Amidst Community Release (Patreon)

so we're going on close to two weeks for the community release and it's honestly been really clean. there have been a few significant issues but they were easy to patch and pretty much all of the reports i've gotten since ohh day four have been quick fixes or just repeats of what's already done...

i mean i was fully expecting to make another post regailing y'all with tales of the bugwars but there's only been one really ?????????? issue and i ended up talking about it on the discord... those who read about it may want to skip the below section since it's just more detailed account of it.

basically aegislash in the fairy tale field were getting way too strong for no apparent reason-- both player and opponent, but opponent moreso. this has been true for all of E18 testing, although wasn't the case during E17.
i first saw this reported during beta, where one of my mono extravagnza friends (tony, on Steel type) OHKO'd Tania's Mega Scizor with the steel-type sacred sword. They ran some calcs and saw it was wrong,... and reported it. So I thought something might have been off with the damage calculations, and had Cass test it (as she has kind of become our go to QA gal somehow), and she found that, although the damage rolls were a liittle higher than expected, we couldn't create the OHKO issue.

Well that's apparently cuz it had to be out for a couple turns first. The issue wasn't with damage calculation, it was with stat stages. We didn't figure that out until someone else on the forum reported it from Community-- that their aegislash had gotten maxed attack before he even tried to use Sword's Dance. 

Cass went back at it again, and we could see an issue, but then things got weird. We saw it was only happening vs Titania, and only on that field, and that's fine but then...
only against Sandslash?
and not when Sandslash was a low level.
and not when Sandslash was alone,
but if it had even one pokemon on the bench it would wig out.
We were deeply confused.

Cass ended up going to sleep a bundle of confusion, and I talked to Marcello about it that night. When I mentioned the level was making it not do the thing, he was immediately suspicious that this could be an AI issue, much like the Memestalk incident, where a misplaced check caused the AI to just make the thing it was asking if it was true, always true. However, the similarity turned out to be invalid since unlike the memestalk, Sandslash had a predetermined moveset...

Still, he remarked the issue might be with the part of the AI we added in E18 that anticipates aegislash's formswap to properly check the damage of its other form. He did a quick check over that code, but saw no issues. That should be fine-- we actually specifically wrote that AI section with this potential issue in mind, to prevent it. In order to do that, we make a theoretical clone of the Aegislash on the field, and then change that to Blade Form and give it Fairy Tale's attack boost so that the actual aegislash on the field doesn't get it.

i suddenly understood.

i am sure i have told before, recently even, how an old issue with how Essentials' duplicates Pokemon caused anything to be done to a Ninjask to also be done to its corresponding Shedinja.
We cloned the Aegislash. It was duplicating on to the real one.

No, that can't be it, Marcello said, minutes before realizing that that was, in fact, it.

The AI was processing that part of the code several times a turn when making its decisions, and so despite only looking at a clone, the real Aegislash was getting the Fairy Tale attack boost several times a turn as well.
thanks obama.

--

but i mean like, other than that it's been so straightforward. If we didn't stop for the Online stuff, which granted, was very important to do, we probably could have gotten public up like four days ago.
I dunno if we just lucked out or what, but fucking hats all the wayy off to the beta testers.
Even the things we had to patch are because of changes we made to the beta (wasteland bug was fixing long issue with hazards, pokedex quest was adding poipole... but honestly that was an Essentials-is-jank issue too...)

idk, it's clean.

Speaking of clean, check out this sweet strat from our bug mono friend, Sheep, who is apparently a true artist:

Sheep [Bug] vs Aya

https://www.youtube.com/watch?v=c2J7FHoJoEs

really, i'm so proud of him finding a way to make that move matter. he saw Aya as his biggest roadblock for most of his run since that's what separates him from the Move Tutor. and Nidoqueen's Stealthrocks would just OHKO most of his pokemon outright.

--

since there's been not so much for me to do on reborn this week, i've been spending a lot of time on starlight, especially in the last few days. since i've been focusing on the battle system, and more specifically fixing some quirks that have arisen in it, i don't have a lot to really show for it.
...did i post the wait thing? i don't think i did.

I finished the wait system:

... but that was a while ago. oops.
this diverges from fire emblem's system, but using facings is a staple of tRPGs that I happen to like, so, you know.

what else did i do...

• reworked cell grid for maps larger than the window size (i was so worried when i found out that was an issue but easy fix ok)
• fixed movement ranges to not display attack cells on spaces that can't be reached due to an ally standing in the way
• implemented fire emblem's one click attack-move shortcut (wow this helps the feel so much, i didnt realize before)
• fixed issue with being able to select to attack empty squares  and just kind of drifting upwards into oblivion instead
• fixed issue where parts of another unit's cell grid could remain left over if you moved very quickly back and forth over a couple adjacent unit
• basically had to rewrite the cell grid handler entirely for the above tbh buuuut this is fine cuz i kept having issues with that and now it's pretty robust and accounts for multiple grids for when i implement fire emblem's click-enemy-range thing on.
• adjusted tbs so that multiple units with the same name can exist (will  make it easy for me to make clone reinforcements)
• probs some other junk

and last night i just implemented overworld unit death. currently every unit dies after fighting at all ever because there's no HP values yet. but that's fine. i'll get there later.

tonight i will implement turn ending when all units have been moved, and then suddenly the whole turn loop will be done. that is basically phase 1 complete.

--

so from the sound of it and assuming nothing major suddenly pops up, chances are the community release will go up tomorrow, after we test the improved online functuality. which isn't a word but looks like it should be one.

in that, i also gotta get some updated screenshots for 18 for the download page. I have a couple already that I wanna use, (battle when partnered with Hardy, sleeping Kommo-o) but I'm curious if anyone has any other particular screenshots /ideas therefor. I'd probably choose at least one more. 

after that! well, actually starting tomorrow Cass is coming to visit and stay with me again. those around maybe remember she did so in January, and it worked out to happen again. maybe I will get her on a stream if she is so inclined.
aside from that though, i don't anticipate either getting a lot of time to work or post during the week she's here, so i may be fairly quiet. there's a strong part of me that wants to just jump right into 19, but there's a solid handful of things i need to work up and plan for properly before diving into it. when i have space, that should be fine. for now, it's probably a break.

cheers until later~