What is in the publish.fish script?

This script publishes my website. It’s written in fish, my shell of choice.

Walkthrough

Here’s what each part of the script does:

Auto-Commit

git add .
git commit -m "backup"

These two lines ensure that any changes I haven’t committed aren’t lost.

Build

zola build

Zola makes this really easy, just one command to build the static output. All the files are in a new folder called public.

Codeberg Pages

Codeberg Pages serves my site at birla.io. Using this is very convenient since I just need to push the static pages to the pages branch of the repository.

git checkout pages

First, I switch to the pages branch. The problem is that Codeberg Pages expects the files to be at the root of the repository, not in the public folder we have at this point.

mkdir todelete
mv * todelete/
mv todelete/.domains .
mv todelete/public/* .
mv todelete/.site_cid .
rm -rf todelete

These commands come next Technically, the IPFS steps come next. to replace the current build at the root level with the new one:

1. Make a temporary folder.
2. Move the both the current build and the public folder to that folder.
3. Bring the content of the public folder and a few other relevant files .domains for Codeberg Pages and .site_cid for IPFS out of the temporary folder.
4. Delete the temporary folder.

The last step to publish the site is to commit and push the changes to Codeberg:

git add .
git commit -m "publish script"
git push origin pages

The actual last step is to go switch back to the main branch:

git checkout main

IPFS

I’m not a fan of “web3.” I have fallen for the excitement of cryptocurrency and blockchain in the past, and am now wiser of it. IPFS is web3-adjacent, but the technology is open-protocol If it is not, please tell me. and free.

I love distributed/peer-to-peer technology, and IPFS is a widely adopted technology of this kind. It’s actually really cool, I think.

Since the entire site is static, it can be distributed easily with IPFS. The following two lines keep my site available to access:

ipfs pin rm -r (head -1 .site_cid)
ipfs add --cid-version=1 -Q -r public > .site_cid

The CID of the site is saved in the file .site_cid. The first line stops pinning the old site, and the second line starts doing so.

Notes: ⊕ I also have pinning set up for the pinata pinning service.

• I stop pinning old versions since I push changes frequently.
• -r means I pin every file involved.
• --cid-version=1 lets me use the newer CIDs
• -Q tells ipfs to output just the CID, which > .site_cid saves to the file.
• IPFS isn’t a good fit for large files, but for the web and many smaller files, I think it is great.

A happy accident of the publishing script is that the CID for the rest of the site is published online. One part of the script on base.html is to check for that file and if it exists create a link to the IPFS site at the bottom of each page.