There are plenty of other ways to access the Anthropic models, eg: OpenRouter. OpenRouter will automatically use Anthropic/Bedrock based on availability and latency.
I can live preview my website from my local server / computer / laptop, while writing content from basically anything. Even a cheap and underpowered writing deck with 400 MHz CPU. The options are limitless.
You're absolutely underestimating the complexity of proper live preview of changes. This is essentially "hot reload" mode, but on the public internet, because it has to run on a public domain. Getting that right is a challenge, and if you don't know why, you haven't attempted to solve it yet.
Why would I want my WIP site to be on the public internet? This has been built into Jekyll for years. Probably other SSGs too but I don't know/use them.
Because you're thinking in the context of a solo developer working on their site locally. You don't need a CMS. People that want to collaboratively work on a website, some of which may lack technical skills, need a way of previewing their edits that doesn't involve running shell commands.
You're conflating the two unnecessarily. There's no reason Jekyll's server has to run on the editor's local machine. See my other comment about jekyll-admin which can be used collaboratively: https://news.ycombinator.com/item?id=47737324
I don't see much of a difference between `jekyll serve` and the e.g. `service nginx start && service php-fpm start && service mariadb start` that would be needed to run WordPress. In fact I don't run my WIP Jekyll site on the same machine that I edit from. Mine is available only on my personal mesh VPN, but it could easily be available publicly if I wanted it to be.
I would appreciate getting into more details. As it sounds like a made-up problem to me. Perhaps you don’t understand how simple it is to have a static website as compared to a dynamic one. Especially to some simple project.
You can always reduce complexity by moving the goal posts. In context of the original proposition—that one of the core values WordPress provides is live preview—we have to assume a reasonably complex website authored by multiple people.
Live Preview means, then, that you need to:
- have a web-based editor behind secure auth,
- create an environment resembling CI to rebuild the site on demand,
- trigger rebuilds when a user modifies the content stored on the server,
- make the new build available in a draft environment
- …where it doesn't affect the live site,
- …only grant access to collaborators,
- …without breaking assets, links in the pages, CORS, or CSP.
There are more constraints and pitfalls that I'm not going to enumerate here. My point is, this stops being simple as soon as you stop being hand wavy about it.
The solution of this problem could be many, depending on the situation. I have my blog and other static sites synced to all my devices, and the server rebuilds them nightly. In most cases it makes no difference whether I’d deploy it right away, or it would be deployed within a day. Testing the website can be done completely offline, all you need is to sync your changes before the night. Triggering the rebuild on the server (which is deploy) can be done via a Shortcut from an iPhone. No way you can make it as easy with a Wordpress website.
Incidentally, also the curse, for the authors who attempt to monetize SSGs: most users prioritize control, and don't clamor for new features. A nightmare combination for *aaS peddlers.
Last time I interacted with them (admittedly years ago) they were even more allergic to self-promotion than HN. I was read the riot act one time when I posted my own blog.
This is so accurate. When you make one component of a system less expensive (in terms of time or money), the complements get more valuable.
What are the complements of code?
- Distribution
- Operation
- Marketing
- User experience
- Attention
All of these are going to get more valuable. We're at the beginning of this, but I don't see those components becoming less important.
This quote also hit home: "But software is mostly a thing people use, and getting people to use things is not a building problem. It never was."
As engineers, we think that everything is a building problem. Building is fun and seems less risky. It's our natural bent. That's why the canonical advice for software startup founders is to talk to customers as much as possible.
Not specifically but it's the same idea. CIMD is perhaps one step too far for the cases I've worked with. We seem to prefer an out-of-band process for establishing trust. Two CTOs exchanging FQDNs at lunch is a fairly robust model.
Yes, it makes sense in terms of output IF you can sustain the water inputs. But the water inputs aren’t sustainable. Cutting back on that sort of farming in the region would make a much greater impact than fractions of percentages from digitizing infrastructure.
Communicating with submarines that are deep underwater is pretty hard. There are techniques like ELF (Extremely Low Frequency) but they require a small power station to transmit a few characters a minute.
Both give you optionality because they support N models.
reply