I'm with you at least 2/3 of the way. My preferred stack is PRQL + DuckDB + Dagster. I evaluated the space for work at my current company (was originally only DE, handling ingests from ~300 sources across various systems, on order of ~1k downstream tables in dbt + hundreds of dashboards + a handful of business-critical data+app features; now leading a small team).
I came away ranking dagster first, prefect second, everything else not close. IMO dagster wins fundamentally for data engineers bc it picks the right core abstraction (software defined assets) and builds everything else around that. Prefect for me is best for general non-data-specfic orchestration as a nearly transparent layer around existing scripts.
I very much agree with English being a hodgepodge of confusing rules, especially those tense-related.
Re rewtriting (3) though: does not the following suffice?
"She must have been being watched".
I can imagine an argument that "She would have had to have" is not equivalent to "She must have" on grounds that the "have had" might suggest additionally that the past-ongoing-watching no longer occurs. I don't really think this flies, though, since the natural reading of "been being watched" already suggests to me that the watching was relegated to the past ("been" being interpreted as only connoting the past, not the "inclusive-or"-type interpretation "previously and perhaps presently"; much as "or" itself is - to my chagrin! - generally interpreted exclusively in standard parlance).
I couldn't think of how to simplify the "been being" though; that is a tough nut.
About this: I have a relation who works for a relatively large upstream auto part supplier. I asked him somewhat jokingly about this fight club scene, and he immediately and unabashedly told me he'd been involved in a number of such conversations. In retrospect, I can't understand why I was at all surprised: how else would the conversations go in a corporation (whose sole or primary incentive is by default monetary)?
(To be clear, I'm not saying that I find this morally correct — I'm not sure how I feel about that aspect, honestly, except icky at the surface level. I more means that it seems retrospectively to me that, well, of course that's how it would go, given the incentive structure)
> how else would the conversations go in a corporation (whose sole or primary incentive is by default monetary)
This isn't really a corporate issue at all though. Given scarce resources (whether physical or human), we need to be able to allocate resources efficiently. A conversation along these lines happens in public health systems all the time: how much money should be spent on medical interventions? There, the concept of a QALY (Quality-adjusted life year) is used, and typically, a price limit is set per QALY. Then, only interventions below that threshold will be funded. The idea is that since the healthcare system has limited funds, it doesn't make sense to spend exorbitant amounts delivering marginal results for one patient.
Now, one could argue this is simply a monetary issue, and if we didn't use money to measure these things, the issue would go away. The thing is, even if money isn't an issue (somehow), scarcity is still something we need to deal with. Developing and administrating medical interventions takes human labour, and spending a disproportionate amount of person hours on small gains is still an issue.
This comment is probably a little rambling, but the TL;DR is that given scarce resources (whether that's money in a corporation, or chemists and doctors in a health system), doing calculations on human lives is necessary if we want to make sure we allocate resources effectively.
I can't speak to "best", but I can say that I've been very happy with BazQux Reader. It has good keyboard shortcuts, is very fast, lets you make filters within your subscriptions, has a good in-reader article view, a good mobile site (no app, which I like), and also allows to subscribe to see social media sites if that's your thing. I had previously used Feedly and much prefer BQR
I apologize if this is too unsubstantive of a comment.
I primarily want to say how much I like the parent comment's writing. It is very evocative — "the tea sat tangibly in my stomach", eg, is the sort of description I'd expect in a well-written novel. Kudos nimbius.
Separately, I fully agree w/ the above point about a 2:100 ratio being well too much. The tea shop from which I buy generally recommends ~2g loose for a 175ml cup, but I still find this to be too strong at anything except a minimal brewing time. I tend to use about 1g/cup, ie a bit over 0.5% by weight.
Logic is my thing (although regrettably not what my supervisor has me doing; that's another can of worms).
For me personally, the above is exactly correct. Logic is wonderful for me in large part because it's so far removed from tactile reality, though so ingrained in everything.
PDEs/analysis are way too applied for my taste; they as such don't give me the same escape into a world of pure thought (although of course they do provide this moreso than applied ML, or -- god forbid, for me -- web development would). Logic is seductive to me this way.
I recently was gifted -- very, very generously! -- a Moonlander Mark I split, ortholinear keyboard [0] (made by the same company that previously made the Ergodox). It's genuinely fantastic, and I can't ever see myself going back to either a non-split or a non-ortholinear board as my desktop's input. It's a huge improvement ergonomically -- (0) no hunched shoulders, with the halves at shoulder width (1) the ortholinear layout results in much more natural-feeling finger motions and (2) the ability to "tent" the individual halves is also wonderful, for not having to twist your wrist out of its natural somewhat-diagonal position. Beyond that, the ability to easily modify up to 32 layers of functionality [1], with macros, really is astonishingly useful once set up to your personal workflow. If I were to lose it, I'd order a replacement immediately. Highly recommend, for anyone who can justify the board's price -- I'm normally fairly frugal, so the fact that I consider a $400 keyboard anything except an egregious waste of money really speaks to how much I like this device.
I couldn’t agree more. I’ve been using a Kinesis Advantage and a Mistel Barocco (when traveling) for years. I recently switched to the Moonlander and I love it.
Coming from the Kinesis Advantage the adaptation period is short, and since the moonlander is so easy to program I didn’t really have to adjust my habits. If anything, I’m trying to learn that I can use its programmability to my advantage to make my life easier.
I’m also very happy that it’s hot swappable, so I can start experimenting with different switches.
Why would you switch to the moonlander from the Kinesis? The curved design of the kinesis is way better for reducing strain than the flat style of the moonlander/ergodox. If the only factor is programmability, then I think that factor may be overexaggerated in how important it is. Know that anyone who wants to riposte that last point should realize that I'm not a novice to making my own keyboards or programming them. While layers seem like a good idea, it's very difficult to actually come up with enough usecases to find them more useful than just the ability to remap keys and record macros, both of which the Kinesis Advantage 2 can already do.
I also find having two separate pieces a bit annoying since I'll tend to want to fidget with them instead.
> Why would you switch to the moonlander from the Kinesis?
QMK. I think QMK is a bit like Excel in that everybody has that subset of functionality they use. I like layers and the leader key.
Open source firmware. I have a last gen Advantage I can't use because there's a bug in the firmware that causes modifier keys to stick and there's nobody left at Kinesis that remembers how to work on it. That will never happen with QMK.
Hot swappable switches.
Edit: I do miss the curve though. I'm using SA Profile keycaps [0] and while they're nice it's not quite the same.
My https://github.com/kinx-project/kint project is a replacement keyboard controller for the Kinesis Advantage keyboards, which allows you to use QMK :)
Yeah this is what I was going to say. There are a few projects which allow one to swap out the firmware for a custom one on Advantage Kinesis keyboards. There's one with a teensy, one with a custom controller. It's not exactly trivial to do, but I think it is the ultimate if you want to go that route. The hardware design of the Kinesis is top tier, but the software is a bit weak.
I gave up on three Kinesis Advantage keyboards in 2006, after they all started dropping key-up events on modifier keys. I then took one apart. Primitive, dated USB hardware, and nothing about the company read that they were eager to modernize. (The ergo keyboard market abounds with never-updated designs.)
Has there been a generational improvement in their circuit quality?
The original ones had some problems that were never fixed (IIRC, the original firmware was outsourced to a company that went under), but the Advantage 2 was a redesign and works great (at least for me).
I had the same bug and it irritated me to no end too. I was about to make my own custom controller when I found out that they made a version 2 which fixed the issues. I have 2 of the version 2 and have been using it for a few years now with no problems. I'd say it was improved.
If you email them, they might give you a discount or some kind of trade-in if you tell them about your previous problems. I've found their customer service to generally be pretty chill.
Is that what's happening? My kinesis advantage (plugged into a Mac) sometimes gets stuck modifier keys and only unplugging it and plugging it back in fixes it. Seems worse than just dropping some keyup events, because it doesn't get unstuck if I hit the modifier key again.
Not Advantage, but my Kinesys Freestyle Edge has great firmware and customize app (Swing) that works flawlessly. The firmware is better in some features than QMK, but worse in some features. (I want LT key on QMK for Kinesis)
The kinesis is great in my opinion, but the two sides are not far enough apart. A little is better than nothing, but you still have to position your hands forward, stretched out in front of you.
I have the same board and I am generally happy with it.
There is one warning I’d like to make though; I have fairly small hands and I feel that my hands have to reach more using this keyboard than with other split keyboards I’ve used.
It’s purely anecdotal though, because I have done some measuring and cannot actually find that there is any significant difference in distance between keys compared to my old keyboard. So, there must be something else at the root of that feeling.
It is, however, something I’ve heard from others as well.
In general though, it is a great keyboard. It is expensive, compared to off the shelf Logitech and Microsoft boards, but seen as a the main tool I use for getting my job done, it’s no cost to pay.
> I have fairly small hands and I feel that my hands have to reach more using this keyboard than with other split keyboards I’ve used
The Moonlander has a lot of key rows for a "flat" keyboard so there's going to be some reaching if you use them all. I have relatively large/long hands and coming from a Kinesis Advantage (which has a similar number of rows) it was a lot of reaching for the top and bottom rows, especially the corner keys. I ended up putting a lot more keycodes on layers to get around it.
I still get QMK and the true split design (which helps with my shoulders) but I also get the curved keywells back so I can hit everything on the keyboard pretty easily. I did lose that bottom row but I just pushed the useful keys to the thumb cluster.
imo a large keyboard is good to start with as most people will be used to having a large standard keyboard. what i've done is, using qmk, slowly updated my keyboard to make less and less use of keys that i wouldn't have on a 3x5 dactyl. it allows me to make my changes slowly but still develop the habits to get away from the number row and what not all while finding what works well for me
I have smaller hands as well and have adapted well to the Moonlander. I have found that I prefer to keep the thumb cluster tilted upwards to reduce the distance to my space/return keys.
Unfortunately this means I can't use the tenting since that requires the thumb cluster to tilt down and act as one of the legs. That's my only real gripe so far.
I have smaller hands as well and I do tent it slightly but then I only use the first "piano" key regularly [1]. The others on the thumb cluster are for layer keys.
My only gripe is that you can't negative tilt it, which I prefer when standing. I do love my Moonlander though.
I have looked at that keyboard a lot and am really tempted to splurge, but ..
What is putting me of is the missing dedicated cursor keys at the bottom right. I use a layout on my current keyboard that has cursor keys on e,s,d,f with a modifier and use that a lot, but not exclusively..
I have the same issue with the ultimate hacking keyboard and others.
Has that bothered you at all, or did you find a layout that works for you and keeps the cursor keys?
You can have the cursor keys where you want on UHK, see the webdemo of the UHK Agent software: http://ultimatehackingkeyboard.github.io/agent/#/keymap/QWR
and if you buy the modules you can also have additional actions, without taking your hand off the keyboard, because you do not even need a mouse.
The Moonlander (and most of the ones being discussed) is fully programmable, so you can move the keys around as you wish.
I have ← ↑ ↓ → in a line on the four lower keys on the right of my ErgoDash. Not having the ⊥ arrangement took a little time to get used to.
I considered some fancy QMK (keyboard firmware) hacking to keep the inverted T, and have the key above ↓ change between Z and ↑ depending on the keypress before, but decided it wouldn't be practical. However, the Z is for Dvorak. A Qwerty layout, where it would be /, might cause less trouble.
I've looked at a few Moonlander setups for the keyboard layout that I use on their site [1][2][3] and to my mind the right half is simply missing one column of keys to get a proper cursor key block.
Even the lying-L option ─┘ would mean missing the right shift key (which of course should probably be on the thumb block) and getting used to something different.
I'm not saying it wouldn't work, but for me being from Germany the keyboard is even more expensive than it is anyway and I'm not sure I'm willing to pay that much for the hope that I can adjust to it.
Those layouts have a lot of layers; I didn't have the patience to learn something that complicated.
This is my ErgoDash layout. I have one less bottom-row key than the Moonlander, so I have combined PgDn/LeftAlt and ←/RightAlt onto single keys. I would uncombine these if I had the extra key, as I occasionally press Alt when I intend to press ←, but am too slow.
I also have Enter/Ctrl as a single key, but it's very rare for this to do the wrong thing.
I mostly learned the layout to reduce stress on my hands due to the more ergonomic placement of each letter.
From all the other layers I only use the cursor keys and L3 for []{}<>() and stuff like that, which are much easier to reach and I use a lot when programming.
No, you can have a normal ⊥ arrows layout on the right half [1]. I do as I use those keys a lot. Or I can use a navigation layer as others have mentioned.
The magic is the double use of modifiers if tapped or held.
Whilst I personally could see a need for another column if you have a Nordic layout like mine, but that would make it not symmetrical and unnecessary for English layouts.
For a querty or qwerz layout that would indeed work, if you moved one symbol key (- for german layout, ? for american).
However in my keyboard loayout there actually is a letter in that place (j) [1]. And while I'm willing to compromise on modifiers, page up/down and the like, I draw a line at the stuff that I spent a lot of effort on learning to touch-type :)
As for using a navigation layer - perfectly happy with that 90% of the time, I already use that today, but for some things I just prefer actual cursor keys.
For example selecting an entire word in an editor, you press shift to select, ctrl to jump a whole word and cursor for the direction to jump in. If you add a modifier to that, you start breaking fingers ..
Ah yes, for that layout with 4 extra keys you would have to compromise. E.g. move the ',' or '.'to be shared with the right shift key so that you can have the arrows keys in normal layout. Or just have the left and right on the base layer and all 4 in another layer.
This really seems like a kinesis advantage but split which is cool. I own an advantage 2 and it has really helped with RSI. Personally I prefer a single unit so it's ideal for me!
It makes such a difference to have everything comfortably laid out at shoulder width.
But what if you need to use a mouse? Then what I've found best ergonomically is to use a "rollermouse". But I don't feel that works well with a split keyboard, and then I prioritize the mouse. (For me my elbow is the biggest issue)
Personally I think a mouse is not replaceable by anything else that even gives remotely the same productivity.
What can be improved is how often you have to reach for your mouse and how much your hands have to move on the keyboard.
I have personally switched to a tiling window manager that allows me to do much with the keyboard, added the surfingkeys plugin to chrome to allow the same when browsing and switched to using the neo2 keyboard layout (this is for Germans, Dvorak or something similar might be better for you).
The combination of these factors has massively improved how much I need to reach for the mouse throughout the day.
I agree with this. I wrote a script that will focus my most commonly used windows with keyboard shortcuts. Super+F for Firefox, Super+T for terminal, Super+S for Spotify, and so on. Between this, vim, Surfing Keys for Firefox, and just an interest in learning keyboard shortcuts I have mostly eliminated the mouse. I still use it when it's the right tool for the job, but I can get by with far less mousing these days.
I found what caused my RSI was going from keyboard to mouse over and over. Getting a keyboard with a trackpoint so I could keep my hands static fixed it. I would eventually like to get a split keyboard with the thumb trackpoint.
I have tried a bunch of things for this issue. I tried placing a trackpad between the two halves, I tried it on either side as well. It didn't really work for me.
What I landed on is a) mapping the mouse to an extra layer that I use when I only need minimal movement, b) using VIM extensions for as many programs as possible, including my browser and c) just using a normal mouse otherwise.
The Moonlander default layout has a layer that emulates mouse controls. It's not great for long sequences but can help with some small tasks where you prefer not to lift your hands from your keyboard.
For those of you with one at home: hold ';' and then control the mouse with E,D,S,F and left and right arrows left and right click respectively.
I overkill. Mostly I have a trackball in the middle as well as an ergonomic Penguin mouse on the side. I rotate and swap between what is on the left, right and middle. A trackball doesn't need a lot of space.
My understanding is that TempleOS was/is regarded as a tremendous one-person achievement (although of course its author had some quite nonstandard beliefs). Whereas Urbit seems to be generally decried as fakery
I came away ranking dagster first, prefect second, everything else not close. IMO dagster wins fundamentally for data engineers bc it picks the right core abstraction (software defined assets) and builds everything else around that. Prefect for me is best for general non-data-specfic orchestration as a nearly transparent layer around existing scripts.
Ofc to each their own based on their usecase.