202 karma · joined October 10, 2017
I don't really find that command names reliably relate to words. Offhand none of these would do what I expect: curl, top, git, gawk. less is a more featureful more, and who knows what an xxd is. Certainly not a dead four eyed monster with its tongue hanging out.
stab is probably Send To A Buffer, just like grep is Global Regular Expression Print
> again, another 2-3 key-presses
I guess atm, I would do `/again<enter>nvi(`. That's based off the fact that I don't think "again" shows up much in the text, and I can jump quickly between instances via `n`. Then once I'm there I realise that the rest in in parens, so I go for `vi(`. In retrospect I should have searched for "again," as the comma is even more specific.
It's funny: "Another benefit is precision. I don't need to pay a lot of attention to the mouse pointer's position." is close to an argument I'd make for the mouse. I think that might be because when I'm thinking, I idle with the mouse, higlighting things that look interesting to me etc. So when I want to move forwards, it's pretty close where I want to be. If I had to select something via keyboard I'd have to look for distinct markers to work out if I can do something like `vi(`, whereas the mouse lets me do fuzzy visual matching (acme would. This is only for mouse though, my touchpad/trackpoint is less precise.
My point was not that I think the mouse is better then the keyboard. My point was that I think the mouse tends towards a fuzzier exploratory mode, and the keyboard tends towards more focused execution modes. Even if you like one mode more, there's no reason not to improve the other.
Now after writing all this, I'm not sure that's correct as I can think of other exploratory keyboard UIs that work (e.g Norton Commander style file managers).
I do still think that there is no reason not to improve your mouse experience, a nd that it does have untapped potential. Acme editor has this thing where you can just right click a word, and it'd jump to the next instance of that word. No selection needed. So if you're reading some code and see an interesting variable you could right click to see where it's next used. This of course assumes you have your hand on a mouse already, which in the case of Acme you probably do as it has a heavy mouse focus.
Hover was a terrible example though. I hate having things appear during hover. I was just trying to think through what the least interaction would be, and you're quite right that if you can focus via keyboard easy it'd also be easy to get hover.
----
A weird thought came to me during this: what if the vimperator/vimium/surfingkeys link selector was ported to the mouse? You could hold down the left mouse button, and this would label all the links with numbers, and highlight number 1 (or 0 if you prefer). Then when you spun the mousewheel down it'd move to the next one. Spinning it up would go to the previous link. Finally when you released the left mouse button, it'd open the selected link.
So how it'd play out is you'd hold left, see you want link 11, spin the mousewheel that many times, and then release. I suspect you'd learn how far the numbers feel over time, so it'd get fairly precise (and you can adjust by spinning back).
No idea how practical it'd be.
Not saying it's a bad experience. I find it's quite nice if I'm currently looking at the correct link so I can get the letters from the label easily.
It's just a different paradigm. If I didn't know what a link was, I wouldn't know that I could press f to open the link selection. This means I can't select a link unless I had a "select unknown" button, and then use that to inspect it.
Select unknown can be done with the keyboard: Emacs' embark[^1] is a good example of it. You use search/arrows to go to an interesting piece of text (e.g something underlined), run embark, and it lists all the potential things you can do with it (e.g open in external browser, or download an archived copy). It's just that keyboard UIs do not tend to be geared around it.
That though is the sort of thing the mouse is good at: you see something interesting, you prod it to see what you can do with it.
Say you saw a phrase you didn't understand in the browser. Currently, with a mouse you'd highlight it, right click, and then select the search with search engine option. With keyboard you'd open the search engine, and enter the phrase. Both get you to the same place.
Now if we extend the mouse to have a "I would like to know idioms better" option. You could just select the phrase, right click, and a definition could be there waiting for you next to the context menu. It could even be there on hover.
With the keyboard though, you'd generally run an "I would like to know idioms better" command. This would then ask you which idiom on the page would you like to understand. You then select it, and presto. A faster version could potentially have idiom meanings show up the moment you ran the command by a similar hover effect (or just listing all discovered idioms in a popup etc).
One of the differences here though is that from the mouse you still have a list of actions on the text open, so maybe if you liked the definition you could then copy the phrase.
This certainly could be done from the keyboard UI: just have it list other applicable commands that could be run on the selection generated from the command that searched for phrases. No reason why it could not then spawn something asking if you wanted to "(c)opy, c(u)t" etc like the context menu.
But that is only visible because there was a phrase you didn't know. If you didn't know that there was a command to define phrases, you wouldn't have found out. In theory you could have commands register everything they could act on, and then have another command let the user select an interactive thing, and list the commands that registered for it. I think that would be overwhelming though, as something like a dictionary command would register for every word.
Ultimately the mouse is just different then the keyboard, and I think that there is interesting potential in having a tool that can interrogate arbitrary things on your screen.
Keyboard UI tends to be verb first. E.g in Vimium if you want to open a link in a new tab you press F to start the "Open in new tab" command. Then it lists the links you can open in a new tab for you to select from.
Whereas mouse UI tends to be noun first. E.g you see a link, and right click it to list the commands. You then select the "Open in new tab" command.
I think the reason for the difference is that the mouse can reach an arbitrary location easier than a keyboard, but once it's there it has much less options to continue.
For me, this means that the mouse is more exploratory, and the keyboard is more task oriented. If I'm toying around with ideas the mouse works better, but if I have something specific I want to do you can't beat the keyboard.
Writing this I've realised that I could repurpose the underutilised side buttons on my mouse as modifiers :p Hold top thumb button: press left click to go previous window, right to go next; hold bottom thumb: press left to go to previous workspace, right to go to next. Could have top thumb + mouse wheel for volume (and middle click for mute). I guess bottom middle could open a radial menu of commonly used apps/commands :)
My config files are a bit more interesting, and are stored in a chezmoi[^1] repo. If I tinker with a program enough, I generally move its config into there so I can keep track of it and the changes (to make this easier I have a bat file in shell:sendto to add files/folders to chezmoi).
Thinking about this, I should probably move the chezmoi repo into my personal directory, and symlink it back to it's normal location.
It's things like that which have made me give up on %USERPROFILE% (and to a lesser extent $HOME), and I now put my files that I care about into a completely separate folder outside of the normal hierarchy.
Looks like the dev has moved to Windhawk, and that at least has partial support for ungrouping[^2].
[1]: https://ramensoftware.com/7-taskbar-tweaker-and-a-first-look...
[2]: https://ramensoftware.com/windhawk-mods-for-the-windows-11-t...
e.g this matches markdown code blocks (and also extracts the language from the start to select the mode to run)
(define-auto-innermode poly-markdown-fenced-code-innermode
:head-matcher (cons "^[ \t]*\\(```{?[[:alpha:]].*\n\\)" 1)
:tail-matcher (cons "^[ \t]*\\(```\\)[ \t]*$" 1)
:mode-matcher (cons "```[ \t]*{?\\(?:lang *= *\\)?\\([^ \t\n;=,}]+\\)" 1)
:head-mode 'host
:tail-mode 'host)
( via https://polymode.github.io/defining-polymodes/ )Makes me wonder if you could encode smell to straight audio though. Then every area could have a sort of hum.
I'm not sure it's linked anywhere visible, but it does have an entry in the html head....
Was testing a new program, and wanted to search the issues to see if someone had the same sort of UI problems I had (and hopefully a fix). No luck there.
I then googled to see if someone had a workaround for the GitLab search issue, and found your post :p
Decided to not switch to the new app.
My current approach is to use a directory.build.props file in the folder where I clone into (i.e. outside of all the repos), and use it to set the package output path to a local folder and timestamp the package version. That way every build creates a new package to test locally, but I still have to update the references in the dependent projects when testing (and build everything separately).
Not particularly happy with it, but I don't have to manage that many shared dependencies either.
I wonder if a precompile target has enough access to tinker with the references? I.e. remove items from the PackageReferences item group, and create corresponding ProjectReference items. Probably will need to also call the MSBuild build task on those projects too so they can have a fresh build in case any changes have been made in them too.
But yes, it would be great if Microsoft supplied something for that.
See: https://docs.microsoft.com/en-us/visualstudio/msbuild/custom...
I don't use them for package references myself, but this blog post has some information on how you'd do that: https://www.mytechramblings.com/posts/centrally-manage-nuget...
Internal changes aren't going to be detectable, so it feels like the best you can hope for is that they don't want the maintenance burden of a patch set on top of your project. At that point it's not much different than MIT.
That said, AGPL would scare off most companies :p
I don't use them enough, but the replacements are interesting: they seem to mostly search for indicators of a forum (e.g the existence of a next page button).
Hah, that resonated with me. I've had anosmia since birth, and I had a moment this week where I was chewing on rosemary to gauge what its impact on my dish would be, and thought to myself "I really don't get herbs".
I can't find when it started happening, but there is at least someone wanting to turn it off Jan 2019: https://developercommunity.visualstudio.com/idea/437785/allo...
They're basic verbs you use to interact with text; not a full command. E.g instead of executing "format document" from a menu, you select the document with the left mouse button, and then middle click on the word "format". This executes the "format" command on the selected text. Somewhat like running gg=G in vim.