zenoprax

On this netbook, which has 1GB of RAM, if you keep the swap partition small you will run into “No space left on device” errors all the time.

The author doesn't explain why but I'm guessing this is because `/tmp` is filling up. Setting a quota on this mount point would help limit the impact that badly behaving programs might have on RAM usage.

Regarding memory compression: isn't the use of `zswap` and `zram` commonplace on embedded Linux at this point?

I've never heard of people having difficulty with inline replies before. Show them how to configure `sed 's/> /You said /g'` to run on all their incoming email...

(I use the same plain text email strategy)

This was such an improvement to mine as well. Lots of strange advice in this thread about "one sentence/question per email" and other arbitrary terseness. Just use a messaging system for single-thread async communication.

Writing the email is part of the thinking process for me so by the time I finish there is a single coherent sentence that I can cut-paste to the top with the rationale/evidence/etc organized below it. If the one-liner is enough they can skip the rest (which I needed to write anyway regardless) and if they need more detail they don't have to ask.

Some other practices I've picked up over the years:

Responses should always be inline between relevant quotes rather than the default of hiding the original at the bottom. Remove quotes that don't matter. My favourite responses are a simple "Agreed." after a long quoted paragraph of justification.

In general the email chain should get shorter with each reply otherwise it suggests that the topic is either too complex or insufficiently defined. It also gives some sense that it will eventually end!

Write subject lines that make searching/filtering easy to do later rather than something to catch the recipient's attention.

With a tagline of "a universal document converter" it is almost a guarantee to become a complicated program but how much of it is being used for any single conversion?

Two more examples:

Rclone is "bloated" but it needs to be in order to fulfill its purpose.

ZFS is "bloated" because it combines volumes and filesystems but breaking the Unix philosophy also enables a different kind of synergy and simplicity elsewhere.

Where is the "data loss window"? Between nodes or between the client and the infra?

The front page lists under Capabilities:

4.8 Honest fsync A successful fsync means every acknowledged write is durable in S3. If a failover may have lost unflushed writes, the next fsync returns an error instead of a false success.

Someone else mentioned: "write() is buffered (that's the batching) and "committed" maps to fsync(), which returns only once data is durable."

---

It sounds like all writes are written synchronously to at least one node but failovers/replicas are just eventually consistent. If so, latency between nodes is not within ZeroFS's control and including. Or are you saying that the latency is impossible for even a single node? If so, that would mean much more than just a footnote is needed.

---

I don't see an issue with the benchmark but I might not be looking in the right place.

The [sequential writes](https://github.com/Barre/ZeroFS/blob/ec32199d48d0409d4cccd44...) and [append-only writes](https://github.com/Barre/ZeroFS/blob/ec32199d48d0409d4cccd44...) start where I'd expect and `success: true,` should equate to `fsync()` as previously mentioned.

(Apologies if I got this wrong, still learning this)

I have a feeling someone is going to jump in with a "solution" to each of these forgetting that lots of work is ephemeral/transient and undeserving of even the slightest bit of automation or pre-configuration.

Others have already backed up bit-for-bit copies on your behalf: you just need to download it from them piece by piece.

Bear in mind that you would likely be infringing on the rights of Apple since US law only permits backups for physical media.

Controlling the TLD has its own benefits and drawbacks (managing email reputation, for example) but as a regular person I have more reason to trust `.cloud` than `.self` purely on the basis of proven continuity. My `.com` domain will almost certainly live as long as the internet does provided that I keep paying to renew.

Regardless, a UUID is probably the right call. It doesn't help with memorability but it's at least more stable than an IPv4/IPv6 address and can be hard-coded. I wonder if you would get a full zone or if it's just an A/AAAA record given their broader goals of email and VPN tunneling.

I think there a simpler way to summarize this list:

Digital Purchases are non-transferable licenses to access "content" from a "provider". The terms of the deal are alterable and revocable at the provider's discretion with no obligation to maintain pricing, quality, availability, or editorial/artistic integrity nor must they provide any advance warning of such changes nor is there any legal recourse in case of disagreement.

Written out plainly it sounds pretty hostile when compared with an immutable Bluray whose limitations and capabilities are known in advance.

image of an apple, stored as an analog signal on a magnetic tape

If I am following along...

analog virtual: representation of apple on photographic film digital physical: disc with film.mkv

Given the two remaining combinations:

analog physical: unexposed photographic film? digital virtual: binary-encoded data (film.mkv)?

Anything beyond this and it becomes a philosophical or metaphysical discussion though.

I think you're playing language games here. What does an "analog virtual thing" look like?

Digital = expressed by discrete bits of encoded digits (1s and 0s). Analog = lossy and necessarily physical

A "digital physical thing" is just a physical thing (disc) with digital things encoded on it.

Doesn't it sound reasonable?

If you were hired with this as an explicit expectation, yes. It's one thing to know that your actions can be audited in case there's some sort of incident but imposing unlimited surveillance and using that information for the purpose of eliminating your job could be argued to be intimidation (ie. "we can't afford mass layoffs but aggressively monitoring employees will force the undesirables to quit").

No one likes the terms of their employment being changed against their will no matter how legal it might be. Why not make it opt-in in exchange for some other perks? If the data is valuable then compensate employees for the added burden/liability of total telemetry.

I'm with you there. I have to either hear the full sentences narrated by my internal voice or see the words flashing in my mind in order to "think". This is great for building and maintaining deep mental models but it is also highly susceptible to "bit rot" (such as forgetting the rationale or evidence for a specific assertion or position) days/weeks/years later. I have a friend who simply can't understand how inter-linked note systems (like the kind Obsidian enables) are helpful. It's just a bewildering mess to them and they think more linearly.

Thus, writing things down is a necessity for me: it's not for a need for structure but rather that my "context window" gets filled too quickly. I can counter my own arguments but it's more fun, and often quicker, to do with someone else. Besides, there is such a diversity of thinking out there it would be foolish to not take advantage!

Middle-click a link to open a new tab. I can't force a link to open in the same window but I can with new tabs.

Sometimes it's easier to follow a link, have a look, and then go back without jumping around tabs.

Reminds me of the show Severance. You don't know what the master plan is for several seasons even with exposure to all the quirky subdepartments: https://www.severance.wiki/lumon_depts

I have emailed two authors now and both responded enthusiastically and answered my questions. Granted, these were also niche texts so I don't imagine it's common for them to get fanmail either!

I have emailed people based on a YouTube video, podcast episode, blog post, or just browsing a project on GitHub. If their email address is available I see that as permission to contact them for "wholesome" purposes. A few things that come to mind:

1. clarification on something in particular that they have already published

2. engage in genuine discussion about adjacent topics in which their opinion is specifically relevant

3. expressions of appreciation

4. corrections of information to prevent genuine harm or significant frustration for others

My success rate is probably 50-75% but I only do it a few times per year.

Cold-calling to get people try try your new app or answer a survey is rude.

People continue to criticize Arch for being elitist or gate-keeping to keep casuals out but there are clear benefits by not allowing dangerous things to be simple. This is true in many aspects of life.

After using Void Linux I switched to `aurutils` to get a similar separation on Arch. I can easily maintain a local AUR repo by compiling/making my own binaries and can use `pacman` to install and manage them which improves the upgrade process overall.

First time seeing an HTTP 451 in the wild for me.

Edit: I take it back. Just a 200 in a trenchcoat.

It is frustrating to know that we can digitally sign and encrypt messages but don't because "it's too hard for normal people".

With HIPAA, is it not possible to simply encrypt the message? The "forgot password" flow for their message center is probably email anyway.

I can upload my public key to SourceHut and all email from them becomes signed and encrypted. It's a one-time process to generate long-lived keys and another to set up with SourceHut and that's all I need to do.

"user@something"@example.com is a valid address

I appreciate your commitment to correctness but like [XKCD 1172](https://xkcd.com/1172/) ... the user is clearly in the wrong at this point.

Inverting the order actually addresses my primary annoyance: what is a feature?!

refactor(core): Update webmcp support to use document.modelContext

As the author points out, the line between a fix, an improvement, and general clean-up is blurry and dividing each semantic change into its own commit (and possibly squashed later anyway) is just creating work for no one's benefit.

I think Conventional Commits are just an artifact of trying to automate SemVer rather than solving any of the other problems directly. I don't think changelogs should be automated anyway - I can `git log` that if I want a list. A changelog is an opportunity to communicate to a wider audience what is actually going on under the hood.

What a great capsule of wisdom!

There is still a tendency within some parts of aviation (safety auditing) to look for root causes and use tools like "fish bone diagrams" despite the more holistic approach used after an actual crash or incident.

Alternative view: it works best for flatter laces. I have a pair of running shoes with thicker round laces that don't stay tied unless I use the traditional method.

I'm curious about the physics involved to cause such an obvious and singular failure.

Cursor has limits even when using your own key. I was even cut off using a local model. I guess they use some sort of harness that requires non-local resources? I'm not sure I've actually tried to use Cursor in a fully-offline scenario yet. Cline works well enough and doesn't require any sign-up.

I didn't know about these options either. I am using Cline: Cloudflare isn't an option but Vercel is. My spending is pretty low overall now that I'm using local models much more but good to know that there are cheaper alternatives to try or at least suggest to others.

Other features I've just noticed: - configurable prompt injection protection using OWASP regex (https://cheatsheetseries.owasp.org/cheatsheets/LLM_Prompt_In...) - configurable PIM protection for outbound prompts - input/output logging - "JSON healing" to auto-correct minor hallucinations

Lots of other stuff too. The business model seems pretty simple and the value-add features don't look particularly expensive or difficult to copy.

I was expecting the use of non-SSH git remotes without network access. Any mounted file system can be a valid remote such as a USB drive. I use file-based remote to keep some repos encrypted on S3 using Rclone.

For example, `git remote -v` would show: `secure-s3 /mnt/fuse/rclone/secure-s3/git/$REPO.git`

I think concurrency is a problem with file-based remotes but for one person keeping a desktop and laptop in sync it is much simpler than running a VPS.

I did the same thing but I'm now pushing it a bit further: POSIX shell rather than Bash for scripts. If what I'm doing can't be done with that it suggests that I should probably just write it in Python or Perl instead.

Fish scripting is limited to functions/aliases and this works out well since they're easy to read and tweak over time.

Awesome, thanks for the explanation. I didn't know that kuid/kgid existed! That also explains why Proxmox manages the re-mapping in the `LXC.conf` rather than the AppArmor profile. The cascade of AppArmor configs seemed to focus quite a bit on access to `/proc` and `/sys` so I think I mixed that up cgroups with my comments about memory access.

I've been reading up on them (https://www.kernel.org/doc/html/latest/filesystems/idmapping...) seeing some of the notation for user IDs (e.g. `u20000`) reminded me that my Hetzner ZFS storage was accessed using a similar UID format for the username.