World Conquest NM

[Natasiel]

Yogibear:
I compared the value screen to screen using 2 clients, :debug and :inspect; finishing the first scenario on one instance. So I suspect "start-of-scenario saves" is what has been used. I have not been careful, but I'd say the result is not persistent within the full range of save type you enumerated. I'll test further and give you back some news about it.

Crab:
I'll test 'no_leader="yes"' prestart recall, and I could also test your patch if you are willing to share it. The latest WC_NM is available from the addon server if you compile from trunk. I'll upload a special version with "easy level" made for debugging.

# where the [side] is initialized on first scenario:
~/.wesnoth1.7/data/add-ons/WC_NM/scenarios/WC_Start.cfg

# where the [side] is initialized in subsequent scenarios:
~/.wesnoth1.7/data/add-ons/WC_NM/utils/scenario.cfg

A question: Should I do those tests against trunk or some other branches?

And again, thank you for your feedbacks.

[ezysquire]

Great work Natasiel: you need a tester - call on me any time.

Any further thoughts on the fixed hero underling list?

[Yogibear]

A question: Should I do those tests against trunk or some other branches?

Trunk, please.

[Natasiel]

Great work Natasiel: you need a tester - call on me any time.

Any further thoughts on the fixed hero underling list?

I am pretty sure we'll get many long nights testing this stuff as soon as 1.8 is out.
But I got no idea about what is going on the fixed hero underling list.

We really are making progress. Crab's workaround for first bug does the trick.

Spoiler:

[event]
name=prestart
{FOREACH_FROM player 1 I}
[store_starting_location]
variable=startpos
side=$I
[/store_starting_location]
[recall]
canrecruit=yes
x,y=$startpos.x,$startpos.y
side=$I
show=no
[/recall]
{NEXT I}
{CLEAR_VARIABLE startpos}
[/event]

Next, We'll have to wait for the patch, as recall lists are screwed upon reload. But that's really strange; when completing first scenario, you get your recall list on arrival in scenario 2. If you save on scenario 2 then reload, your recall list will still be right. But, if you save in the middle of scenario 1, then reload; upon completion, you will land in scenario 2 without anything to recall as in v1.7.12+svn (40862).

[Rigor]

oh wow im gonan testificate this right now!

[Natasiel]

I am logging that 1.7.12+svn(40927) has the same behavior. Whenever there is a reload of scenario 1, either auto or manually saved, once entering scenario 2, we got no recall list at all. Not even the leader is available.

[Rigor]

unfortunately, NM on difficulty "Nightmare" is shockingly easy.

did u forget that in vanilla WC there were lots of dangerous lvl2 & 3s in the first few turns ?

we r trying out playing WC NM on "hard" difficulty now and tell u later about it.

[Yogibear]

I am logging that 1.7.12+svn(40927) has the same behavior. Whenever there is a reload of scenario 1, either auto or manually saved, once entering scenario 2, we got no recall list at all. Not even the leader is available.

I can confirm this and we are trying to fix that.

[Crab]

Wesnoth 1.7.13 includes some patches by Yogibear and me, to fix 'leader's ids are changed' and 'recall list is gone' issue.
can you retest WC to see if any recall-list-related bugs remain?

[Natasiel]

unfortunately, NM on difficulty "Nightmare" is shockingly easy.

did u forget that in vanilla WC there were lots of dangerous lvl2 & 3s in the first few turns ?

we r trying out playing WC NM on "hard" difficulty now and tell u later about it.

The random nature of the campaign can greatly reduce the difficulty level. eg: You are mainly playing woses and you meet nothing besides spears. The formula for AI spawns hasn't been touched, so you probably just hit some generous randomness or you tried an easier to debug version from the dev server.

Wesnoth 1.7.13 includes some patches by Yogibear and me, to fix 'leader's ids are changed' and 'recall list is gone' issue.
can you retest WC to see if any recall-list-related bugs remain?

Thanks to both of you. Of course I'll test, that is the least I can do to help you out on this. Time to svn sw back to trunk and recompile on my old berta, then I'll post a full status report.

On a side note, can you tell me the nature of the bug?

[Crab]

On a side note, can you tell me the nature of the bug?

Of course

The first bug is with the fact that id's of leaders were changed to player's names, preventing correct replacing of leader in the second scenario.
Fixed by Yogibear.
crab: do not override leader's id with player's name in MP. Patch by YogiHH [rev]41101[/rev]
jhinrichs: Fix AI sides not keeping their side id for mp scenarios: [rev]41104[/rev]
jhinrichs: Fix generating a wrong player_id for the side config (thanks to Crab_ for spotting this). [rev]41111[/rev]

The second bug was with code which prepared start-of-scenario save after first scenario ended. Savegame code (correctly) assumed that "if there is no scenario information in the starting pos, add the (persistent) sides from the snapshot, else do nothing, as persistence information was already added at the end of the previous scenario". But, at the end of previous scenario, persistence info was lost. So, I added an one-liner to write persistent sides to starting pos of the next scenario

crab : fix persistence in MP campaigns : [rev]41102[/rev]

[Natasiel]

As on 41175

Warning: The file you have tried to load is corrupt. Loading anyway.
Unexpected characters at the line start, value '=' at <unknown>:19090

...appears when loading either a normal or autosave. "Loading anyway" shows up only when loading as a single player. The bug is a show stopper in multiplayer as the game won't load at all.

Besides this, the recruit list no longer disappears but the hack is still required to start next scenario with the proper leader.

[edit]...same on 41192[/edit]

[Crab]

Warning: The file you have tried to load is corrupt. Loading anyway.
Unexpected characters at the line start, value '=' at <unknown>:19090

can you post a bug report about "Unexpected characters at the line start, value '=' at <unknown>:19090 " ? it can be caused by some other unrelated changes (e.g. those which were done in [rev]41154[/rev])

"but the hack is still required to start next scenario with the proper leader."

ok, I'll try to reproduce this.

[Crab]

but the hack is still required to start next scenario with the proper leader.

I downloaded WC and saw the cause of the problem: leaders are recalled by id. the bug with overriding of ids was fixed, but, of course, you still need to assign that id manually, it doesn't have a default value.
if you have leaders in [side] definitions, just add things like "id=WCp1" to leader config there in each scenario.

[Glowing Fish]

One idea I had for World Conquest (and it is just a random idea), is that it would be interesting to spice up the recruit list of both the players and the enemies with some non-faction, but still default units. (Scorpions, Giant Spiders, Wolves, that type of thing...)