Hacker Newsnew | past | comments | ask | show | jobs | submit | klez's commentslogin

> Anything that doesn't need external "life" to come alive, I would consider as "life" in various states

Doesn't almost any biological entity need external life to come alive via, e.g. reproduction or mitosis or what have you?


Yes but you are talking about the threshold of existence, and the cell is alive as soon as it starts existing. For a virus you also have the threshold of "application", when the viral code is applied to something alive. Before that the virus exists but is not alive itself. After the application it's modifying other life which maybe technically can be considered alive.

This is why I said "to come alive" instead of "to be created". The virus is something that just exists but only becomes alive when mixed with something that's already alive.


Sure, but considering how central and defining the concept of "life" is to biology (the study of life and living organisms) you'd think we wouldn't have a fuzzy definition for that specific concept. I can see why it's tricky, though.

Life is very useful as a term because it allows you to define a 'not living' term as well: dead. And it has meaning at the highest level. But if you start looking at things in a more detailed way even death doesn't arrive 'all at once' for multi cellular organisms, for instance a dead person's hair still grows to the point that corpses need to be shaved. And a virus may be dead by one persons view on what 'life' is all about but alive by someone else's definition. And depending on the context both of them may be right.

The definition is fuzzy because the concept is fuzzy! Even something that we in every day life see as settled such as a species is not always clear-cut. Cat or dog? That's usually easy. Member of a species yes or no? Not so easy, and in some cases subject to considerable debate and even then unresolved.

For instance:

https://en.wikipedia.org/wiki/Species_complex

And:

https://en.wikipedia.org/wiki/Ring_species

Boundaries are hard, just like naming things.


Agree on the fuzzy term.

It is misconception that hair and nails continue to grow. What happens is the that kind and soft tissues dehydrate and shrink and the hairs and whiskers stick out more. Growth stops soon after oxygen and nutrients stop being delivered.


Ah good one! Another childhood myth killed off.

We really only have one example of life (or at least all our examples are interconnected), so I don't expect great definitions.

Just like geology doesn't have a great definition for their subject of study (the earth). They have a definition that works really well, but because they only have one example, the definition ain't stress tested.

Slightly less silly: it took the discovery of lots more bodies inside and outside the solar system (dwarf planets here, exoplanets elsewhere) for astronomers to really nail down the definition of planet.


I find Maturana’s definition satisfactory. Best overview in his book with Francesco Valera:

Autopoiesis and Cognition: The Realization of the Living (1980)

https://en.wikipedia.org/wiki/Autopoiesis_and_Cognition:_The...

Unfortunately out of print.


how does it exclude fire, something well known for making more of itself (which seems to be the main criteria in Autopoiesis)

I think that depends on your prior expectations about how biological concepts should be structured. I was trying to make the case that we should expect that they're fuzzy when we're dealing with very complex phenomena that exhibit a lot of variation. The fact that these kinds of phenomena happen to exhibit clustering is what makes (fuzzy) classification possible, but we also find that many phenomena or organisms are in borderland areas between clusters, so the classification doesn't work as well with them.

The problem is that it's really hard to come up with a definition that includes all of the things we agree are obviously life (e.g. mold), does not include fire, and does not just appeal to the particular structure that most or all life on earth seems to have (the cell).

The result is a landscape of fuzzy definitions mostly centered around that last one.


Analyze a word you think has a crystal clear definition.

We only have a single tree-of-life (or possibly, several syncretic trees-of-life from a single planet) as an example. Makes it a little difficult to discern the true principles.

I, for one, still need to look it up every time I see it mentioned. Not everyone is talking or thinking about LLMs every waking minute.

Are you looking up what the abbreviation stands for, or what an MCP is?

The first case doesn't matter at all if you already know what an MCP actually is.

At least for the task of understanding the article.


MCP being the initialism for "Model Context Protocol", the specification released by Anthropic, generally dictates you shouldn't say "an MCP" but simply "MCP" or "the MCP". If you are referring to a concrete implementation of a part of MCP, then you likely meant to say "an MCP Server" or "an MCP Client".

I figured with all the AI posts and models, tools, apps, featured on here in the last year or two that it was a given. I guess not.

One lesson I learned from this latest upgrade: DON'T SKIP READING THE DOCUMENTATION BEFORE UPGRADING.

I rebooted after upgrade and found myself dropped in a console-only environment. Turns out the (proprietary) drivers for my video card are not supported anymore. Now I'm on nouveau and it sucks.

My fault for having an Nvidia card, I guess, but an AMD equivalent is on its way.


For this exact reason I enable Timeshift on every computer I run. If an upgrade fails spectacularly like that, recovering and reverting to the old edition is as simple as picking a pre-upgrade snapshot from the boot menu.

If you're not on btrfs, I'm sure there are similar tools for your favorite file system. For me, it's been a life saver.


OpenSUSE has this sort of thing configured by default using BTRFS and snapper. Automatically creates copy-on-write snapshots before and after every time you use the package manager. I think this should be the default on most end-user oriented distros.

Fwiw there's a really nice write-up on how Steam Deck does A/B updates, using rayc and casync/desync. https://github.com/randombk/steamos-teardown/blob/master/doc...

Also Collabora had a nice update on tuning work they did. https://www.collabora.com/news-and-blog/news-and-events/stea...

I'm a little surprised they use casync and multiple partitions, given that root is on btrfs. As they have it is more reliable if anything does go bad, so makes sense, but: given that steam decks are on a given snapshot of the btrfs system, it seems like they could just send an incremental update to the next release and try to boot that. Oh ah, that breaks down if, like me, you disable the readonly root.


how much space does that take up for you?

At least for btrfs, very little.

It's also almost instantaneous.

You can think about it like this: the snapshot creates a bunch of hard links to the existing files. So every file basically has 2 pointers to it.

The upgrade now updates the files in the non- snapshot paths, creating new files/keeping the snapshots on the original file. (That's afaik why it's called a CoW filesystem - copy on write)

Now, every changed file will be stored "twice", but all unchanged files are still only on the disk once.

As most upgrades still keep the majority of all files unchanged, there is effectively little space used for the snapshot


definitely going to check that out, thanks

Similarly, Trixie contains an updated version of Dovecot that (even though the version number seems to indicate otherwise) has a new configuration format that is not backwards compatible. This is clearly stated in the release notes but may be surprising nevertheless.

I have too developed PTSD from nvidia + debian (even worse, nvidia optimus on a laptop).

To the point where I would not upgrade my system for months at a time, when I was in the midst of important work, because I didn't want to spend a day on tty trying to make the DE work again.

In the words of Linus: FUCK NVIDIA


If your card is that old, shouldn't the nouveau drivers be quite full-featured anyway? Regardless, you may also want to run a Bookworm kernel with a full Trixie userspace. It should install cleanly once you add the Bookworm repos, and you'd be fully supported for quite some time thanks to LTS. It would technically be a FrankenDebian, but one that might be quite acceptable nonetheless.

> If your card is that old, shouldn't the nouveau drivers be quite full-featured anyway?

Normally they would probably be, but it looks like my ultra-wide screen is too much to handle for those drivers.

> you may also want to run a Bookworm kernel with a full Trixie userspace.

Good idea indeed. But at this point I'm afraid of borking everything again. I'll just wait for the replacement card I ordered and hope it doesn't make things worse.


> If your card is that old, shouldn't the nouveau drivers be quite full-featured anyway

Unfortunately, nouveau is very limited and barely works (when it works).


Especially on older cards actually! The new hardware, since the 3000 generation or so, moved some "secret sauce" (probably bullshit) into the firmware so that everything that the drivers need to do is something that Nvidia is willing to disclose, so there can be open source drivers that can get the hardware out of low power mode. That lack of "reclocking" crippled performance with nouveau.

Eh, I'd still buy AMD if possible.



Can't you install the drivers from nVidia directly?

I tried, but it fails at compile time and I can't be bothered to try and understand why it doesn't find a stdargs.h that is _clearly_ there. I attempted various solutions I found online for half a day, then just threw the towel and ordered a 25€ replacement from AMD which I'm told has better support than nvidia on linux anyway.

Ah, yeah, that tracks. I've heard the same, though I haven't replaced my nVidia yet.

Reading this I feel blessed having found https://universal-blue.org

You don't need a special OS to have backups or even just btrfs/lvm snapshots.

Don't understand the downvotes.

Ublue family of atomic os, provides snapshot restore which helps exactly in these scenarios.


In some parts of Italy, the name will make people's inner 12-year-old giggle. It sounds like a childish name for "penis". Do what you will with this information.


Also, in French, pipeau is a kind of flute. "Playing flute" meaning bulshitting. Do what you will with this information.


Luckily, at Pipo360, we’re doing the opposite: cutting through BS and giving devs real, working backends in 60 seconds.

No flute-playing here. Just code.

French devs — we see you


This reply was 100% generated by chatgpt. It always gives me that style. "No <X> here. Just <Y>"


No prompts here. Just a dev building fast and figuring it out in public


You literally have em-dashes in your reply and original post. You are following the very standard "No X, no Y - just Z" ChatGPT speech pattern.

You are either lying or you've learnt english from ChatGPT.


Very rude to continue to lie about it


Thanks for the heads up - that's exactly the kind of cultural insight we need as we build for global markets. We're evaluating our options and really appreciate you taking the time to share this


Is this the usual anti-java sentiment or is there something in particular with java >= 17 that you have something against?


It has obvious annoyances like these deeply nested paths:

quarkdown-quarkdoc-reader/src/main/kotlin/com/quarkdown/quarkdoc/reader/dokka/DokkaHtmlContentExtractor.kt

https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpris...


I can't seem to understand how their syntax is "markdown-like" in any way. All I see is some sort of declarative language surrounded by code.


Totally confused. Lost 10 minutes.


So that's what it is! Just the other day I was visiting an old website on the wayback machine which I didn't expect to work because it was flash-based. To my surprise everything worked fine and I kept seeing those raffle logos for a second while it loaded. Looks like the internet archive is applying it automatically when it shows website containing flash animations.


You can automate commits and push on save. I had a similar setup for vimwiki before migrating to a web-based wiki system (dokuwiki).


Out of curiosity, what other solutions did you explore and why were they lacking?


(Not OP)

I've used LibraryThing for book boxes. Using smallish boxes (30-40 paperbacks each) so that carrying them is not a backbreaker. Scan the ISBN barcodes with phone app, fix old ones/whatever on web app, tag with box number written on at least two sides. No problems found so far.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: