mercy-coding (Patreon)

probably a short misc update today... i haven't spent much time with the beta yet but that's likely going to change realll quick here. in the mean time, jazzy volunteered to make a custom mega ring icon since we decided to combine it to work as both a mega ring and a z bracelet-- or whatever that thing was called. above is the result. if i say so, he did a pretty damn good job.

as always, the big ticket item for the test is AI. i want to be surprised, but I have kind of put off improving the AI for the last sixteen episodes, so I suppose this amount of extra time is just the price i pay. it's development debt. i think a lot about that concept lately. everything i decide to do puts me in a debt of work that must be slowly paid off through development. in a sense, i just took out a loan as far as the AI debt goes.

incidentally, with each new game that gamefreak releases, the development debt also increases since those pokemon etc have to be added/updated as well..... that's the most frustrating part. it's out of my control. and god, so many flashy new moves that will need animations. animations are another thing we took out a loan on. i don't want to think about that right now though.

so let's talk about the AI. there have been a few reports of AI inactivity in the beta that we didn't see in the alpha, and also we were having some realllll trouble reproducing them. we've seen these cases in: silver wind, autotomize, and focus energy. and we haven't been able to recreate a damn one.

so that's frustrating.

i did a little work around silver wind a few days ago to try and figure out the root of the problem, but no matter what i did with the test trainer i didn't seem to have any luck. corrected an issue with its scoring! actually there's a larger issue to be fixed there but we'll talk about that later. but i didnt find the inaction issue. we both basically came up dry.

i was thinking with as little luck as we've had with this-- with all of these--
we might just have to let it go and wait for more reports to come in during community. that hyper sucks, but what can we do?

then marcello mentioned that all three of these moves could be because of the same problem. what? well, after reports in the alpha one thing we did was add a check to all status moves that cuts their score if the user has a move that would KO, that way pokemon don't get so greedy. and apparently-- as marcello told me-- he considered silver wind and focus energy set up moves (rightfully so, but i wouldnt have thought of it) so he counted it on those too.

okay, so that's a lead, but it doesn't really explain why we can't reproduce the issue with a n y of them. out of curiosity i started taking a look at the reports for any patterns. silver wind and autotomize were both doubles with an AI partner, and they had the moves in similar menu positions, but focus energy threw a wrench into all of that. was the focus energy issue a separate, but similar one after all? i did notice that all three cases -did- have a killing move though, which correlated to the section we had added. i suppose that's some compelling evidence.

marcello took a closer look at that section, but he didn't see anything. and this was yesterday-- lemme tell you about my yesterday: it was awful. the reasons aren't worth going into and i'm fine now and there's no need to worry but i was frustrated and shaking due to unrelated things and i was not prepared to sit down and peel through a load of test scenarios and code that i could barely mentally parse. 

i had not communicated this to him though, and so not knowing my poor condition, he sent me the section of code in question. i just about snapped at him. no real reason, mind you-- but he sends code to me like that sometimes and it has absolutely never been especially helpful because i didn't even notice the three extra trees in my front yard so how the heck am i supposed to notice a typo in a wall of text??? that's not worth snapping at anyone over but i just about lashed out at everything yesterday. i didn't. but boy i was close. and for no reason. still, this was annoying. don't you realize that sending me code walls never works ou--

actually... there's three similar sections in the text wall, for best/high/medium skill trainers. the second two have their code ordered differently than the first. would that be a problem?

...no, nevermind, that doesn't make sense, i had to be misreading it. i'm tired, and out of it and actually nothing at all makes sense and i have no idea what i'm sayi--

"Ame i think that's it"

what...
oh.

all of the reports had been vs medium and high skilled trainers, but our test trainer is always set best skill to get the most out of it. of course we couldn't replicate it, because our method of testing was flawed in the first place. the best skill section had been ordered properly, but the other two needed to match. best of all, this approach applied to all three of our reported moves (and a lot of others) so we got it all in one sweep. 

one very tired, frustrated, half-assed, confused sweep.

some call it luck. others would say skill.
i call it mercy. at least one thing went right. sometimes, that's all you need.