@tess@mastodon.social
@tess@mastodon.social avatar

tess

@tess@mastodon.social

Software wizard. Roller derby athlete. Mom. Gamer.

Warning: crush hazard

This profile is from a federated server and may be incomplete. Browse more on the original instance.

tess, to random
@tess@mastodon.social avatar

Logged into my bank account to pay bills tonight and my entire payee list is gone. I'm going to chalk it up to a microservice going down at their end and not immediately panic; it's a bank and way after hours so the odds that they have any IT on the problem is low.

Now, if that shit isn't back tomorrow, I will absolutely panic.

tess,
@tess@mastodon.social avatar

It's not that I couldn't reconstruct the list (literally, from paper bills as they come in) but rather that I am paranoid about ever paying a bill late and am very bad at checking the mail so I need that list to remind me when I might owe something 😅

aeva, to random
@aeva@mastodon.gamedev.place avatar

so I'm trying to work out what the exact flags cmake is setting for gcc for a debug build, and the docs just say "the usual ones", and my searches are saturated with stackoverflow posts that are topically adjacent but unfortunately irrelevant to my problem at hand because I'm trying to work out why my debug builds have horrible numeric issues or something but my release builds don't

and so I get the clever idea of looking at what cmake generated, and oh my god that is a lot of cmake

tess,
@tess@mastodon.social avatar

@aeva @mcc after looking at both the GLM documentation and source code: this tracks

Holy shit, www appears to be them using macros to implement a poor man's version of extension methods from C#; this is EXTREMELY CURSED

(Also, vecN appears to support /= with a scalar, so you could just do Tmp /= w; return Tmp?)

tess,
@tess@mastodon.social avatar

@aeva @mcc I mean, it's also likely they have defined scalar operator/ in some equally absurd sort of way, so honestly I'd just go for it and let the compiler tell you know.

tess,
@tess@mastodon.social avatar

@aeva @mcc but based on 20+ years of dealing with gcc I would say that it is almost certainly some sort of cursed interaction of a macro referencing the same object several times in the middle of a return statement, interacting with local destructors in a way that's not technically incorrect under C++ rules but also extremely unintuitive*

  • by which I mean "wrong for any sensible definition of wrong created by a human being who has read the C++ spec, but not to the kinds of people who code gcc"
tess,
@tess@mastodon.social avatar

@aeva @mcc the fact that compiler optimization seems to eliminate the issue adds some strength my conviction on this; the ordering (or even execution) of destructors/memory cleanup would depend heavily on what kind of build configuration you were using.

tess,
@tess@mastodon.social avatar

@aeva @mcc I mean frankly, I'm also riding with andi on the "what the hell does dividing two vec3s do?" train; it really feels like you should be dividing by a scalar

tess,
@tess@mastodon.social avatar

@xgranade @aeva @mcc it's clear glm is from the OVERLOAD ALL THE OPERATORS EVEN IF IT MAKES NO SENSE era of C++, but the net result of that is that a lot of them end up just kind of being mystery operations

tess,
@tess@mastodon.social avatar

@xgranade @aeva @mcc I created a string library in middle school where unary + was "to upper case" and / was "remove all occurrences of the second operand from the first".

It was extremely stupid but I felt very clever.

tess,
@tess@mastodon.social avatar

@xgranade to be fair, << has the benefit of being lower precedence than all ref/deref and arithmetic operators, and more or less make sense in context.

They were also added way before variadic templates, which is how you do that sort of thing today (though in some ways they are more elegant).

The thing that was gross about my library wasn't so much the operators, but the ring buffer it used for intermediates, so you could overrun it with too complicated of an expression.

tess,
@tess@mastodon.social avatar

@aeva @mcc I suppose it's silly to ask if there's any chance your LastFoldInverse is horked from a different operation?

Also, have you tried breaking out the operations, something like this?

 vec3 Tmp2 = Tmp.xyz;  
 Tmp2 /= Tmp.www;  
 return Tmp2;
tess,
@tess@mastodon.social avatar

@aeva @mcc yeah I guess I would probably play it safe and avoid using more than one swizzle per line of code? I dunno. Clearly between the library and the compiler there is something bad happening.

mcc, to random
@mcc@mastodon.social avatar

Been trying to learn basic Korean. Duolingo hasn't let me graduate out of Hangul yet but I seem to be doing ok with the Hangul at least. The one thing I'm really struggling with is the distinction between ㅔ and ㅐ. Every source I've seen attempt to explain it says that they used to be different vowels but over time they became the same vowel. But you still need to know which is which for spelling words.

tess,
@tess@mastodon.social avatar

@mcc they do have different pronunciation in careful speech and in North Korea (though for most speakers they are merged).

But basically this is no different than in English how you can spell /e(ɪ)/ "ai", "ae", "a_e", "eigh", etc.

tess,
@tess@mastodon.social avatar

@mcc at the same time, not having to remember which words have open vs closed e pronunciation is a blessing (contrast: French, Italian)

foone, to random
@foone@digipres.club avatar

I was about to complain about this program using memcpy(foo,bar,4) so often, on a 32bit architecture, but then it hit me:

ALIGNMENT RESTRICTIONS.
shakes fist in x86-background

tess,
@tess@mastodon.social avatar

@foone apropos of nothing, there needs to be an alignment alignment chart.

tess, to random
@tess@mastodon.social avatar

People who are Very Concerned About Free Speech come in exactly three flavors:

  • Rich assholes who already have massive platforms

  • White people who really want to say the n-word without repercussions

  • Doctors and teachers who are literally being banned from talking about specific topics by conservative state legislatures

🎵 one of these things is not like the others 🎵

tess, to random
@tess@mastodon.social avatar

Tried some new tactics and finally beat Alecto. Now that I have Tiche, I'm coming for you, Malenia!

tess, to random
@tess@mastodon.social avatar

I wrote a thing on historical and present exclusion of minority groups from sports and specifically women's sports.

CW for racism, lesbophobia, and transphobia

https://cohost.org/leftoblique/post/1467361-some-thoughts-on-mi

tess, to random
@tess@mastodon.social avatar

Malenia thoughts:

I've gone at her 15-20 times now, and gotten her about halfway through phase 2.

Ducky dance isn't the hard part; I can usually dodge most/all of it. It's just the absolute relentlessness of the fight; she never lets up and any mistake sets you back significantly.

It's a good fight. Fair. But man it's hard as fuck (especially for a mage build with vigor in the low-mid 30s - yikes!)

tess,
@tess@mastodon.social avatar

@onelson To be fair, I'm using spirit ashes. Though, with her, it's always a trade-off between "helps you distract/stance-break her" and "provides her with a personal HP juice box".

In my first playthrough with my unga bunga build the Mimic was the braindead option, but I'm not intentionally using her this time (and it probably wouldn't help much anyway tbh).

tess,
@tess@mastodon.social avatar

@onelson yeah there aren't very many good options. Redmane Soldiers seems to be a net positive, since they're hyper-aggressive, have good defense, use fire, and do immense poise damage.

Stormhawk Deenh and Lhutiel also seems like good choices purely as a distraction, but they're not really gonna help you fight her, just make it so she's not targeting you 100% of the time.

(Having a flying or teleporting summon changes things a LOT since the AI struggles with their movement.)

tess,
@tess@mastodon.social avatar

@onelson I'm doing this run with very specific summons:

  • gargs (my boys!) are my default, especially against soft targets
  • greatshield dudes when I need space
  • redmane soldiers for tree spirits and avatars, because fire
  • lhutiel for pure distraction
  • deenh for when the floor is lava

I'm also not summoninig for every boss, and I've cleared every field boss and evergaol except Alecto. But I also don't want to skill my way through 140+ bosses.

tess, to random
@tess@mastodon.social avatar

Things that would make quote-boosting better than quote-tweets:

  • opt-in, or at least opt-out
  • bottom text on quoted toots instead of top

The first has been requested by a number of people. The second, I think, changes the tenor of the quote-boost completely.

Top text allows you to set the context for the boost before people read it; it's perfect for bad faith boosts and dunking and people trying to spread intentional bad takes.

Bottom text makes a QB much more of a "yes, and..."

tess,
@tess@mastodon.social avatar

Top- vs. bottom-text is one of those ways even a small difference in UX can completely change how people use a tool.

Look at Tumblr, which is all bottom-text; even when people reblog to dunk it's almost always dunking on a reply to a post they agree with.

Reblogging a thread with a horrible or offensive OP into into your followers' streams is generally considered bad form.

The same would almost certainly be true of bottom-text quote-boosts.

tess,
@tess@mastodon.social avatar

Another benefit of bottom-text quote-boosts is, when signal boosting someone (and especially someone from, say, a community you don't belong to) it lets them speak for themselves.

This is one of the things that always made me uncomfortable about Twitter's top-text QTs: I have to provide context before my followers even read the tweet I'm trying to promote; it forces me to basically speak for that person, and I'm not always comfortable doing that.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • ngwrru68w68
  • everett
  • InstantRegret
  • magazineikmin
  • thenastyranch
  • rosin
  • GTA5RPClips
  • Durango
  • Youngstown
  • slotface
  • khanakhh
  • kavyap
  • DreamBathrooms
  • provamag3
  • tacticalgear
  • osvaldo12
  • tester
  • cubers
  • cisconetworking
  • mdbf
  • ethstaker
  • modclub
  • Leos
  • anitta
  • normalnudes
  • megavids
  • lostlight
  • All magazines
    •