flipboard - Link to source

Nigel Farage Breached The MPs' Code Of Conduct 17 Times, Watchdog Confirms
https://www.huffingtonpost.co.uk/entry/farage-registered-more-than-ps380000-financial-interests-late_uk_69709305e4b084d70182d794?utm_source=flipboard&utm_medium=activitypub

Posted into UK Government @uk-government-FlipboardUK


Nigel Farage Breached The MPs' Code Of Conduct 17 Times, Watchdog Confirms

The Reform leader registered more than £380,000 of financial interests outside of the accepted 28-day period.
Kate Nicholson (HuffPost)
@UK Government

mastodon - Link to source

🔴 Wat gebeurt er als Trump met één druk op de knop alle Amerikaanse systemen van Nederlandse organisaties uitschakelt?

We deden onderzoek. Wat blijkt?

🎓 81.4% van de onderwijsinstellingen raakt toegang tot documenten en e-mail kwijt

🏛️ net als en 72.6% van de gemeentes

🏢 en 70.4% van de grootste Nederlandse bedrijven

Dit en meer kun je vinden op zijnwealautonoom.nl. Daar verzamelen we van ruim 13.000 Nederlandse organisaties bij wie ze hun websites en e-mail hebben ondergebracht.

Zijn we al autonoom?

Zijn we al autonoom brengt in kaart van welke organisaties het internet in Nederland het meest afhankelijk is.
Zijn we al autonoom?
in reply to NV Access

mastodon - Link to source

Jonathan

@NVAccess github.com/nvaccess/nvda/issue…

Incompatible Add On (requires NVDA Update) shows no warning dialog when trying to install and just gets ignored

Brief summary Clicking on an .vda-addon file which manifest specifies a higher minimum_version than your installed NVDA version, it should prompt you with a dialog saying that the add on is incompa...
Jonathans859 (GitHub)
@NV Access

mastodon - Link to source

On this day in 1985, 41 years ago, Greensboro, NC, the city in which I grew up, experienced it's coldest temperature on record, -8 degrees F, or -22.22 degrees C.

I was alive for that, but being less than a year old at the time, don't have any memory of it.

My baby bedroom was the least insulated in that house, though, so that must have been fun keeping it at a reasonable temperature.

Mount Mitchell, NC recorded a temperature of -23 degrees F, or about -31 degrees C, on that same day.

This entry was edited (6 days ago)

mastodon - Link to source

honestly, the janky-ass superflous animations & effects even with all the “reduce motion” #accessibility options enabled seem to exist only to slow down my previously very snappy recent-gen (n-2) #iPhone – this isn’t event planned obsolescence, it feels like hastened obsolescence to make the #Apple money printer go brrr just a little sooner than usual, as a treat 😤

on a device that has the hardware to support 120Hz refresh rates, why am I ever seeing folders & icons get visibly drawn on my screen? 🤪

#LiquidAss #enshittification on display yet again 💩

#Accessibility #apple #iphone #enshittification #liquidass

mastodon - Link to source

Oh, cool. My NAS back at home is apparently suffering from Dying Power Supply Syndrome (DPSD) or something. It apparently improperly shut down twelve times yesterday, according to the logs.

Of course, it would do this just a few months after I spent over 2 grand replacing the storage on it with 8 new 12 TB hard drives, replacing the 6 TB drives that were all starting to die.

Looks like maybe a new NAS is in my future when I get back... whenever that is.

This entry was edited (6 days ago)

glitchsoc - Link to source

Brrr... new introduction time cause my old one is already outdated lol.

I am Sinclair-Speccy, your local resident computer nerd that can’t actually do a lot of tech stuff like people think. My primary interests are #Linux, #OldTech, and #OStan. You’ll usually find me boosting retro tech content and occasionally posting it myself, mostly retro tech ads.

My interest in emulation all started when I managed to run Yggdrasil Linux from the '90s (I know, so interesting /s). Despite having used XP as my first OS other older OSes (mostly Linux ones) interest me more if you look at the media I've posted on my account besides the tech ads.

I’ve moved around a bit online as I started on .art as my friend was there, then bitbang.social and now oldbytes.space.

One of my favourite niche fandoms is OS-tans which is turning operating systems and hardware into anime characters, though I tend to call them tech gijinkas. I even have my own tech gijinkas universe that is hosted on my site.

As for some other stuff I enjoy #mythology, #transformers, #cats, #fantasy, #nature such as #flowers and #plants and #animals, #steampunk, #starwars, #photography, #art, #space, #cyberpunk and #scifi... there is more but I'm not writing them all.

You can check out my website here: sinclair-speccy.github.io/The-… and I hope to expand it into a tech journalism platform or something, though motivation and its static nature kind of stop me...

Unlike in the past I'm not active as I used to be, sadly.

#RetroTech #Retro #RetroTechnology #RetroComputing #VintageComputers #Introduction

#linux #art #cats #introduction #photography #scifi #cyberpunk #nature #fantasy #starwars #retrocomputing #animals #space #retro #steampunk #mythology #retrotech #plants #flowers #oldtech #ostan #transformers #retrotechnology #vintagecomputers

reshared this

mastodon - Link to source

Edit: I didn’t notice that this article is from last June.

Farewell to autumn: Danish Digital Ministry turns its back on Microsoft.

At the Danish Digital Ministry, all employees should be able to do without Microsoft. Instead, Linux and LibreOffice will be used, says the minister.

heise.de/news/Von-Word-und-Exc…


Abschied bis Herbst: Dänisches Digitalministerium kehrt Microsoft den Rücken

Beim dänischen Digitalministerium sollen alle Angestellten ohne Microsoft auskommen. Stattdessen werde man Linux und LibreOffice nutzen, sagt die Ministerin.
Martin Holland (heise online)
This entry was edited (6 days ago)

mastodon - Link to source
Sorry guys. I had to remove the timeline gaps system. It was just not working correctly at all, and was only causing issues and annoyances. Not sure if I will ever reimplement something like this in the future, however if it's of concern, you can disable caching entirely, at which point the app would load a fresh copy of the timeline upon every client start like before.

mastodon - Link to source
Thank you for using FastSM! We update the app hourly to keep your experience running smoothely! For best results, keep your app up to date!
FastSM will never ask you to review the app. Mainly because it can't, because there is no review system attached to GitHub.
Anyway, This update includes but is not limited to bugfixes, performance improvements, new bugs, unintentional performance degradations, issues, disappointments, problems, issues, and many more issues. Please keep your updates turned on to experience all of this, and more! Our developers will never tell you that they are taking a nap, or mowing the lawn, unless, of course, they are indeed taking a nap, or mowing the lawn. Why our developers would be mowing the lawn in a changelog, who actually knows. Anyway, more bugfixes and performance improvements time! Please keep your updates turned on for more bugfixes and performance improvements! Thank you for using FastSM! We promise that as we continue to add more bugfixes and performance improvements, eventually we will add so many bugfixes and performance improvements that we will be forced to rename the app to SlowSM, or potentially even BuggySM. So stay tuned for this eventuality. Anyway, again we hope you enjoy using FastSM!

reshared this

mastodon - Link to source
I don't think I like this new world, you guys. Every time Youtube Music plays me a song I haven't heard before, I wonder if it's AI. Hmm, that's a lot of releases in a short time. Is the voice quality a bit odd because the track quality is low, or because I'm imagining it, or because it's AI? Wait, those rhymes are really weak. Did an LLM just spit those out?

mastodon - Link to source
There are two primary reasons I like watching the output from Claude or Gemini, or any model for that matter: #1. you can learn some really cool shell tricks, #2. I love watching how it transforms my thoughts, e.g. from a markdown spec file, into a plan. Even the most careful planning on my part always has some gaps, high verbosity, etc. For that reason, I always request the model to ask followup questions.
#1 #2

mastodon - Link to source
In today's world of autocorrect, why do so many places where people write from their phones always include contractions with missing apostrophes? You'd think any modern operating system would see "isnt" and immediately correct it to "isn't". Same for "ive", "dont", and so on. I can understand the apostrophe being multiple taps away, and so not getting added manually, but why wouldn't software put it back in?

mastodon - Link to source
#TimeToTehOn...Good morning. I like FastSM rather more than tc, only probably because it loads faster, and takes up less RAM space. However, I won't be running it networkly like I can TC, for the simple reason that I cannot localise user data, or not that I know of. I don't use Telegram that much on the pc anyway, because most who have my details there, also follow me on mastodon, so no use boring people on telepaw if I already posted here anyway. BlueSky I don't have, and probably will never have either...Between my facebook/mastodon.africa/fluffy.family accounts I already have 3 social media related accounts, and that seems to be enough for now. However, knowing me, I may just fall for Bluesky, like I fell for the Seika Mini2 Smart Braille Notetaker...so I'll never say never, simply because I know myself well enough to know that curiosity might indeed push me to get bluesky regardless of the current reluctance. Anyway, have a nice day on this humpday.
#timetotehon

mastodon - Link to source
It occurs to me that I haven't yet tested FastSM's ability to upload media to Mastodon. So I hereby present something which I call, "Girl, You Know it's False." I take the intro to the Milli Vanilli song "Girl, You Know it's True," where a man and a woman are talking to each other on opposite stereo channels, then I deliberately use and normalize the channel where each person is *not* talking. What you hear is a really echoy conversation, with almost certainly tape hiss and even some 50 Hz hum in the background. Then at the end the two orchestra hit type things also sound weird. Anyway here goes.

mastodon - Link to source
Well, I got GPT5 Pro mode again, asked it to help refactor my driver because I don't do a good enough job myself of that, and now it's been thinking for 88 minutes. Either it's going to give me a big disaster pile, or I somehow got it stuck. Longest time I've seen it think was like 65 minutes haha. But what the heck can you do for 88 minutes! This better be good. I feel like I'm asking a real coder to do the work, but worse, because it's not an actual real coder but a program, so it's more disappointing.
in reply to Brandon

mastodon - Link to source

Tamas G

@serrebi @FreakyFwoof @stormproductions @xogium oh interesting! May give it a try at work, we have access to both. I have Google One plan which I share with my partner so we both get Gemini, they prefer it and honestly OpenAI has never allowed for family sharing like that, quite unfortunate. I feel like I need both in my toolbox. OpenAI (pro for now, we'll see if I make enough use of it this month to count) for those "think through and patch files direct inside container for you" moments, Gemini for the "review my code I wrote and that it's not trash and passable to an AI to work on", and they both look at different areas of the code. GPT will get it done even if you give it a bad idea to get done, Gemini's for sure a lot more cautious. So I can't decide on either these days, usually my workflow is augmenting one's response with the answer of the other, which makes me wonder what a powerful tool could be built if it just passed your prompt between the 2 AIs to solve, I'm sure someone's thought of that by now.
@Brandon @Andre Louis @Casey Reeves @Storm
in reply to Brandon

mastodon - Link to source

Tamas G

@serrebi @stormproductions @FreakyFwoof @xogium yeah, even in the "AI pro" plan :) + the 2 TB Gmail is nice, well, that's why I was initially subbed to it but they migrated people over with a slight price bump. economically speaking I think if I wanted to get the most use out of AI, Gemini does win by a longshot, even if GPT's now trying to compete with their $8 "go" plan LOL
@Brandon @Andre Louis @Casey Reeves @Storm
in reply to Brandon

mastodon - Link to source

Storm

@serrebi @FreakyFwoof @xogium I have installed Gemini's CLI, but I'm either too stupid with CLI's in general, or I don't even know how to navigate with NVDA. I tried typing /settings so I can see what settings I can tweak, and I couldn't even see the darn list of settings with NVDA. Any tips? I mean, any shortcuts for either NVDA or Gemini's CLI that I have to learn to start using it properly? Where do I actually pass the screen reader flag? My questions probably sound sily and stupid for somebody used to CLI's, but I truelly can't get the hang of it.
@Brandon @Andre Louis @Casey Reeves
in reply to Casey Reeves

mastodon - Link to source

Tamas G

@xogium @FreakyFwoof @stormproductions @serrebi oh gosh. that's aweful. Do you know if it behaves any better with Mac OS VoiceOver? For work I'm using both side by side, so perhaps I'd have more luck with Gemini Cli under Terminal? Even though VoiceOver's terminal support isn't known for being great, I do wonder if it would do better.
@Brandon @Andre Louis @Casey Reeves @Storm
in reply to Winter blue tardis

glitchsoc - Link to source

Casey Reeves

@tardis @FreakyFwoof @stormproductions @serrebi In any case: xogium.me/the-text-mode-lie-wh…

Sections titled Case Study: The gemini-cli Madness and The “Stale Bot” excuse: A Case Study in Neglect seem pretty relevant to this.

It doesn't matter if you use the screen reader option, or if your gemini-cli config has it enabled, this will happen.

And here's to why you shouldn't do stupid apps for a terminal that try to be too modern and get written in node.js using the ink framework.

The Inclusive Lens

2026-01-05 05:20:28

writefreely - Link to source

The text mode lie: why modern TUIs are a nightmare for accessibility


The mythical, it's text, so it's accessible


There is a persistent misconception among sighted developers: if an application runs in a terminal, it is inherently accessible. The logic assumes that because there are no graphics, no complex DOM, and no WebGL canvases, the content is just raw ASCII text that a screen reader can easily parse.

The reality is different. Most modern Text User Interfaces (TUIs) are often more hostile to accessibility than poorly coded graphical interfaces. The very tools designed to improve the Developer Experience (DX) in the terminal—frameworks like Ink (JS/React), Bubble Tea (Go), or tcell—are actively destroying the experience for blind users.

The Architectural Flaw: Stream vs. Grid


To understand the failure, we must distinguish between two distinct concepts often conflated under “terminal apps”: the CLI (Command Line Interface) and the TUI.

  1. The CLI (The Stream): This operates on a standard input/output model (stdin/stdout). You type a command, the system appends the result below, and the cursor moves down. This is linear and chronological. For a screen reader, specifically kernel-level readers like Speakup, this is ideal.
  2. The TUI (The Grid): This treats the terminal window not as a stream of text, but as a 2D grid of pixels, where every character cell is a pixel. It abandons the temporal flow for a spatial layout.


Case Study: The gemini-cli Madness


Let's look at a concrete example: gemini-cli, a tool written in Node.js using the Ink framework. On the surface, it looks like a simple chat interface. But underneath, Ink is trying to reconcile a React component tree into a terminal grid.

When you use this tool with Speakup (Linux) or NVDA (Windows), the application doesn't just fail; it actively spams you.

Because the framework treats the screen as a reactive canvas, every update triggers a redraw. When the AI is “thinking,” the tool updates a timer or a spinner. To do this, it moves the hardware cursor to the timer location, writes the new time, and moves it back.

For a sighted user, this happens instantly. For a screen reader user, this is what you hear:“Responding... Time elapsed 1s... Responding... Time elapsed 2s... [Fragment of chat history]... Responding...”

It drives the screen reader mad. The cursor is teleporting all over the screen to update status indicators, spinners, and history. Speakup tries to read whatever is under the cursor at that exact millisecond. You end up hearing random bits of conversation mixed with timer updates, making it impossible to focus on what you are actually typing.

Worse, lets pretend that you've somehow managed well with speakup so far, but that you want to do some work with nvda. Maybe paste an error you're getting on windows. So you open your terminal, ssh into your linux box, attach to your screen session and paste your text.

The result is an immediate crash of the screen reader (NVDA) or massive system instability. Why? Every time you type a character or paste text, the application triggers a state change. The framework decides it needs to re-render the interface. Because the conversation history is part of that state, the application attempts to redraw or re-calculate the layout for thousands of lines of text instantly. The more messages you have in a conversation, the more this will happen. And no, you can't just avoid this by using insert+5, the key combo supposed to avoid announcing dynamic change of content.

The Lag Loop


Furthermore, frameworks like Ink running on single-threaded environments (like Node.js) suffer from massive performance degradation when the history grows. If you paste a large block of text, the system has to calculate the diff for thousands of lines.

This causes input lag. You press a key, and you wait. You can wait up to 10 seconds for a single character to echo back. The system is too busy calculating how to redraw the screen to actually process your input.

Why The “Old Guard” Works (nano, vim, menuconfig)


Sighted developers often ask: “If TUIs are bad, why do you use nano, vim, or menuconfig?”

The answer is not that these tools handle the cursor perfectly by default. The answer is that they allow you to hide the cursor entirely.

1. Hiding the Cursor (nano, vim)


In tools like nano or vim, usability depends on turning off features that track cursor position. If you run nano with options that show the cursor position (like --constantshow), or if you use vim without specific configuration, the experience is broken.

When the cursor is visible and tracking is active, Speakup prioritizes the cursor's location update over the character echo. Instead of hearing the letter “a” when you type it, you hear “Column 2”. You type “b”, and you hear “Column 3”.

These older tools succeed because they allow you to disable this noise. You can configure them to suppress the visual cursor or status bar updates, forcing the screen reader to rely on the character input stream rather than the noisy coordinate updates. Modern frameworks rarely offer a “no-cursor” or “headless” mode; they assume the visual cursor is essential.

2. Single Column Focus (menuconfig)


Tools like the Linux kernel's menuconfig work because they enforce a strict, single-column focus. Even though there are borders and titles, the active area is a vertical list. The cursor stays pinned to that list. It doesn't jump to the bottom right to update a clock, then to the top left to update a title. The spatial complexity is kept low enough that the screen reader never gets “lost.”

3. The Lost Art of Scrolling Regions (Irssi)


Irssi is the gold standard for accessible chat, but not because of luck. Irssi was built over 20 years with a custom rendering engine that utilizes VT100 Scrolling Regions.

When a new message arrives in Irssi: 1. It tells the terminal driver: “Define a scrolling region from line 1 to 23.”2. It sends a command: “Scroll up.” The terminal moves the bits up. 3. It draws the new text at the bottom of that region.

Crucially, it handles this in a way that minimizes interference with the input line. It relies on the terminal's hardware capabilities rather than rewriting every character on the screen manually. Modern frameworks ignore these hardware features in favor of “diffing” the screen state and rewriting characters, which is computationally heavier and hostile to accessibility.

The “Stale Bot” excuse: A Case Study in Neglect


Google and the maintainers of gemini-cli pretend to care about accessibility. “Pretend” is the operative word here. If you look at the repository, critical accessibility regressions like Issue #3435 and Issue #11305 have been left to rot. There is no discussion, no roadmap, and no fix. Even worse is the fate of Issue #1553, which was supposed to track these accessibility failures. It didn't get solved; it got silenced. It was closed automatically by a bot with this generic dismissal: > Hello! As part of our effort to keep our backlog manageable and focus on the most active issues, we are tidying up older reports. It looks like this > issue hasn't been active for a while, so we are closing it for now.”

This is unacceptable. Closing an accessibility report because the maintainers haven't touched it in months is not “tidying up”; it is hiding evidence. It effectively says that if a bug is ignored long enough, it ceases to exist. It boosts the project's “Closed Issues” metric while leaving the actual software unusable for blind users.

Conclusion


If you are building for the terminal and care about accessibility, stop using declarative UI frameworks that treat the terminal like a canvas.

The “modern” TUI stack has optimized for the developer's ability to write React-like code at the expense of the machine's ability to render text efficiently.

If you cannot guarantee that your application allows the user to hide the cursor, or if you rely on aggressive redrawing to show spinners and timers, you are building an inaccessible tool.

For the blind user, a dumb, linear CLI stream is infinitely superior to a “smart” TUI that lags, spams, and scatters the cursor across the screen.


@Brandon @Andre Louis @Winter blue tardis @Storm
in reply to Winter blue tardis

glitchsoc - Link to source

Casey Reeves

@tardis @FreakyFwoof @stormproductions @serrebi It works, up to a point. And I've done a whole lot this past month alone to really know what I'm talking about and not just say nonsense.

The web ui will work but when your code is now spanning multiple files -- over 40 in my case, it creates problem. Gemini on the web constantly loses track of which file does what, which leads to very stupid things like:

Me: your fix does not work, I still get the error about session not properly disconnecting
Gemini: well of course it won't work, I've analyzed your code and it's still the older one, do this and it will work.
Me: ... of course it doesn't work, and besides you're refering yourself to an old version of this code.

@Brandon @Andre Louis @Winter blue tardis @Storm
in reply to Storm

glitchsoc - Link to source

Casey Reeves

@stormproductions If they are still up after all the crap they go through right now, that is...

Trials for the death of several people that trusted gpt with things, claims that chatgpt can even be trusted with health advices now when it clearly can't -- I'm not joking! I saw an ad about it!

Then also the Musk case. Oh dear oh dear. Now I don't like Musk at all but truth is, he's apparently got 60% chance of winning his case against OpenAI. And he asked for a jaw dropping $134b iirc.

I think OpenAI will sink sooner than later.

@Storm
in reply to Casey Reeves

mastodon - Link to source

Tamas G

@xogium @stormproductions I do use Gemini, quite often to double-check code GPT wrote, because it will scan through it a lot faster. But (at least on web) it's horrible at pasting back code you gave it exactly 1 to 1 still, like it'll omit sections that it's pasting back of the code I gave it, and I'm yelling at it "give me full pieces! No omissions!" Haha. It's gotten better but GPT having an entire container inside to access and run Linux commands, not to mention I've uploaded a 30-40 MB zip file to it with my entire Gitub repo and Gemini just yells at me, "file is too large." LOL.
@Casey Reeves @Storm

bridgy-fed - Link to source

Pravidelné kombinování různých druhů fyzické aktivity může být nejlepší pro prodloužení délky života, ale souvislosti nejsou lineární, což naznačuje možný optimální prahový efekt www.news-medical.net/news/2026012...

Diverse exercise routines asso...


Diverse exercise routines associated with reduced risk of death

Regularly doing a mix of different types of physical activity may be best for prolonging the lifespan, but the associations aren't linear, pointing to a possible optimal threshold effect, suggests research published in the open access journal BMJ Med…
News-Medical

mastodon - Link to source
God, Mac OS is really awful these days! A notification comes in, no VO focus on it, so hit VO+F1 twice, notifications menu has "go to Notification Center" and..."No items in Notification center!" Also, "Go to Notification Center" doesn't actually do anything! If something this basic doesn't work, it's no wonder VO is so bad. Case in point, every time my Mac restarts now, it comes up saying "Installer has no Windows." I have to use VO+Shift+Space to click the mouse because it's hanging out in the Login Window, while Vo's focus refuses to move from this blank Installer Window no matter what. Pressing VO+F1 just says, "Authentication required." Shit, maybe I just need to blow the OS away and start over.

mastodon - Link to source
New: NV Speech Player finally gets a "pause mode" setting, like Eloquence has. This should not chew into lines with say-all, but rather does clause-based punctuation pause detection. And, it does appear to work. You can choose between "off," "short," and "long." This should solve a lot of complaints around words "merging together" after punctuation. I also began my first pass at abstracting parts of the driver out: _frontend.py and _dll_loader.py. Logging was also improved, I cleaned up over 40 try/catch exceptions, We're not trying to play Pokemon here.
download: eurpod.com/synths/nvSpeechPlay…
Oh! and for those wondering "where's classic pitch?" It's located in the Voice settings dialog, you'll find it as a checkbox.
This entry was edited (6 days ago)
in reply to Tamas G

mastodon - Link to source

Cleverson

thanks for the last corrections concerning m/n between vowel and consonant for pt-br. Now, I've just added a phoneme which I've called "brã", for Brazilian ã, which consisted of a single change: take ã and change cf4 from 3300 to 2600. then I added to pt-br.yaml:
- from: "ã"
to: "brã"
Now, when I command it to speak a word like "região", the new phoneme appears to be used, but it's much less stressed and somewhat shortened, or at least de-emphasised. Can you give a look please?
in reply to Cleverson

mastodon - Link to source

Tamas G

@clv1 hmm. try and replace your packs folder at eurpod.com/synths/packs-ptUpda… - I'm experimenting on this one. What you’re hearing isn’t really “less stress,” it’s that the new vowel is darker, so it feels less punchy even if the timing didn’t change. In our latest pack we kept your approach, but made the change in a more balanced way so the synth doesn’t “half use” the old sound: we added a dedicated Brazilian ã variant and mapped pt-br to use it, making sure the vowel stays consistent across the voice filters. If you still want it to pop more, the best next tweak is a tiny loudness bump on that new ã variant rather than changing stress/timing.
@Cleverson
in reply to Zvonimir Stanecic

mastodon - Link to source

Tamas G

@asael @clv1 haha not sure. Depends on how well base pairs are defined right now. You could always add it inside the driver, here we have the list of languages, if you formatted it the same way as that string and used HR for Croatian it would speak today but with perhaps missing sounds. Then we map those missing sounds to specific phoneme pairs in hr.yaml. Generally many sounds can be reused across languages but there's definitely edge cases especially with glides, fronted sounds, ETC.
@Zvonimir Stanecic @Cleverson
in reply to Tamas G

mastodon - Link to source

Cleverson

Now to another issue: I need the distinction to be more noticeable when there is an r between consonant and vowel, i.e. in syllables like bra, cra, fra, tra, the consonants probably need to be more separate from each other. I have added some settings to pt-br.yaml, namely those concerning stop gaps, but I don't know yet how to deal with them... What is the correct setting for this case?
⇧