geokon

the number of gotchas involved in bash scripting indicates to me it's very hard to make robust solutions using the terminal. Trying to understand how things like colors work in the terminal makes the whole thing look very duct-taped together

the video claims it "Steam Ship" and not "Screw steamer". But later claims TSS is "triple screw steamship". So I'd be hesitant to trust the rest of the video's content.

terminal emulation keeps playing a central role in "computing" not withstanding it being an ugly prehistoric crocodile. Do you ever envision creating something modern, without the warts and with a design that incorporates half a century of lessons-learned? Something that fills this need but in a more elegant package? I'm asking because you're one of the few people that seems well-positioned to make a clean-slate replacement

I personally avoid using the terminal as much as possible b/c the whole things feels very archaic/fragile/baroque - but the central idea seems great and I'd love a modern take on it

very cool!

and yeah, I dunno what's with Java apps being clunkers haha. I've come across games in Java that didn't feel bloated, but for some reason desktop apps almost always have a certain bloated feeling to them. Though I feel the same with Electron/VSCode (the input delay to me is very noticeable)

thank you for the work you do :)

> the JVM isn't the only reason Clojure starts slowly, but it's still a major contributing factor

In case you didn't actually look at the links, but the evidence seems to point to this not being true (JVM startup is ~10% of the time)

> you're still hauling around a heavy runtime

> around 8 to 13 megs in size

I think if you jpackage a Clojure app with no GUIs or anything (that then bundles only the strictly necessary parts of the JVM) - it's around 50MBs. It's still embarrassingly slow to launch, involves ugly installers, doesn't provide crosscompilation and the memory usage makes Electron apps look lean.

That's to say, I think Jolt has a very valid place in all of this :) If I could plug in Clojure code and get out small fast executables for all major OSs (cosmopolitan?) that I can then share, that'd be useful for the rare situation I don't have any JVM dependencies

Oh interesting! So they manage to not watermark their maps. OSM seems to heavily imply you should watermark, but as I understand it, it's not strictly required

I often have the experience of going to a museum, seeing a painting I like, and then not being able to find any image of it online - let alone a highres scan. In my casual searches I've found museums generally don't share high resolution scans of their collections. Try to find high resolution images of famous, but not super famous paintings. It's often virtually impossible.

I remember in college my art history professor had a personal digital collection of high quality scans that he'd procure somehow that were impossible to find online. The data seems to be highly guarded, I'm guessing b/c it can be used to make merchandise

So my impression is that the Rodin Museum is not really the exception here

The subtext seems to be a hostility towards people making money from what you've shared.. which I find sort of baffling - but is very on-brand for the general anti-capitalist sentiments in the memesphere. People building on your work feels like something to be celebrated, but I guess it's a difference of opinion. I'm just generally skeptical that this would be "killing the project" at all.. it feels like it'd be the other way around? If OSM data were for instance used by Google and Apple Maps.. it feels like it'd just have made it more foundational and would get people even more excited to contribute. It'd be the world's central repository of map data everyone builds off of.. like the Debian of maps. Instead it's a parochial project for map nerds.

Can you imagine if GIMP has a similar policy to watermark all it's output with a "Made my GIMP!" in an effort to "attracts the flow of new contributors" and to make it hard for people to build a product with it?

A license notice and attribution are not the same though? Furthemore, it's a bit different when it's buried in the meta data somewhere.

The OSM situation seems more akin to GIMP insisting you have to watermark every image with "Made thanks to the GIMP Corporation!"

I understand that there is precedent and there is a logic to it... but to me it rubs me the wrong way. I think text and code are not very analogous situations. If I think in terms of software licenses..:

- I sort of respect the copyleft ethos. You put out a thing with the understand everything it touches is also going to be openly available. I think attribution isn't really part of the central idea there.

- I also respect the MIT/BSD style thing where you're just putting stuff out there in the public domain and it's part of the corpus of human knowledge. You leave your mark so to speak.

The middle ground of "You can use this but you gotta promote our service and stamp our name on it" just feels icky. I don't want to contribute to that. I feel I'm helping OSM the organization and.. I don't know them .. are they good stewards of the data I'm giving them? Are they going to be good stewards in 10 years? Hopefully that kinda makes sense? I guess the same can be said for Wikipedia, but Wikipedia for better or worse is very siloed. You have to attribute stuff you copy from Wiki but ..

A: Realistically nobody is actually copying wiki articles other than lazy high school students. It's just not very reusable outside of Wikipedia

B: This is more of a plagiarism issue. It encourages people to disclose it's not their own words (this is not an issue with maps.. nobody thinks you surveyed your city to draw the map)

If Wiki became CC0 tomorrow nothing would really change. If you needed to fork Wiki and put an attribution in the footer, it'd be very innocuous..

click on the link:

> Licences that are not compatible

> Licences that require downstream attribution

> For practical reasons we require that users of OpenStreetMap data attribute the project as a whole and we in turn provide attribution to third parties via the "Contributor" pages. Licences and terms of use that require attribution of the third party data source directly on derived works are incompatible.

"require attribution of the third party data source directly on derived works" is exactly what OSM requires. It's hypocrisy of a sorts

Is there evidence that LLMs generate better code in more popular languages? I get the sense the "experience" translates between languages and it can reason in any language just fine. I write Clojure code using a rather esoteric framework (Pathom3). There is probably very little similar code out there (it's definitely a tiny fraction of the training dataset) but it seems to do just fine

Not saying you're wrong, just curious if there are numbers backing this up.