Crazy to me how GPT-5.1 now really pushes you to get pro. After a certain point in your conversation, it will tell you that you've hit your web browsing limit for that chat. Not for your entire session, mind you, just that conversation's allowance gets capped. So then it gives you this little hint of encouragement, "upgrade to pro for less limits." Really? OpenAI realized that $20 a month from their small customer base isn't enough profit so now the hammer comes down on it. Let's face it, most people use GPT for free, because even if it has less limits, it is enough for basic needs to many. Unlike the annoyances of skip limits, ads everywhere, or other usage limits, the free tear of GPT is relatively the same, again minus some robust tools and higher limits, but these are things not everyone needs. I'm very amused. Of course ads in free GPT are hard to do because nobody wants their conversation scanned for contextual ads. Google did end up doing this in GMail to an extent but were very careful and explicit. Then, Google announced in June 2017 that it would no longer use content from consumer Gmail to personalize ads, ending that feature. GPT? I don't know what they can do besides asking the people who use their service for more than just basic brainstorming to pay up.
It's already been suggested that they'll be adding ads. Maybe people will object, maybe they won't.
I suspect most people don't consider their chats with an LLM to be anywhere near as private/sensitive as email, even though in reality they could be even moreso.
@jscholes oh wow yeah, that would probably just use AdsKit or whatever framework the OS provides, but I wouldn't be surprised if they became an Ads partner directly rather than sourcing that to another company, at least it does sound like this would be the case. That might give them more control over which ads ultimately appear, and I wouldn't be surprised if at first they opted people out of targeted ads but later dipped their toes in once partners come online and it becomes ever too hard to resist chat info targeting the ads.
all these new generation copy-on-write filesystems, and yet no one has been brave enough to name theirs simply "cowfs". then they could call the package containing its mkfs, fsck, etc. "cowtools"
It's fortunate that REAPER also has an open enough plugin system that a third party could do the work to make it practically accessible with a screen reader via a plugin. Most conscious attempts to make such efficient software wouldn't have that additional feature.
oh, yeah, i didn't think about that. I knew they used their own toolkit for the GUI, which usually means poor accessibility. Do you know if there are any DAWs that have good accessibility from the vendor without a plugin?
I have no experience with accessibility for desktop software, as I've only ever worked on the problem for websites. I should read up on AccessKit, I guess.
President Trump’s latest cardiovascular MRI shows he’s in “excellent health” and there are “no acute or chronic concerns.” - White House Press Secretary Karoline Leavitt
i've been growing pre 1800's apples and i confirmed i'm not allergic to raw apples pre 1800. there's a genetic bottleneck for all modern apples around 1800 that bred out the phenolic acid, the lack of phenolic acid makes my tongue itchy and my throat swell up. but this orleans reinette i grew didn't cause any allergic reaction to me. i've grafted a bunch of other pre 1800s varietals and i'm converting my front lawn into an apple dominant food forest style garden with berries and perennial edible plants. i have twelve different apple seedling trees and i'll graft some into frankentrees with many apple varietals on one tree.
This is the full text of a book that I am offering for free. Barnes and Noble Kobo Amazon Google Books Apple Books It’s on Goodreads. Or you can download a PDF by clicking here.
Wer Charly Hübners Film "Element of Crime - Wenn es dunkel und kalt wird in Berlin" letztes Jahr im Kino verpasst hat, findet ihn jetzt in der Mediathek.
Der Film 'Element of Crime – Wenn es dunkel und kalt wird in Berlin' dokumentiert die einzigartige Geschichte von Element of Crime. Seit ihrer Gründung 1985 hat sich die Band zu einer festen Größe in der deutschen Musiklandschaft entwickelt.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
The 64-bit NVDA migration is obviously going to mean a lot of older speech synthesizers (such as SAPI4 voices) and braille displays relying on drivers that use 32-bit libraries will stop working. I was really happy to see the issue was being taken so seriously that there were/still are plans to make a compatibility layer, and this was considered a blocking issue, meaning NVDA 2026.1 would not be released without this being addressed. You can read that issue here: github.com/nvaccess/nvda/issue…
And then it was "addressed"… By removing any mention of SAPI 4 and other legacy drivers. There are still plans to return to this, and I sincerely hope this happens, as part of work on a different somewhat controversial topic - a new addon environment, with a more restricted API access which will be running in a separate process.
Now, anyone who’s already using Espeak or Microsoft voices with a modern braille display is probably going to just shrug, but this change is going to have some very serious implications that I suspect may cause some people to skip these newer NVDA releases. From my comment on the GitHub issue: Many people still rely on older speech synthesizers for various reasons. There are languages where the only good speech options are older legacy synthesizers, and many people who use specific voices due to a hearing impairment, because they are more intelligible. For these people using Espeak, OneCore or purchasing a commercial speech addon may simply not be an option.
The only way out will be for addon developers to come up with their own bridging solutions, assuming an addon for the synthesizer is even still being supported. Eloquence already has at least 1 project to do this, and I’d be really curious to hear from people who use it if performance is worse as a result, whether the supposed performance gains are lost due to using a bridged synthesizer. It’s possible to make a fast screen reader with bridged synths — Take ZDSR, which is a 64-bit app but with support for SAPI 4 and 32-bit voices, and it’s blazing fast, faster than NVDA. But no amount of bridged addons will help anyone using a SAPI 4 voice which currently doesn’t have an addon. I’m lucky enough that for Polish, an amazing community sprung up around RHVoice, an open-source synthesizer that now has a number of fast and intelligible voices, but other languages are not so lucky. (1/2)
Is your feature request related to a problem? Please describe. Yes, no 32-bit speech synthesizers work on 64-bit versions of NVDA. This includes all SAPI 4 synths Version used: nvda_snapshot_alpha-...
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
So, at the end of all this I’m personally unsure if the 64-bit transition was necessary or worth all the collateral damage - are there any libraries that NVDA depends on which are now 64-bit only that forced this? If you know I’d love to know as well! ☺️ But at the end of all this, I feel a decision like this should have been consulted with users, instead of just doing the upgrade because it’s the more modern thing. They could have ran a survey to see how many people would be disproportionally affected - there are plenty of international NVDA communities where it could have been circulated. At the end of all this I just hope we don’t lose too many developers, addons, and especially voices in less popular languages from this. For me, I’m probably going to hold off from upgrading for a while. It’ll be like the 2019.2 days pre speech refactor. Here’s hoping things will work out like they did then while also keeping NVDA as performant as it is. (2/2)
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
Python itself intends to drop 32 bit support in the foreseeable future. I think they should have offered two releases in 2026 and pulled the switch in 2027. But either way the transition is not optional. It has to happen in the next couple years. And I use eloquence64 as my daily driver. It’s fine for the most part. The main limitations are NVDA and windows ones. For example my bridge is unsigned so it can’t run on secure screens. And for the record it’s based on threshold. 99 percent of the code is human written, and the other one percent has been reviewed and debugged by multiple humans. I didn’t just shit out an AI conversion and call it a day.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
@fastfinge Good to know, I'll edit the post slightly, and I appreciate that you and other people put in the effort. Yeah I agree the transition should have been more graceful. Foobar did it this way for example.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
As far as I understand, they are using telemetry to decide. Onnx is 64-bit only, which powers the GPT-2–based image describer that I personally still feel should’ve been an add-on until a proper plugable model framework is developed, which is apparently in the works (though hopefully it happens before 2026.1 else the community will invent their own solutions that might break down the line). CC @NVAccess
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
Even the Sonata TTS voices will not work on 64 bit NVDA as I'm afrade. Nor @mush42 nor @ZBennoui had responded me for the addon or for the updated notebooks so I can continue training my voices. If that even breaks compatibility with older legacy TTS systems like Eloquence or even BestSpeech I don't know if I'll upgrade. Even if IBM TTS will stop working for that, I don't know
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
You'll laugh, but even now, this is one of the reasons I'm using NVDA remote. I dislike what happened to the IBMTTS add-on voice after the 2022 versions, so I just changed the manifest and use 2025.1 on the remote machine with the local machine doing the talking. If Eloquence/IBMTTS is not available in the 64 bit version, I'm happy to forego the 64 bit version. Eloquence is important, the features in that version are not. I realize this may become difficult over time but don't expect any issue for two or three years, at least.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
I'm not too worried about Eloquence, at least. It's used by many hackers, one of us will make it work. As far as making NVDA 64-bit, there are a lot of reasons to do this. It will fix loads of random issues like all tree view items not announcing their position properly in 64-bit apps like Paperback, as just one small example. Not to mention, Microsoft themselves aren't even giving us 32-bit images of their operating systems anymore.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
Happy for a pr on the eloquence64 repo if you can increase speed at all. What we really need is a DLL wrapper. Bridging in NVDA comes with a set of annoying issues.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
That's what I've already done: github.com/fastfinge/eloquence_64/ However, if we had an API compatible 64-bit DLL rapper, it would mean eloquence can be used in other 64-bit apps, and the dlls could be a drop-in replacement without updating the addons at all. As well, having a bridge means the "copy config" button breaks the addon, because executables aren't included to run on secure screens. As well, it's possible (if NVDA hard crashes) to get in a place where the eloquence bridge is already running, tries to run a second time but fails, and because of threadsafe issues causes the first bridge to also crash.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
The main place you will notice delays is in extremely long strings that can’t be split into words. For example a 400 character url. For various reasons streaming a string doesn’t start until a word boundary is found. So long urls can lag the synth.
NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
@fastfinge To be clear: I appreciate these efforts, despite wondering if there's a faster IPC mechanism that the community will eventually settle on. But the add-on does seem to work well in some light testing.
One note: the speech rate in this add-on and IBMTTS don't seem to quite match up. E.g. in IBMTTS, rate 81 seems closer to rate 86 in Eloquence64. @amir @serrebi @pitermach
RE: 🔒, NVDA 64-bit migration breaking older voices, very long (2 posts)
Sensitive content
The code is based on the Threshold addon, rather than IBM TTS, so that probably explains the difference. I didn't use IBM TTS because the need to support every eloquence version IBM or Eloquent Technologies ever released, thought about releasing, or had a dream about one time, massively overcomplicates the code, when 99 percent of people only want 6.1 anyway. Pull requests are welcome, but I've successfully scratched my own ich, as they say.
Schon beeindruckend, wie @inwx das Login so kaputt gemacht hat, dass 1Password nicht mehr funktioniert. Gebt mir doch einfach ein Formular auf einer Seite und nicht so ein komisches Fenster.
Und die Barrierefreiheit ist auch ganz enttäuschend, vor allem, da man ja ganz klar unter das Barrierefreiheitsstärkungsgesetz fällt. Ich verstehe einfach nicht wie man 2025 relauncht ohne sich darüber Gedanken zu machen.
Ja, sie haben es geschafft, eine ohnehin nicht wirklich schöne Seite noch schlimmer zu machen. Die Inkonsistenzen bei der Barrierefreiheit sind schlimmer als vorher. Vorher wusste man wenigstens, dass alles irgendwie Mist war, konnte das aber gut mit GreaseMonkey für den Eigenbedarf fixen. Jetzt ist es wirklich sehr inkonsequent umgesetzt.
Just saw "awareness of ubuntu" in a post on my feed, and I'm fairly sure that the writer was using ubuntu in its original meaning, rather than saying that Linux is a good thing, but I almost never see the word used in the original way.
Here is a pointless internal challenge to myself. For the rest of December, I will make no mention of holidays of any type on this feed, nor will I boost any such content. Here we go.
Remind me: what's the name of that island and special territory of Chile in the southeastern Pacific Ocean? There's a Kris Kristofferson album with the same title.
Urgh! Message at work: "Do you have time for a quick call?" With no further information what about or with what urgency. HATE it with a passion!! #ActuallyAutistic
The lovely Neil Miles has a 30-minute bumper batch of 1980s Christmas idents, announcements and trails on his YouTube channel. youtube.com/watch?v=84EZGb__7B…
So the Calendar UI is getting a lot of love right now to tackle issues like this and make it easier to use and more modern. We share plans and invite users to studies on our Design mailing list, if you'd like to check this out! thunderbird.topicbox.com/group…
Lately @mozilla @thunderbird has been a bit ass. It lets spam through to the desktop and mobile client, where other email clients (notably Evolution on Linux) don't. Even though I clearly mark something as spam, I still see it.
And this is even though Gmail itself blocks it and sends it to the spam folder.
I prefer the desktop version for email. But seeing spam just sucks.
Unfortunately, the Android app doesn't have a spam filter built in like the desktop. But we're sorry you're having issues on the desktop! We've got a blog post with some helpful advice on fine-tuning the desktop spam filter: blog.thunderbird.net/2024/09/t…
Lotus Diplomat: Smartphone with keyboard, 24GB RAM, 1.5TB storage and Snapdragon 8 Elite aims to fill gap left by BlackBerry notebookcheck.net/Lotus-Diplom…
The Lotus Diplomat is a new smartphone that comes with a keyboard and a secondary display. This handset should be well suited for typing longer text and is expected to be offered in several configurations.
@pixelate OOOH you know it's tempting. But not sure. It's croud-funded and no pricing yet. The larger keys though make me really hopeful. I'm stuck with my aging S22 because I haven't wanted a newer Galaxy phone and didn't enjoy that Galaxy experience as I remembered it in my S4 days, so I can't say I'm not on the hunt for an Android phone down the line. The S22 is still getting updates for a little while.
oh god did Qwen vibe code their email password reset because it thinks my email isn't valid so I can't recover whatever my password was (missing from my password manager...)
hahaha I had Claude comb through their 17MB of minified JS (yes it's that big on this page) to find it
● Excellent! I found it. The verifyEmail function uses a regex pattern to validate emails: verifyEmail=e=>/^[a-zA-Z0-9_.-]+@[a-zA-Z0-9-]+(\.[a-zA-Z0-9-]+)*\.[a-zA-Z0-9]{2,6}$/.test(e)
FreeBSD 15.0 (almost)-RELEASE, using pkgbase, on my Ryzen 9 MiniPC (and compared to openSUSE Tumbleweed):
- Full disk encryption works beautifully via GELI, as usual.
- Installing KDE is easy and it works perfectly on Wayland.
- All my main apps work. Others will run via the Linuxulator or Wine (Linux browsers, WinBox for MikroTik, etc).
- The fan seems more relaxed.
- The system generally feels snappier.
- Native ZFS. I can autosnapshot every 5 minutes. If I try to do this with btrfs - snapshots of the home directory included and quotas enabled - the system hangs while handling them (which is why Tumbleweed doesn’t snapshot home by default).
- The media keys on my keyboard work, but volume control uses huge steps and 30 percent is already extremely loud. This can be fixed. The monitor brightness setting is also a bit off, but I don't care.
- amdgpu works perfectly.
- The wifi card works. I haven’t tested the speed because I immediately installed the realtek-re-kmod driver to use the 2.5 Gbit ethernet connection.
- Suspend doesn’t work. This is a big problem for me. It’s probably more psychological than technical, but I can’t leave the computer powered for hours when I’m not using it. I already have servers running 24/7 here. I even considered putting my Qotom FreeBSD server in a VM. It would probably work, but next summer it might be an issue because temperatures here aren’t low and spinning disks don’t love heat (and I don’t love their noise).
- It’s stable and reliable. I’ve done almost everything and it just works, as expected.
- Some small glitches remain, mostly due to missing configuration or packages (I didn’t tune anything. I just installed it and started using it).
A much smoother experience than a year ago, when I bought it.
Will I keep using FreeBSD on this minipc? I’m not sure yet, since Tumbleweed works great and the lack of suspend really influences my choice. I'll contact Aymeric and try to offer some help to improve this.
For now, I’ll keep it on an external SSD and switch from time to time, especially when I know I’ll be using the minipc for hours.
ok you kinda skipped over the release stuff… youre using pkgbase I guess? You want to wait for the official release when the mirrors are released and then fix up your /etc/pkg/FreeBSD.conf and reinstall this bits. Deliberately left underspecified as I’m not sure what state your system is actually in. Come to irc or discord for more help @Tubsta @stefano
You may have earlier pkg-base configurations floating around. It is noted that even the RC and RELEASE have duplicates in places and this probably needs to be mentioned for a manual cleanup within the release notes.
You /etc/pkg/FreeBSD.conf should have (but not enabled):
@mason I don't think the OS is driving the fan. It's a consequence of the load, so it seems that FreeBSD is lighter on the resources. It might also be because of a more conservative throttling
Lower load under FreeBSD is still interesting. I thought Linux still had a serious advantage with fancier interrupt handling in its tickless kernel, timer coalescing, etc., but maybe I'm behind the times. Is it apples to apples in terms of what's running in userland under each of Linux and FreeBSD?
I don't know if that's ever been true. I've had the same large workload on identical servers that takes Linux to hundreds or even 1000+ load average but FreeBSD handles it just fine staying under 100
Also FreeBSD can still be ssh'd into when the load average is near 1000, but Linux... good luck. You'll be waiting tens of minutes just for the shell prompt if you can get past the sshd
@feld My impression is that it didn't show up in the ability to process things in bulk - all scheduler stuff - so much as the ability to not use power at idle; think laptop battery life, for example.
But it's been a while since I've spent significant time with FreeBSD in an environment where idle-time power consumption was measurable, let alone mattered.
@feld Oh, interesting. I'm curious why that's not a default! I wonder what the best way is to list driverless hardware.
Is this documented somewhere? The default is zero, and if "3" is magic there must be a range of options. Also, must it be in /boot/loader.conf? Seems like it could live in /etc/sysctl.conf, or is that too late?
Look, if Kubernetes insists on acting like a moody teenage kaiju who needs fifteen YAML scrolls and a blood sacrifice just to deploy a Hello World, then it deserves a nickname like kubey-boi.
He’s that problematic friend who: • promises to “self-heal” but actually means “restart forever until you cry,” • insists on a full 27-layer abstraction stack, • shows up to your house, rearranges all your furniture, and calls it “desired state,” • gaslights you with: “It works on my cluster.”
Wenn man in Thunderbird auf dem Desktop links unten bei Einstellungen auf "Auf Mobilgerät exportieren" klickt, dann zeigt es einen QR-Code an, den man mit der frisch installierten Thunderbird-App auf dem Smartphone scannen kann.
Und schon ist die App auf dem Smartphone eingerichtet. 😍
feld
in reply to Rebane • • •Rebane
in reply to feld • • •feld
in reply to Rebane • • •