Now up on the OSHWA site! Our third Healthware Highlight exploring the ways that MakeGood INC is working with 3D printing and accessible design with folks with disabilities. They can do incredible things with the support of community and being engaged with person first design work! Check it out here: oshwa.org/announcements/health…
Substack’s extremist ecosystem is flourishing. “The Nazi push alert was only possible because of Substack’s continued commitment to not only hosting but actively promoting authoritarian, Nazi-sympathizing, and other bigoted forms of extremism.” theracket.news/p/substack-s-ex…
I don't care what people who I otherwise respect like Om Malik and Dana Blankenhorn say about network effects blah blah blah . It is wrong to be on a platform that openly allows hate speech. If the Nazi Party's newspaper paid you $USD1000 / month in 1939 to write an op-ed monthly would you do it? No. om.co/2025/08/03/the-why-of-su… is wrong. sorry not sorry Om et al!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Not for the first time, the newsletter platform Substack is coming under criticism for hosting and now (accidentally) promoting Nazi content. Today, John Gruber of Daring Fireball is leading this c…
I really hate the framing of "content with which you fundamentally disagree." It's Nazis. It's not content, it's Nazis. I don't just disagree with it, it's Nazis. If we can't draw the line at fucking Nazis, I'm pretty sure there are no actual lessons history can teach us.
@poploser @roland also dislike Om’s analysis here because it sidesteps the most critical part. Peddling hate is lucrative — but it’s still peddling hate. Is evil only unacceptable when it’s unprofitable?
Love LibreOffice development? Want to turn your passion into a paid job? We are The Document Foundation (TDF), the non-profit entity behind LibreOffice.
😡 "A Canadian company is providing the muscle for a new Florida detention centre dubbed “Alligator Alcatraz” — & the Canadian govt isn’t ruling out working with the security giant in the future. …
Quebec-based GardaWorld, which has reportedly been awarded a US$8-million contract for work on the U.S. detention site, has also been awarded > $100 million in Canadian government contracts since Carney won the Liberal leadership in March …"
“This wishy-washy stance when it comes to jaw-dropping cruelty tied to the U.S. is yet another stroke in an increasingly clear portrait: one of PM Mark Carney with his elbows firmly planted to his sides when it comes to big business.”
“This wishy-washy stance when it comes to jaw-dropping cruelty tied to the U.S. is yet another stroke in an increasingly clear portrait: one of PM Mark Carney with his elbows firmly planted to his sides when it comes to big business.”
Hardly a day goes by without President Donald Trump boasting about the record tariff revenue the US government has been collecting since he ratcheted up taxes on almost every imported good.
We at Tuta offer an end-to-end encrypted calendar alternative to Google. If you're interested in finding out more do take a look at our website tuta.com or feel free to ask here. 🔒
ICYMI: earlier this week I wrote a blog post disclosing my tech stack… arguing that anyone who writes about technology should be transparent about the tools they use.
TL;DR: I’m doing well on the software front. Hardware? Not so much: 🍎
Let's normalize sharing the tech we use - and the platforms we purposefully avoid. I'm sharing my tech stack in this post to let you judge for yourselves if I'm "walking the talk" when I discuss resistance to Big Tech.
We had the privilege to chat with Matthias Baenz, a seasoned tax lawyer about why secure communication is non-negotiable in the legal field, and why he chose Tuta to protect his clients’ confidentiality.
Netušil jsem, že Svijany dělají nealko. Jednu dobu dělali slušné pivo, ale údajně to šlo poslední dobou do kopru. Navíc majitel je babišonek. denikn.cz/644208/majitel-pivov…
Pivovar Svijany hlásí kvůli pandemii ztráty v desítkách milionů korun, jeho majitel Tomáš Kučera ale věří, že v dlouhodobém horizontu je to jen „malinkatá etapa“.
To mi řekla výčepní, ale možná se spletla. Protože mi nejdřív natočila klasické pivo. Zjistím zítra co tam mají. On myslím ani Permon nedělá nealko pivo.
Now that my one joke post about how there are only seven posts on LinkedIn has comfortably done better numbers than everything else I'd ever put on that godforsaken cringe factory of a website combined, I can only conclude that what people really want from LinkedIn is a momentary respite from LinkedIn.
There are only seven Linkedin posts:
“My pre-literate child accidentally microdosed on an adversity. Did they react like a normal human infant? No! They said something suspiciously adult. Here’s what it taught me about b2b sales.
it’s not its primary use, but LinkedIn is also baffingly bad to get eyeballs on your writings. You get more likes on your LinkedIn posts than visits coming from there, it’s purely performative appreciation.
Over the past 12 months I’ve had 49.2K visits on my website, and 299 came from LinkedIn. Post anything with “stoked to,” “celebrating,” or “starting an adventure” and you’re showered with empty likes.
It's frankly absurd, and I think - and I want to rope @grimalkina into discussions like this basically 100% of the time - that there's something deeply, pathologically _wrong_ with working cultures where people feel the least bit compelled seek out this kind of meager, hollow, empty-calorie high-fructose-corn-syrup relationship-microdosing version of human connection on a regular basis.
What nightmare working culture has people self-medicating on a hit of parasocial emoji?
my uneducated guess is that people want to get some form of recognition they don’t get at work (or elsewhere?) and treat LinkedIn praise as transactional. If I like their post, surely they will like mine?
There’s also the fact that your boss is in the room. They will see who and what you like, and what you write.
LinkedIn has the same vibe as the watercooler in the company of The All Seeing Eye or Sauron.
The fascinating part of the dynamic is that (1) your boss is in the room and (2) your HR department, because recruiters are paying for access to the data, can see who at your company is updating their profiles and CVs, and when.
It's pretty incredible, and no wonder people both feel like it's required to engage with them and desperate for escape.
Trotz prekärer Finanzlage organisierte die Gnome Stiftung ein Entwicklerinnen- und Entwicklertreffen in Italien, um über die Zukunft des Projektes zu beraten.
I sometimes think about text-to-speech in comparison to visual text rendering. For the latter, there's a whole suite of different solutions, from segmented LCDs (en.wikipedia.org/wiki/Segmente…) to monospace bitmap fonts to proportional vector (e.g. TrueType) fonts. And somewhere in the current genAI boom, I'm sure someone is generating something that looks like handwriting. But in TTS, the industry largely jumped to generating human-sounding speech from human voices as soon as they could.
It seems to me that, from a technical point of view, good formant synthesis is the TTS analogue of modern proportional fonts. But I see why most of the world doesn't see it that way. Humanity had centuries of getting used to reading typeset text before computers came along. Listening to speech that's obviously synthetic has no such precedent.
What got me thinking about this is that doing text rendering on a microcontroller is commonplace. Maybe not using a full TrueType stack, but at least using proportional bitmap fonts. But to do TTS on a microcontroller, you have to use a technique that's broadly considered archaic.
The current preference for human-sounding TTS, except for blind power-users, is as if all text in a GUI were rendered using generative models that emulate handwriting. Maybe that would have happened in an alternate history where computers somehow came before typesetting.
as a sighted text to speech user, I find context dependent voices - be it deliberately synthetic or deliberately humanlike - roughly match my preferences for fonts. I want my code read cold and robotic, but comments comforting and soothing. It also makes it much easier to hear multiple voices at the same time.
@letoram Interesting. I guess another way of comparing TTS voices to fonts is that robotic voices are like monospace fonts while human-like voices are like proportional fonts.
I'll be honest, I didn't think I could get Hacktivate working on iPhone, but… 😎 It's really packed with fun cybersecurity challenges, and it's both enjoyable to play *and* enjoyable to build. I can't wait to share it with you all!
@nuno_nuno I'll be adding accessibility support for VoiceOver, Switch Control, Voice Control, Reduce Motion, and more, but – and this is a massive but – it will never be a great VoiceOver experience. Some tasks are things like "look through 500 lines of binary gibberish to find GPS coordinates at the end," and I don't think that can be exposed to VoiceOver in any meaningful way.
Well, I worked with helping to make experiences accessible. I can't say for now because I haven't hehe seen the game yet but when I do, I hope we can discuss it further.
Random pop song that just popped in my head: "Try" by Pink. youtube.com/watch?v=yTCDVfMz15… Not anything particularly special, just that a random phrase made me think of it, and I think it's a good song.
Official Video for "Try" by P!nkListen to P!nk: https://Pink.lnk.to/listenYDSubscribe to the official P!nk YouTube channel: https://Pink.lnk.to/subscribeYDWa...
400 years to Proxima Centauri means velocity of roughly 1% of c. To accelerate to that, then decelerate, takes energy on the order of 1-2% of the ship's rest mass.
Ship is 36 miles (50km) long.
So it obviously masses many gigatonnes (1 cubic kilometer of water—0.15 the density of steel—masses 1 billion tonnes). Conservatively this masses 10-100gt.
Fusion bomb yield is about 2% of the payload mass. Our current 10,000 H-bombs probably contain 1000kg of fuel.
@nyrath Upshot: I think this thing will take 100-1000 years of our current worldwide civilizational energy budget to propel.
And at the other end? Congratulations: a colony of 2400 people is at least 3 orders of magnitude too small to sustain a self-training technology base able to service an autonomous space colony. (Because resupply with finished products is impossible at that range.)
TLDR: magic wands or scientific breakthroughs required.
This is what 5 years of reforestation looks like ✨
Once cleared for a palm oil monoculture, this forest is now lush and diverse again.
Together with Leuser Conservation Forum, we’re restoring the Leuser ecosystem in Indonesia — supporting natural regeneration and planting native trees with local communities.
5 years of progress, powered by the Ecosia community 💚
Love LibreOffice development? Want to turn your passion into a paid job? We are The Document Foundation (TDF), the non-profit entity behind LibreOffice.
Intéressant : réadapter l'interface utilisateur avec les concepts Apple… Ça pourrait effectivement améliorer l'ergonomie, parce que calquer sur MS Office est un non sens ergonomique global (c'est fou que MS Office soit si populaire alors que son interface est une torture) (bon d'accord : c'est surtout parce que MS Office est imposé aux utilisateurs et que les pauvres n'ont jamais connu autre chose que cet instrument de torture !)
And in the end it does not even mention the weird behavior: it stores errors in an internal queue which mysteriously makes the *next* invoked function fail...
OpenSSL is a 5 million USD revenue/year company with more than ten full-time employees.
Makers of some the Open Source world's worst documentation. For an API that is enormous, quite weird and full of hidden magic that is not what you would expect.
Disclosure: I'm the OpenSSL Foundation community manager and I was part of a (failed) effort to get a Documentation feature added to Stack Overflow. (See my writing on the topic on my blog: jlericson.com/tag/documentatio…)
I agree the OpenSSL documentation is lacking. (Which is the nice way of saying it's inconsistent, fails to provide necessary information needed to use the software and disorganized.) The problem is multifaceted, but the shear size of the API is, indeed, at the root. Quite a lot of the library isn't documented at all:
$ perl ./util/find-doc-nits -u <br># 1243 libcrypto names are not documented<br># 30 libssl names are not documented<br># 45 macros undocumented (count is approximate)<br>
The current policy (enforced by CI) is to not allow new undocumented functions, but that's just preventing the coverage problem from getting worse. It doesn't address the problems you highlighted with d2i_X509. (Though I should mention you linked to the 3.3 version and there has been some work done on that page that you can read if you switch to the latest documentation: docs.openssl.org/master/man3/d…)
Writing code and writing documentation are separate skills that aren't always found in the same person. As far as I know, there haven't been contributions from technical writers. Personally I'm an advocate for bringing on (at least) one person dedicated to documentation.
@jericson I say those are just lame excuses. OpenSSL has soon existed for *thirty* years and there's clearly still no serious attempt to even try to do good documentation. Lots of the function calls even remain the same so there has been some time to do something.
I honestly cannot understand how a thriving project can claim to be for users and at the same time ignore their docs to this degree.
They can only be lame excuses if they were excuses. I agree with you that documentation needs help. It's something that came up time and again in our survey in June. And, of course, I've consulted the documentation myself to try to understand the library. It's a problem.
You know the history of OpenSSL better than I do. Nothing anyone can do about the past. The question is how do we fix the situation? My answer is that we probably need a technical writer who knows how to structure useful documentation.
In my experience, documentation doesn't come naturally. Everyone wants it, but few people want to write it. (You can kinda see that in the OpenSSL forks which don't seem to be much interested in improving the documentation either.) You are absolutely correct that user-oriented projects make documentation a priority.
@jericson a modern engineer writes code, test cases and documentation. There are no "technical writers" in most of those projects out there with decent documentation. It is enough if engineers just add to the documentation every time a user asks about the functions for things not mentioned there. And undocumented functions would in most projects simply not be accepted. Stop what you do, add a few lines of docs for each one of them.
To go back to your initial post, is this issue related? "Horrible d2i_TYPE*() reuse behavior leads to crashes on error" sounds pretty similar to what you are describing.
(I disagree with the idea that undocumented functions are worse than functions with "a few lines of docs", by the way. Better to have an obvious problem than paper over it.)
d2i_X509_ALGOR() etc. behave badly as explained in doc/man3/d2i_X509.pod: In some versions of OpenSSL the "reuse" behaviour of d2i_TYPE() when *a is valid is broken and some parts of the reused str...
@jericson Not sure I understand your position re documentation. *Some* documentation is better than none, for publicly declared functions. It'd better be correct in what it says, though.
@jericson Totally with ya there... and like I said, lots to catch up on. Could we say there's quite a bit of doc debt, that has accrued over the decades?
@jericson That being said, it's a bit interesting to see how the discussion has glided from the original complaint about *existing* documentation with a gigantic of names 😂
@jericson I do note that you're talking about the next function being invoked, that's affected by the error queue (???). I'd love to know which one, 'cause that sounds real weird. The issue that @jericson talks about is something different, has nothing to do with the error queue per se.
oh maybe we should not rewrite curl but write a new openssl instead ss my next hobby prroject. It can not be that difficult ...<irony off>. I use two simple functions of open ssl and now they are deprecated. How to replace it? no idea. Documentation doesnt give a clue. And yes its huge
The #OpenBSD devs spend a lot of time improving the man pages. LibreSSL is not OpenSSL but I look up their docs a lot. Check man.openbsd.org and Ingo Schwarze's presentations at openbsd.org/events.html
In fairness, I'd like to point that the amount of developers you mention is a very recent development. There is indeed quite a bit of catching up to do.
I tend to encourage people interested in TLS/SSL coding to watch Bob Beck's "First 30 days of LibreSSL" to get some orientation - youtube.com/watch?v=GnBbhXBDmw…
LibreSSL is a recently started initiative by the OpenBSD project to fork and clean up the OpenSSL code base. They have been working on this rather intensivel...
lol. I have used that one. It does suck. OpenSSL has a lot of API level documentation some better than others. I am never quite convinced I am using it correctly even after reading quite a bit of it. The examples are often either too simplistic or not quite what I need. Of course I don't think I have sent them a PR to fix any either...
@andrew whatever the reason, the error queue behavior is not mentioned in the man page. This thing made a bus load of people stare at code for a long time and try all sorts of wrong things until this detail was discovered.
> when you read a block of text, you are not consciously moving your eyes from word to word like you’re dragging a mouse cursor, repositioning continuously.
When I read visually, I have to get my head close enough to the screen or page that I do in fact move it continuously back and forth. Or if I were to use a screen magnifier at a zoom level high enough that I could lean back, I would in fact be literally moving the mouse back and forth, or perhaps doing equivalent keyboard commands.
To be clear, not saying Glyph should have added a caveat or shouldn't have written that, just sharing.
I spent hours and hours reading print textbooks in school. Sometimes the textbooks themselves were reprinted in large print; sometimes I used a video magnifier, a magnifying camera hooked up to a monitor. At home I had a unit where I would put the book or paper on a tray under the camera and move the tray back and forth.
I do wish now that I had spent more time learning to read Braille going clear back to elementary school, rather than reading print textbooks. But I see the logic in not spending extra resources to ship Braille books, and get some things transcribed on-demand (teachers' handouts and such), when I *could* read print. At least I started getting more books on tape (via the National Library Service talking book program and Recording for the Blind) sometime in middle or high school.
As for computers, the earliest computers I used (Apple IIe at school, Apple IIGS at home) had no way of running a screen magnifier as a generic OS component usable across a broad range of software. (The IIGS later did get a screen magnifier in GS/OS, but only for proper Toolbox GUI apps, and those weren't nearly all I used even in the early 90s.) And like I said, with a retrofitted screen magnifier, you _are_ moving the mouse back and forth. Might have still been better for my health.
In the public schools I attended through 7th grade, I had access to Apple IIe computers with an Echo card and a primitive screen reader that only worked for text-mode programs that used ROM routines for screen output and keyboard input, so, mostly BASIC programs or programs that were specially designed to work with the screen reader. The Apple IIGS at home never had the Echo card. I don't know why I never straight out asked my parents, "please buy me an Echo card."
In 7th grade when the school got an MS-DOS PC with then-current assistive technologies, the teacher tried to teach me to use the ZoomText screen magnification program, but I preferred to play around with the screen reader. When I got a laptop in 8th grade, I did use screen magnification when using it as a laptop. But when I later connected it to a monitor at my desk, old habits prevailed and I used the screen unmagnified. The latter did become practically necessary as I ventured into Linux.
These days I use a screen reader pretty much all the time for non-coding tasks, and often when reading code I use a combination of visual reading and a screen reader. So far I haven't made the effort to retrain myself to exclusively use a screen reader when coding.
Getting the laptop in 8th grade is a story of its own. That was my first year at the private (fundamentalist Christian) school that I attended through high school. As I recall, the principal learned about the difficulty I had handwriting notes and assignments, and decided to do an all-school fundraising program to buy me a laptop and screen magnification software. If I remember correctly, there were a few large donors that probably made it work more than the little donations from kids.
Perhaps I thought that since the Apple IIGS was capable of synthesizing speech in real time on stock hardware (although, as has more recently been pointed out to me, not with near as much responsiveness as the Echo and Textalker allowed), I shouldn't have to have the special hardware. And like I said, Textalker only worked with a limited selection of programs anyway, excluding many that I wanted to use.
@jaybird110127 You haven't heard the First Byte engine on the GS, in SmoothTalker or Kid Talk? Bryan Smart was the one that said, a year or two ago, that it was way less responsive than the Echo.
@jaybird110127 Let me see if I can record a quick demo. Will probably demo Kid Talk, since that's the disk image I happen to have, and I got that as a Christmas present when I was 8. Not a screen reader, just a talking word processor for kids.
@jaybird110127 And I don't think it self-voiced the whole UI; it could just read out loud what you wrote. It wasn't for blind kids specifically; it was a mainstream product.
Got a reply from someone who never heard speech synthesis on an Apple IIGS using stock hardware. Specifically, this was the First Byte engine, as shipped in the Kid Talk and SmoothTalker products. Kid Talk was a sort of talking word processor for kids. Not blind kids; the UI wasn't fully self-voicing. Nevertheless, my parents gave it to me for Christmas. For me it was nothing more than a toy. Demo coming...
Here's a demo of Kid Talk running on an emulated Apple IIGS (using MAME). I booted it from an emulated hard drive for speed, and because MAME's emulated 3.5" drive sounds aren't that authentic anyway. Note that the speech is only in the right channel. Not sure if that's a MAME bug, or just that Kid Talk wasn't designed to be used witha stereo card. And sighted viewers get to see me struggle to target the "No" button with the mouse.
The First Byte engine was also available as what we'd now call an SDK, a product called Talking Tools. I got that for Christmas at age 13 in 1993 (I stubbornly continued to play with the IIGS through 1994, though my PC and online activity was also ramping up at the same time). In 1994 I attempted to write a screen reader using Talking Tools, in assembly language, for the text-mode environment used by some 16-bit IIGS programs including the development environment I was using. I failed, hard.
I tried to patch the Text Tools function table (similar to a vtable today) so I could intercept the main text output function call to speak text as it was written to the screen. I wrote my first attempt, loaded it onto a bootable disk as a GS/OS system extension or whatever they called it, booted it, and... it spoke, but it read some random text and then crashed. I didn't know how to debug it and just gave up. Well, it was a side project, and I had other things I had to do, school and such.
And no, I don't have that code. My parents threw out all of our Apple II disks long ago, including my proudest programming accomplishment on that platform (a program that, ironically, wouldn't have been accessible to totally blind kids).
Well actually, I guess it would have been accessible with Textalker GS, just not using 8-bit Textalker like what we had at school. It was text mode; it just didn't use the ROM routines for keyboard input and screen output, and it was a ProDOS system program, not something you could launch from BASIC. The program itself was a simple spelling drill program that my aunt (a teacher) asked me to write. There were surely a hundred other better programs out there, but I wasn't about to say no.
The Apple II computers had both 40-column and 80-column text modes, and there were a few programs that would let you choose which one you wanted, presumably for backward compatibility. I remember one time I was working with my uncle Eric (a significant mentor when I was learning to program as a child), and he automatically chose 80-column mode for one such program, then remembered that I preferred 40-column mode.
And then there's the fact that I had to handwrite assignments, moving my head back and forth across the paper as I wrote. Again, I see the logic, being able to use the same worksheets as sighted peers, not having to have stuff transcribed (if I had written in Braille). Starting in third grade, my mom convinced my teachers to let me use the home computer for homework when practical. But I still had to do a substantial number of assignments in handwriting.
@MikeGorse My family bought our first computer in 1988. By that point, the IIGS had been out for almost two years. Plus my uncle was a big fan of all things Apple II, and he had a IIGS himself, so he probably advised my parents on what to get.
In 8th grade, when I started attending a private school with no assistive technology resources, I got a portable video magnifier where you put the camera on top of the page and hooked it up to a TV. So the school set up a TV on a desk in a handful of classrooms. I continued to use the non-portable unit with the tray and its own monitor at home.
"…but the children!" they scream, and claim they want to protect them from e.g. Transgender folks. Now, someone run stats on a full year of child abuse cases. Interesting that those who scream have a pretty high number of cases – while those they want to protect them from, seem to have none at all.
Fun fact: until today, my blog did not feature even *one* occurrence of the word "saccade". I have now corrected this oversight. blog.glyph.im/2025/08/the-best…
> when you read a block of text, you are not consciously moving your eyes from word to word like you’re dragging a mouse cursor, repositioning continuously.
When I read visually, I have to get my head close enough to the screen or page that I do in fact move it continuously back and forth. Or if I were to use a screen magnifier at a zoom level high enough that I could lean back, I would in fact be literally moving the mouse back and forth, or perhaps doing equivalent keyboard commands.
The curl project welcomes its newest sibling into the family: wcurl. I already wrote about wcurl. I will try to not repeat myself too much here, but starting now wcurl has its new home under the curl organization umbrella.
I often hear or see people claim that HTTP is a simple protocol. Primarily of course from people without much experience or familiarity with actual implementations.
@hyc I think that's taking it too far. As I wrote: other protocols have had similar development tangents, plus HTTP is unique so we can't really compare with something else to see what fared better.
It was designed to be stateless and handle one-shot requests, but was mated to HTML, which may require multiple requests to retrieve the complete contents of a page. That right there was a step 1 design fail.
It lacks message IDs, which are common best practice and are used by every protocol underneath it, for many good reasons. That glaring lack, and the lack of well defined message lengths, has hindered all subsequent progress. Another step 1 design fail.
@hyc and yet HTTP has conquered the world and while you say it "hindered all subsequent progress" HTTP/2 and HTTP/3 have sprung out from version one and they too are used more than most other protocols on the web.
you already said that HTTP/2 and /3 are more complex because they must include /1 so yes, it has hindered progress. The fact that /2 and /3 even need to exist shows that /1 was inadequate. And being used most doesn't mean it's best. Look at how much network bandwidth is wasted sending the same headers over and over on related requests. The percentage of network traffic that's pure overhead is insane, especially if accounted over all the years HTTP has been in use.
@hyc In the last 30 years I've implemented several HTTP servers and HTTP clients for various purpose and I always considered it a simple protocol.
If you want to do something fancy, client and server can agree to do it. But for my rather simple applications, code just didn't need to care for these special cases and was simple.
Compared to that: Try to implement SMB, CIFS or ADS...
pinging you @DavidGoldfield as this sounds like something your community might like to hear about: "Are you passionate about accessibility in tech? The Open Source email client Thunderbird is starting a new Accessibility Standards & Compliance Committee, and we’re inviting one community member to be part of it. We’re looking for someone who cares about inclusive design, development, or content and wants to make a real impact. You’d help review features, share ideas, and make sure accessibility stays front and center in how we build Thunderbird. If that sounds like you, send an email to community@thunderbird.net with "Design Committee" in the subject line to request more information. We’d love to hear from you! Let’s make email better for everyone." www.reddit.com/r/accessibility/comments/1mkdt8q/want_to_help_make_thunderbird_more_accessible/
. It is wrong to be on a platform that openly allows hate speech. If the Nazi Party's newspaper paid you $USD1000 / month in 1939 to write an op-ed monthly would you do it? No. 
I agree with you that documentation needs help. It's something that came up time and again in 
Hubert Figuière
in reply to Hubert Figuière • • •