iiinova — where your digital self keeps evolving

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 15/10/2026

No. I don't like this. I am displeased.

$Some unholy C++ code that reads... ''' ErrorTypeUtils::ErrorType TypingWeighting::getErrorType(const CorrectionType correctionType, const DicTraverseSession *const traverseSession, const DicNode *const parentDicNode, const DicNode *const dicNode) const { switch (correctionType) { case CT_MATCH: if (isProximityDicNode(traverseSession, dicNode)) { return ErrorTypeUtils::PROXIMITY_CORRECTION; } else if (dicNode->isInDigraph()) { return ErrorTypeUtils::MATCH_WITH_DIGRAPH; } else { // Compare the node code point with original primary code point on the keyboard. const ProximityInfoState *const pInfoState = traverseSession->getProximityInfoState(0); const int primaryCodePoint = pInfoState->getPrimaryCodePointAt( dicNode->getInputIndex(0)); const int nodeCodePoint = dicNode->getNodeCodePoint(); const int keyIndex = traverseSession->getProximityInfo()->getKeyIndexOf( primaryCodePoint); // TODO: Check whether the input code point is on the keyboard. if (primaryCodePoint == nodeCodePoint) { // Node code point is same as original code point on the keyboard. return ErrorTypeUtils::NOT_AN_ERROR; } else if (CharUtils::toLowerCase(primaryCodePoint) == CharUtils::toLowerCase(nodeCodePoint)) { // Only cases of the code points are different. return ErrorTypeUtils::MATCH_WITH_WRONG_CASE; } else if (primaryCodePoint == CharUtils::toBaseCodePoint(nodeCodePoint)) { // Node code point is a variant of original code point. return ErrorTypeUtils::MATCH_WITH_MISSING_ACCENT; } else if (CharUtils::toBaseCodePoint(primaryCodePoint) == CharUtils::toBaseCodePoint(nodeCodePoint)) { // Base code points are the same but the code point is intentionally input. if (keyIndex == NOT_AN_INDEX) { return ErrorTypeUtils::MATCH_WITH_MISSING_EXPLICIT_ACCENT; } return ErrorTypeUtils::MATCH_WITH_WRONG_ACCENT; } else if (CharUtils::toLowerCase(primaryCodePoint) == CharUtils::toBaseLowerCase(nodeCodePoint)) { // Node code point is a variant of original code point and the cases are also // different. return ErrorTypeUtils::MATCH_WITH_MISSING_ACCENT | ErrorTypeUtils::MATCH_WITH_WRONG_CASE; } else { if (keyIndex == NOT_AN_INDEX) { return ErrorTypeUtils::MATCH_WITH_MISSING_EXPLICIT_ACCENT | ErrorTypeUtils::MATCH_WITH_WRONG_CASE; } // Base code points are the same and the cases are different. return ErrorTypeUtils::MATCH_WITH_WRONG_ACCENT | ErrorTypeUtils::MATCH_WITH_WRONG_CASE; } } break; '''$

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 14/10/2026

Oh no, #creativeassembly what did you do to my precious #total_war_shogun_2, I can't even bring myself to install it after someone I played with sent me these. I assume the code you the same treatment, after they told me about their performance woes. My deepest […] [Original post on mstdn.social]

An upscaled image of a Portuguese terco waving a finger from a 6 fingered hand at a samurai who is looking away in disgust.

A unit card of a samurai on horse back with an upscaled version that clearly went through generative AI, the samurai's face is gone.

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 13/09/2026

[Analysis of perplexing slop] Saw this in a comment about AI in mathematics. Was posted in _favour_ of human creativity. A large number of people don't consider things they aren't practised in to be within the domain of meaningful creative tasks. When one […] [Original post on mstdn.social]

An incredibly busy diffusion created pixel arrangement meant to be Terrance Tao, Nikola Tesla?, Anubis, a magic cat in a hat, and an army of identical white men wielding giant pencils and math equations fighting an army of corporate robots.

There is a lot of text in the image, often labeling various actors in the arrangement as holding certain principles. Such as a robot with "corporate control" on their armour.

101 1

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 10/09/2026

Read a java string into an array of c++ chars, with a 1 byte pad, that you then write a null character to the last index... sometimes twice in a single function call! It is happening in more than half the functions I am reading right now. I know why they are […] [Original post on mstdn.social]

$c++ code that reads: char localeChars[localeUtf8Length + 1]; env->GetStringUTFRegion(locale, 0, env->GetStringLength(locale), localeChars); localeChars[localeUtf8Length] = '\0';$

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 08/09/2026

How to remove the dpad key in #heliboard 4.1 @vincentvalentine @eqe Hope you both find that easy to follow.

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 02/09/2026

Someone I know sent me a great little example for mult-touch drift typing. :-) I am super excited to get to work on this next year, once the grant is over and I can start focusing on things not on the milestones for the grant.

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 28/08/2026

Well, that is all the gains I am going to try for, in #drift_type during the grant. I think that is pretty good, considering I was aiming for 60% The problem is, we need a much faster dictionary to get the execution time back down. Right now, it is taking […] [Original post on mstdn.social]

A terminal output that reads:

loaded 1095608 gestures, in 4.298s
loaded 77 lexicons, in 3.178s
resolved 1095608 gestures, in 503.430s
naively adjusting for 16 threads...
averaged 7351µs per gesture.
baseline: 88.440
test score: 72.764

A stacked paired bar chart, the google library on the left of each pair, and drift_type on the right of each pair.

drift_type is not doing as well, but the gap is definitely closing.

overlapping line filled graphs of gesture accuracy vs word length in graphemes:
- drift_type in blue
- the google library in orange

drift_type is a fairly-consistent 15% lower until word length gets to 25 graphemes, and then the gap closes noisily.

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 21/08/2026

Ya no way someone did this twice, if they did, I quit.

visualisation of a gesture of the word "program" on qwerty

100 1

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 15/08/2026

79561 words gathered all together. Not sure how much more active gathering I'll do. I think I might be done setting a target per day, and may only do it when I am stuck waiting for things. How's everyone else doing? 😅 #heliboard

000 0

eclexic @theeclecticdyslexic.mstdn.social.ap.brid.gy · 11/08/2026

uhhh... I am sorry? Anyone wanna explain what happened with this one? Don't think I have ever had package manager tell me it doesn't want to serve me an update, because the compressed data to move from one patch version to the next is too big. #flathub #firefox

terminal output that reads:

~ $ flatpak update
Looking for updates…

ID Branch Op Remote Download
1. [✗] org.mozilla.firefox stable u flathub 8.4 MB / 123.6 MB

Error: While pulling app/org.mozilla.firefox/x86_64/stable from remote flathub: Decompressed delta part exceeds configured limit of 55726617 bytes
Updates complete.
error: There were one or more errors

000 0