on the Interplanetary File System
A blog on IPFS about a blog on IPFS
server-free, cloud-free, hack-free, spy-free, censorship-free
One thing I constantly regret is not doing a good enough job at checking for errors in the blog before uploading to IPFS. This is particularly bad after I publish it to IPNS and ask people to pin it, because my name publish TTL is set to 42 days. This causes, I believe, some version conflict in the network as different nodes resolve the name to different Content ID’s while the link propagates through the network.
One thing that never helps is that viewing output in the browser locally after generating is that certain assets like CSS and JS have absolute URI’s that are impossibly not-findable locally, so the page rendering is incomplete and it’s hard to notice things out of place. Invariably, I notice a botched image URI after uploading to IPFS, publishing, and pinning. It’s not nice to fix a little typo and then have repeat the whole process again.
I never questioned the reason for making those asset URI’s absolute as they are by default when installing Frog. Today I did, and decided to test changing them to relative URLs in the page-template.html —as I did with the image URL’s.
Here’s a quick tutorial on using your IPFS site as a force for good in the fight against evil.
UPDATE: Now includes an example of a download scheme different from YouTube, with the convenience of the curl command.
How often you encounter deleted videos in your playlists or on other sites might depend on how controversial your interests are. I decided to put my skills to the test and try to rescue a video published in 2011 by InfoWars, whose account is now banned and deleted from YouTube. Here are the steps I followed.
YouTube account deleted
“At 3pm ET, it appears that Google Cloud (affecting Gmail, YouTube, SnapChat, Instagram, and Facebook among others) mysteriously (and almost unprecedentedly) went offline.”
— ZeroHedge: Did The Government Just Test The Internet Kill Switch?
All Your Base Are Belong To Us
It’s time to let go of the obsolete client-server, surveillance capitalism model. Holochain works. IPFS works. Just say ‘no’ to central points of failure!
Found at: discuss.ipfs.io:ipfs-deploy npm package helps deploy static websites to free IPFS pinning services
I am excited by this tool from @AgentOfUser!
I couldn’t do ‘npm install -g @agentofuser/ipfs-deploy’ because I removed XCODE (it’s too huge for my SSD). So I used the live run method:
npx @agentofuser/ipfs-deploy _site
I’m going to call it: Xfinity is a failed experiment. If I don’t warn you now, I’ll have given you bad advice.
This Zer0-G experiment does show, however that a small (not yet popular) site hosted on IPFS can survive having its seeding node off-line for about two days if you follow the advice outlined on Exospheric tutorials. We have also discovered along the way that among the contributing factors to resilience is a more sane TTL setting for IPNS than the default.
The main reason I strongly recommend against using Xfinity as your primary internet connection is its absolutely stunning unreliability. The craptastic Arris cable modem is fundamentally different than just a router. It crashes and restarts all the time —often several times in an hour. It always takes up to about 10 minutes to come back on-line, and about a third of the time needs to be hard restarted by unplugging the power.
This is a quick update on improving IPNS performance and reliability. I read through the source code and issues on Go-IPFS
publisher trying to decode what’s being done on this.
I don’t want to abandon using IPNS because it does eliminate dependency on the DNS system. For the same reason, I have not worked on using DNSLink. In a worst case scenario, a publisher’s Domain Name can be seized (hijacked).
Also, it would be a pain to update it every time a new version is published. Unless there is an API on your registrar’s site for updating (I think this is doubtful because it is risky) there is no way to pipe the version CID to update the TXT record, so a publisher would have to log in to the registrar’s site and manually administer it. Openprovider.eu, the registrar I use, while having a good reputation and reasonable policies, has a confusing user experience. I want to log in as seldom as possible!
Since one of the aims of this log is to help journalists live on a shoestring budget while keeping their unhackable journal online, I’m eating my own dog food and trying experiments to see what realistically works. I’m documenting what circumstances a static site disappears from IPFS.
If you’re not a Luddite and want to engage the internet, there is only so much you can do to cut expenses. I don’t recommend eating less. So I decided to bite the bullet and say goodbye to T-M*bile and its pillaging. If you’ve had similar experiences with other mobile carriers, the same applies: ask yourself whether you really need them.
I decided to see what it’s like to live in permanent Airplane mode:
- no mobile phone
- no decimeter-accurate body tracking
- no instant and always availability
Technically, “Airplane mode” turns off 2G/3G/4G and Wi-Fi, but you can turn the Wi-Fi back on. So “Zero-G” seems more descriptive.
In my case, I’ll save about $1,400 a year. What could I do with that money? Go see the Pyramids, or buy a HoloPort, or get a nice e-bike.
Wrath of Comcast
I often confuse myself in getting image links to work right in what the author of Frog calls “non-posts” —so here are the results of some experiments to determine a work-around for the way Frog parses site URIs for non-posts. Image URIs don’t get parsed and rendered as one would expect.
What is a “non-post” according to Frog? It’s simply a page that is not a blog entry because it has no date in its source filename (this is the determining factor) and it can lack a date in its metadata. Metadata is ignored for non-posts and is rendered as plain text, so I often omit it for aesthetic reasons. “Non-posts,” ironically, are rendered to the
/posts/ directory, whereas blog post html files end up in the site root.
Above all, a “non-post” is not indexed at all in the blog index page and tag index pages. I think it’s better to call them “non-indexed” pages.
Would it help you if I pinned your content to my node?
Reader Hodlon wrote:
I am new to IPFS and have been playing around with the software some. I came across your blog and enjoy the content you’re publishing. Would it help you if I pinned your content to my node?
To do this I believe all I need to do is pin the hash QmZJBQBXX98AuTcoR1HBGdbe5Gph74ZBWSgNemBcqPNv1W. Is that correct?
I beg your pardon for the long absence! I’ve been fully immersed in what I’d call a third world stress test simulation. Back in November, I thought I’d just say goodbye to T-mobile and go full 0G Wi-Fi hopping in airplane mode. But this is Florida, and the local internet duopoly is Comcast, who took over a month get my house online and provide me with correct information, and to bypass the conflicting, duplicate accounts caused by their system engineers who never imagined what real customers have to put up with if they don’t have a mobile number for 2-factor authentication.
My IPFS blog can’t stay online more than two days under these third world conditions. At least not yet. Some life hacking will need to be done.
It’s time to give some quasi-scientific observations about trying to use Uptime Robot to prod the IPFS gateways to not forget my IPNS lookup and to keep my most recent snapshot alive in the network. In summary, it seems to help, but has its limits.