && slays_are_equal(obj->slays, slay_cache[i].slays))...

Since the check is vs known brands/slays, it probably should be:

if (brands_are_equal(obj->known->brands, slay_cache[i].brands) && slays_are_equal(obj->known->slays, slay_cache[i].slays))...

[PowerWyrm]

While implementing object knowledge for my variant, I saw a difference in the code for make_fake_artifact(), which is used for artifact knowledge menus and artifact power: at the end, my code does

    /* Learn brands and slays */
    object_know_brands_and_slays(obj);

which seems logical, since slay_power() will only use the known brands and slays to evaluate the power of brands/slays. This means brands and slays never contribute to artifact power...

[Nick]

While implementing object knowledge for my variant, I saw a difference in the code for make_fake_artifact(), which is used for artifact knowledge menus and artifact power: at the end, my code does

    /* Learn brands and slays */
    object_know_brands_and_slays(obj);

which seems logical, since slay_power() will only use the known brands and slays to evaluate the power of brands/slays. This means brands and slays never contribute to artifact power...

I don't believe this is a problem, because the only place artifact power is used is in randart generation, where all brands and slays are assumed known anyway.

For similar reasons, the slay cache being broken doesn't matter much - it just means it's not used when an individual ego item is being assessed for price, while its main use is speeding up stats gathering (where, again, all slays and brands are assumed known).

I'm not planning to do anything about either of these at this stage, or about the various ID crashes, because the rune-based ID branch changes a lot of the relevant code.

[Pete Mack]

Has anyone using Linux64 tried running a build with CLang Memory Sanitizer? I just hit an assertion in realloc, which is almost certainly a sign of memory corruption. (I can't really see what happened unfortunately.) Lost a level where I killed Mouth of Sauron, too. :/

[Pete Mack]

I don't understand the fractional blows system. I have 18/220 on both STR and DEX, but for weapons weighing 18lbs are getting only 4.5 blows, and the descriptions says I need +2 more DEX to get 5.8 (4.8+1 for high level warrior)
Is this intentional, and if so, why? Truly, 18/200 should be enough DEX to get max blows with any weapon.